diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,142406 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999788748758899,
+  "eval_steps": 1000,
+  "global_step": 23668,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0281293952180027e-06,
+      "loss": 1.6563,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0562587904360055e-06,
+      "loss": 1.4033,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0843881856540082e-06,
+      "loss": 1.5079,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.112517580872011e-06,
+      "loss": 1.3928,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.140646976090014e-06,
+      "loss": 1.4684,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.168776371308017e-06,
+      "loss": 1.5503,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1969057665260198e-06,
+      "loss": 1.4389,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2250351617440226e-06,
+      "loss": 1.4129,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2531645569620253e-06,
+      "loss": 1.4857,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.281293952180028e-06,
+      "loss": 1.4406,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.309423347398031e-06,
+      "loss": 1.4417,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3375527426160337e-06,
+      "loss": 1.3166,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3656821378340365e-06,
+      "loss": 1.4005,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3938115330520392e-06,
+      "loss": 1.4046,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.421940928270042e-06,
+      "loss": 1.3417,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.450070323488045e-06,
+      "loss": 1.3923,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.478199718706048e-06,
+      "loss": 1.3688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5063291139240504e-06,
+      "loss": 1.4044,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5344585091420536e-06,
+      "loss": 1.2498,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5625879043600563e-06,
+      "loss": 1.4288,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.590717299578059e-06,
+      "loss": 1.3658,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.618846694796062e-06,
+      "loss": 1.3695,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6469760900140647e-06,
+      "loss": 1.3905,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6751054852320675e-06,
+      "loss": 1.3453,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7032348804500702e-06,
+      "loss": 1.4288,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.731364275668073e-06,
+      "loss": 1.359,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7594936708860758e-06,
+      "loss": 1.2784,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.787623066104079e-06,
+      "loss": 1.4178,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8157524613220813e-06,
+      "loss": 1.318,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8438818565400845e-06,
+      "loss": 1.3026,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8720112517580873e-06,
+      "loss": 1.3991,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.90014064697609e-06,
+      "loss": 1.2211,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.928270042194093e-06,
+      "loss": 1.3886,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9563994374120957e-06,
+      "loss": 1.3065,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9845288326300984e-06,
+      "loss": 1.3889,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0126582278481012e-06,
+      "loss": 1.2823,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.040787623066104e-06,
+      "loss": 1.2679,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0689170182841068e-06,
+      "loss": 1.1773,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.09704641350211e-06,
+      "loss": 1.3811,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1251758087201123e-06,
+      "loss": 1.2857,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1533052039381155e-06,
+      "loss": 1.2451,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.181434599156118e-06,
+      "loss": 1.3733,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.209563994374121e-06,
+      "loss": 1.3617,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.237693389592124e-06,
+      "loss": 1.318,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2658227848101267e-06,
+      "loss": 1.3163,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2939521800281294e-06,
+      "loss": 1.3323,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3220815752461322e-06,
+      "loss": 1.2101,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.350210970464135e-06,
+      "loss": 1.2305,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3783403656821378e-06,
+      "loss": 1.3876,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4064697609001405e-06,
+      "loss": 1.237,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4345991561181433e-06,
+      "loss": 1.1852,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4627285513361465e-06,
+      "loss": 1.2397,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.490857946554149e-06,
+      "loss": 1.269,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.518987341772152e-06,
+      "loss": 1.2536,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.547116736990155e-06,
+      "loss": 1.3164,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5752461322081576e-06,
+      "loss": 1.3419,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6033755274261604e-06,
+      "loss": 1.25,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.631504922644163e-06,
+      "loss": 1.2434,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.659634317862166e-06,
+      "loss": 1.2751,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6877637130801688e-06,
+      "loss": 1.323,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7158931082981715e-06,
+      "loss": 1.302,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7440225035161743e-06,
+      "loss": 1.2925,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7721518987341775e-06,
+      "loss": 1.2695,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.80028129395218e-06,
+      "loss": 1.3898,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.828410689170183e-06,
+      "loss": 1.2238,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8565400843881854e-06,
+      "loss": 1.3654,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.884669479606189e-06,
+      "loss": 1.2566,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.912798874824192e-06,
+      "loss": 1.256,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.940928270042194e-06,
+      "loss": 1.2897,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9690576652601966e-06,
+      "loss": 1.3349,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9971870604782e-06,
+      "loss": 1.2583,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.025316455696203e-06,
+      "loss": 1.2374,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.053445850914205e-06,
+      "loss": 1.2907,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.081575246132208e-06,
+      "loss": 1.3248,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.109704641350212e-06,
+      "loss": 1.3012,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.137834036568214e-06,
+      "loss": 1.2809,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.1659634317862164e-06,
+      "loss": 1.2612,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.19409282700422e-06,
+      "loss": 1.3207,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.222222222222222e-06,
+      "loss": 1.303,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.250351617440225e-06,
+      "loss": 1.2021,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2784810126582275e-06,
+      "loss": 1.2089,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.306610407876231e-06,
+      "loss": 1.1564,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.334739803094234e-06,
+      "loss": 1.264,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.362869198312236e-06,
+      "loss": 1.2611,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.390998593530239e-06,
+      "loss": 1.238,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.419127988748243e-06,
+      "loss": 1.1705,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.447257383966245e-06,
+      "loss": 1.2438,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.475386779184247e-06,
+      "loss": 1.2626,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.503516174402251e-06,
+      "loss": 1.2461,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.531645569620254e-06,
+      "loss": 1.3176,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.559774964838256e-06,
+      "loss": 1.3113,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5879043600562585e-06,
+      "loss": 1.2479,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.616033755274263e-06,
+      "loss": 1.2541,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.644163150492265e-06,
+      "loss": 1.2923,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.672292545710267e-06,
+      "loss": 1.2969,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.70042194092827e-06,
+      "loss": 1.2745,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.728551336146273e-06,
+      "loss": 1.2059,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.756680731364276e-06,
+      "loss": 1.2701,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.784810126582278e-06,
+      "loss": 1.3825,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.812939521800281e-06,
+      "loss": 1.2798,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.841068917018284e-06,
+      "loss": 1.297,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.869198312236287e-06,
+      "loss": 1.2591,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8973277074542895e-06,
+      "loss": 1.3117,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.925457102672293e-06,
+      "loss": 1.2606,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.953586497890296e-06,
+      "loss": 1.4064,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.981715893108298e-06,
+      "loss": 1.1938,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.009845288326301e-06,
+      "loss": 1.2704,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.037974683544304e-06,
+      "loss": 1.3278,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.066104078762307e-06,
+      "loss": 1.2404,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.094233473980309e-06,
+      "loss": 1.2651,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.122362869198313e-06,
+      "loss": 1.2539,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.150492264416316e-06,
+      "loss": 1.2118,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.178621659634318e-06,
+      "loss": 1.2684,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2067510548523205e-06,
+      "loss": 1.2748,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2348804500703246e-06,
+      "loss": 1.2056,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263009845288326e-06,
+      "loss": 1.2386,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.291139240506329e-06,
+      "loss": 1.2297,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.319268635724332e-06,
+      "loss": 1.1487,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.347398030942335e-06,
+      "loss": 1.1811,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.375527426160338e-06,
+      "loss": 1.2794,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.40365682137834e-06,
+      "loss": 1.2754,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.431786216596343e-06,
+      "loss": 1.2376,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.459915611814346e-06,
+      "loss": 1.2737,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.488045007032349e-06,
+      "loss": 1.2299,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5161744022503515e-06,
+      "loss": 1.2401,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.544303797468355e-06,
+      "loss": 1.2135,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.572433192686358e-06,
+      "loss": 1.1779,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.60056258790436e-06,
+      "loss": 1.2898,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.628691983122363e-06,
+      "loss": 1.1665,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.656821378340366e-06,
+      "loss": 1.2687,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.684950773558369e-06,
+      "loss": 1.2074,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.713080168776371e-06,
+      "loss": 1.2808,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.741209563994375e-06,
+      "loss": 1.2596,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.769338959212378e-06,
+      "loss": 1.2047,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.79746835443038e-06,
+      "loss": 1.2173,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8255977496483825e-06,
+      "loss": 1.2155,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.853727144866385e-06,
+      "loss": 1.2824,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.881856540084389e-06,
+      "loss": 1.2213,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.909985935302391e-06,
+      "loss": 1.2977,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.938115330520394e-06,
+      "loss": 1.2706,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.966244725738396e-06,
+      "loss": 1.2215,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9943741209564e-06,
+      "loss": 1.2333,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.022503516174402e-06,
+      "loss": 1.2581,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.050632911392405e-06,
+      "loss": 1.2604,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.078762306610409e-06,
+      "loss": 1.2629,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.106891701828411e-06,
+      "loss": 1.2008,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.1350210970464135e-06,
+      "loss": 1.2558,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.163150492264416e-06,
+      "loss": 1.2778,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.19127988748242e-06,
+      "loss": 1.2621,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.219409282700422e-06,
+      "loss": 1.1963,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.247538677918425e-06,
+      "loss": 1.262,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.275668073136429e-06,
+      "loss": 1.2725,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.303797468354431e-06,
+      "loss": 1.2566,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.331926863572433e-06,
+      "loss": 1.1922,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.360056258790436e-06,
+      "loss": 1.2006,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.38818565400844e-06,
+      "loss": 1.223,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.416315049226442e-06,
+      "loss": 1.2399,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4444444444444445e-06,
+      "loss": 1.194,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.472573839662447e-06,
+      "loss": 1.2067,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.500703234880451e-06,
+      "loss": 1.3445,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.528832630098453e-06,
+      "loss": 1.2552,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.556962025316456e-06,
+      "loss": 1.2961,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.585091420534458e-06,
+      "loss": 1.2478,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.613220815752462e-06,
+      "loss": 1.2375,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.641350210970464e-06,
+      "loss": 1.2764,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.669479606188467e-06,
+      "loss": 1.2616,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.697609001406471e-06,
+      "loss": 1.2258,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.725738396624473e-06,
+      "loss": 1.2442,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7538677918424755e-06,
+      "loss": 1.1389,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.781997187060478e-06,
+      "loss": 1.2798,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.810126582278482e-06,
+      "loss": 1.3108,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.838255977496484e-06,
+      "loss": 1.2596,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.866385372714487e-06,
+      "loss": 1.3261,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.894514767932491e-06,
+      "loss": 1.1952,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.922644163150493e-06,
+      "loss": 1.2244,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.950773558368495e-06,
+      "loss": 1.2659,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.978902953586498e-06,
+      "loss": 1.2128,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.007032348804502e-06,
+      "loss": 1.2766,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.035161744022504e-06,
+      "loss": 1.1933,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0632911392405065e-06,
+      "loss": 1.2205,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0914205344585105e-06,
+      "loss": 1.2495,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.119549929676513e-06,
+      "loss": 1.2117,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.147679324894515e-06,
+      "loss": 1.1675,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.175808720112518e-06,
+      "loss": 1.2815,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.203938115330522e-06,
+      "loss": 1.2536,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.232067510548524e-06,
+      "loss": 1.2431,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.260196905766526e-06,
+      "loss": 1.2843,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.28832630098453e-06,
+      "loss": 1.2848,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.316455696202531e-06,
+      "loss": 1.2447,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.344585091420535e-06,
+      "loss": 1.2288,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.3727144866385375e-06,
+      "loss": 1.183,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.40084388185654e-06,
+      "loss": 1.2325,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.428973277074542e-06,
+      "loss": 1.2686,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.457102672292546e-06,
+      "loss": 1.2973,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4852320675105486e-06,
+      "loss": 1.251,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.513361462728551e-06,
+      "loss": 1.2492,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.541490857946555e-06,
+      "loss": 1.2274,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.569620253164557e-06,
+      "loss": 1.2152,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.59774964838256e-06,
+      "loss": 1.2799,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.625879043600562e-06,
+      "loss": 1.2478,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.654008438818566e-06,
+      "loss": 1.2556,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.682137834036568e-06,
+      "loss": 1.1489,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.71026722925457e-06,
+      "loss": 1.2534,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.738396624472575e-06,
+      "loss": 1.2425,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.766526019690577e-06,
+      "loss": 1.2537,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.79465541490858e-06,
+      "loss": 1.2894,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.822784810126582e-06,
+      "loss": 1.2701,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.850914205344586e-06,
+      "loss": 1.2355,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.879043600562588e-06,
+      "loss": 1.1945,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.90717299578059e-06,
+      "loss": 1.2153,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.935302390998595e-06,
+      "loss": 1.1711,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.963431786216597e-06,
+      "loss": 1.232,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.9915611814346e-06,
+      "loss": 1.2496,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.019690576652602e-06,
+      "loss": 1.1544,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.047819971870606e-06,
+      "loss": 1.2223,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.075949367088608e-06,
+      "loss": 1.2937,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.10407876230661e-06,
+      "loss": 1.1893,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.132208157524615e-06,
+      "loss": 1.2323,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.160337552742617e-06,
+      "loss": 1.2714,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.18846694796062e-06,
+      "loss": 1.2396,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.216596343178622e-06,
+      "loss": 1.3389,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.244725738396626e-06,
+      "loss": 1.2821,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.272855133614628e-06,
+      "loss": 1.1103,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.30098452883263e-06,
+      "loss": 1.3031,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.329113924050634e-06,
+      "loss": 1.2566,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.357243319268637e-06,
+      "loss": 1.2875,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.38537271448664e-06,
+      "loss": 1.2486,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.413502109704642e-06,
+      "loss": 1.2143,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.441631504922646e-06,
+      "loss": 1.2036,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.469760900140648e-06,
+      "loss": 1.1991,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.49789029535865e-06,
+      "loss": 1.2245,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.526019690576653e-06,
+      "loss": 1.222,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.554149085794655e-06,
+      "loss": 1.2478,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.582278481012659e-06,
+      "loss": 1.1796,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.610407876230661e-06,
+      "loss": 1.3054,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.638537271448664e-06,
+      "loss": 1.2382,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.666666666666666e-06,
+      "loss": 1.3028,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.69479606188467e-06,
+      "loss": 1.305,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.722925457102673e-06,
+      "loss": 1.217,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.751054852320675e-06,
+      "loss": 1.2405,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.779184247538679e-06,
+      "loss": 1.2145,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.807313642756681e-06,
+      "loss": 1.2346,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.835443037974684e-06,
+      "loss": 1.2747,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.863572433192686e-06,
+      "loss": 1.2899,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.89170182841069e-06,
+      "loss": 1.269,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.919831223628692e-06,
+      "loss": 1.1788,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947960618846695e-06,
+      "loss": 1.2185,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.976090014064699e-06,
+      "loss": 1.1465,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.004219409282701e-06,
+      "loss": 1.241,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.032348804500704e-06,
+      "loss": 1.2343,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.060478199718706e-06,
+      "loss": 1.2646,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.08860759493671e-06,
+      "loss": 1.1439,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.116736990154712e-06,
+      "loss": 1.2434,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.144866385372715e-06,
+      "loss": 1.261,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.172995780590719e-06,
+      "loss": 1.1933,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.201125175808721e-06,
+      "loss": 1.2022,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.229254571026723e-06,
+      "loss": 1.2399,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.257383966244726e-06,
+      "loss": 1.1767,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.28551336146273e-06,
+      "loss": 1.2522,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.313642756680732e-06,
+      "loss": 1.2299,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.341772151898735e-06,
+      "loss": 1.2843,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.369901547116739e-06,
+      "loss": 1.1955,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.398030942334741e-06,
+      "loss": 1.208,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.426160337552743e-06,
+      "loss": 1.207,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.454289732770746e-06,
+      "loss": 1.249,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.48241912798875e-06,
+      "loss": 1.2039,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.510548523206752e-06,
+      "loss": 1.1412,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.538677918424754e-06,
+      "loss": 1.2645,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.566807313642758e-06,
+      "loss": 1.2025,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.59493670886076e-06,
+      "loss": 1.2327,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.623066104078763e-06,
+      "loss": 1.2268,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.651195499296766e-06,
+      "loss": 1.2477,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.679324894514768e-06,
+      "loss": 1.3019,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.70745428973277e-06,
+      "loss": 1.2737,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.735583684950773e-06,
+      "loss": 1.2277,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.763713080168777e-06,
+      "loss": 1.2112,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.791842475386779e-06,
+      "loss": 1.1756,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.819971870604781e-06,
+      "loss": 1.1923,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.848101265822785e-06,
+      "loss": 1.1391,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.876230661040788e-06,
+      "loss": 1.317,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.90436005625879e-06,
+      "loss": 1.3603,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.932489451476792e-06,
+      "loss": 1.2067,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.960618846694797e-06,
+      "loss": 1.2314,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.988748241912799e-06,
+      "loss": 1.2454,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0016877637130801e-05,
+      "loss": 1.1377,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0045007032348805e-05,
+      "loss": 1.2378,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0073136427566808e-05,
+      "loss": 1.1205,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.010126582278481e-05,
+      "loss": 1.1929,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0129395218002812e-05,
+      "loss": 1.2513,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0157524613220816e-05,
+      "loss": 1.2535,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0185654008438819e-05,
+      "loss": 1.2595,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0213783403656821e-05,
+      "loss": 1.2246,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0241912798874825e-05,
+      "loss": 1.2205,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0270042194092827e-05,
+      "loss": 1.2495,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.029817158931083e-05,
+      "loss": 1.1857,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0326300984528832e-05,
+      "loss": 1.1286,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0354430379746836e-05,
+      "loss": 1.2546,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0382559774964839e-05,
+      "loss": 1.3001,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0410689170182841e-05,
+      "loss": 1.1356,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0438818565400845e-05,
+      "loss": 1.2536,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0466947960618847e-05,
+      "loss": 1.2064,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.049507735583685e-05,
+      "loss": 1.1471,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0523206751054852e-05,
+      "loss": 1.2652,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0551336146272856e-05,
+      "loss": 1.1523,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0579465541490858e-05,
+      "loss": 1.2483,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.060759493670886e-05,
+      "loss": 1.2607,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0635724331926865e-05,
+      "loss": 1.2337,
+      "step": 307
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0663853727144867e-05,
+      "loss": 1.1772,
+      "step": 308
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.069198312236287e-05,
+      "loss": 1.1757,
+      "step": 309
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0720112517580872e-05,
+      "loss": 1.2011,
+      "step": 310
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0748241912798876e-05,
+      "loss": 1.1965,
+      "step": 311
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0776371308016878e-05,
+      "loss": 1.1214,
+      "step": 312
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.080450070323488e-05,
+      "loss": 1.1906,
+      "step": 313
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0832630098452883e-05,
+      "loss": 1.2178,
+      "step": 314
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0860759493670885e-05,
+      "loss": 1.1865,
+      "step": 315
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0888888888888888e-05,
+      "loss": 1.2486,
+      "step": 316
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0917018284106892e-05,
+      "loss": 1.2414,
+      "step": 317
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0945147679324894e-05,
+      "loss": 1.1916,
+      "step": 318
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0973277074542897e-05,
+      "loss": 1.168,
+      "step": 319
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.10014064697609e-05,
+      "loss": 1.2392,
+      "step": 320
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1029535864978903e-05,
+      "loss": 1.2052,
+      "step": 321
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1057665260196905e-05,
+      "loss": 1.3188,
+      "step": 322
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1085794655414908e-05,
+      "loss": 1.0825,
+      "step": 323
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1113924050632912e-05,
+      "loss": 1.1842,
+      "step": 324
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1142053445850914e-05,
+      "loss": 1.2982,
+      "step": 325
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1170182841068916e-05,
+      "loss": 1.2925,
+      "step": 326
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.119831223628692e-05,
+      "loss": 1.274,
+      "step": 327
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1226441631504923e-05,
+      "loss": 1.2011,
+      "step": 328
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1254571026722925e-05,
+      "loss": 1.1878,
+      "step": 329
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.128270042194093e-05,
+      "loss": 1.2362,
+      "step": 330
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1310829817158932e-05,
+      "loss": 1.2162,
+      "step": 331
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1338959212376934e-05,
+      "loss": 1.2247,
+      "step": 332
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1367088607594936e-05,
+      "loss": 1.157,
+      "step": 333
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.139521800281294e-05,
+      "loss": 1.2381,
+      "step": 334
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1423347398030943e-05,
+      "loss": 1.2022,
+      "step": 335
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1451476793248945e-05,
+      "loss": 1.236,
+      "step": 336
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1479606188466949e-05,
+      "loss": 1.2365,
+      "step": 337
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1507735583684951e-05,
+      "loss": 1.1941,
+      "step": 338
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1535864978902954e-05,
+      "loss": 1.1612,
+      "step": 339
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1563994374120956e-05,
+      "loss": 1.1502,
+      "step": 340
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.159212376933896e-05,
+      "loss": 1.2438,
+      "step": 341
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1620253164556963e-05,
+      "loss": 1.2748,
+      "step": 342
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1648382559774965e-05,
+      "loss": 1.1745,
+      "step": 343
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1676511954992969e-05,
+      "loss": 1.2505,
+      "step": 344
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1704641350210971e-05,
+      "loss": 1.2394,
+      "step": 345
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1732770745428974e-05,
+      "loss": 1.23,
+      "step": 346
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1760900140646976e-05,
+      "loss": 1.0944,
+      "step": 347
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.178902953586498e-05,
+      "loss": 1.2037,
+      "step": 348
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1817158931082982e-05,
+      "loss": 1.1946,
+      "step": 349
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1845288326300985e-05,
+      "loss": 1.1758,
+      "step": 350
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1873417721518989e-05,
+      "loss": 1.1969,
+      "step": 351
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1901547116736991e-05,
+      "loss": 1.2282,
+      "step": 352
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1929676511954994e-05,
+      "loss": 1.2487,
+      "step": 353
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1957805907172996e-05,
+      "loss": 1.1098,
+      "step": 354
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1985935302391e-05,
+      "loss": 1.2088,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2014064697609002e-05,
+      "loss": 1.217,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2042194092827005e-05,
+      "loss": 1.1628,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2070323488045009e-05,
+      "loss": 1.1727,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.209845288326301e-05,
+      "loss": 1.1832,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2126582278481013e-05,
+      "loss": 1.2679,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2154711673699016e-05,
+      "loss": 1.1567,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.218284106891702e-05,
+      "loss": 1.1671,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.221097046413502e-05,
+      "loss": 1.2253,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2239099859353025e-05,
+      "loss": 1.2175,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2267229254571027e-05,
+      "loss": 1.161,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2295358649789031e-05,
+      "loss": 1.2028,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2323488045007032e-05,
+      "loss": 1.1904,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2351617440225036e-05,
+      "loss": 1.1772,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2379746835443038e-05,
+      "loss": 1.2487,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2407876230661042e-05,
+      "loss": 1.2474,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2436005625879044e-05,
+      "loss": 1.1885,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2464135021097048e-05,
+      "loss": 1.2835,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2492264416315049e-05,
+      "loss": 1.2169,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2520393811533053e-05,
+      "loss": 1.2187,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2548523206751056e-05,
+      "loss": 1.2593,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.257665260196906e-05,
+      "loss": 1.1943,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.260478199718706e-05,
+      "loss": 1.141,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2632911392405063e-05,
+      "loss": 1.2606,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2661040787623067e-05,
+      "loss": 1.258,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2689170182841069e-05,
+      "loss": 1.2653,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2717299578059071e-05,
+      "loss": 1.2217,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2745428973277074e-05,
+      "loss": 1.1253,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2773558368495078e-05,
+      "loss": 1.1497,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.280168776371308e-05,
+      "loss": 1.1594,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2829817158931084e-05,
+      "loss": 1.1385,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2857946554149085e-05,
+      "loss": 1.2111,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2886075949367089e-05,
+      "loss": 1.238,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2914205344585091e-05,
+      "loss": 1.2676,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2942334739803095e-05,
+      "loss": 1.1621,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2970464135021096e-05,
+      "loss": 1.249,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.29985935302391e-05,
+      "loss": 1.2212,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3026722925457102e-05,
+      "loss": 1.2483,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3054852320675106e-05,
+      "loss": 1.1625,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3082981715893109e-05,
+      "loss": 1.1985,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3111111111111113e-05,
+      "loss": 1.2382,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3139240506329113e-05,
+      "loss": 1.2504,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3167369901547118e-05,
+      "loss": 1.1457,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.319549929676512e-05,
+      "loss": 1.2145,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3223628691983124e-05,
+      "loss": 1.2117,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3251758087201125e-05,
+      "loss": 1.2273,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3279887482419129e-05,
+      "loss": 1.1969,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3308016877637131e-05,
+      "loss": 1.2126,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3336146272855135e-05,
+      "loss": 1.2277,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3364275668073136e-05,
+      "loss": 1.168,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.339240506329114e-05,
+      "loss": 1.2129,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3420534458509142e-05,
+      "loss": 1.1802,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3448663853727146e-05,
+      "loss": 1.1963,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3476793248945149e-05,
+      "loss": 1.2446,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3504922644163153e-05,
+      "loss": 1.2345,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3533052039381153e-05,
+      "loss": 1.2235,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3561181434599157e-05,
+      "loss": 1.2546,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.358931082981716e-05,
+      "loss": 1.2472,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3617440225035164e-05,
+      "loss": 1.1675,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3645569620253164e-05,
+      "loss": 1.1527,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3673699015471168e-05,
+      "loss": 1.2735,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.370182841068917e-05,
+      "loss": 1.223,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3729957805907175e-05,
+      "loss": 1.2249,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3758087201125175e-05,
+      "loss": 1.21,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3786216596343178e-05,
+      "loss": 1.2037,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3814345991561182e-05,
+      "loss": 1.1935,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3842475386779184e-05,
+      "loss": 1.1674,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3870604781997188e-05,
+      "loss": 1.2095,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3898734177215189e-05,
+      "loss": 1.2487,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3926863572433193e-05,
+      "loss": 1.1532,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3954992967651195e-05,
+      "loss": 1.1411,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.39831223628692e-05,
+      "loss": 1.199,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.40112517580872e-05,
+      "loss": 1.1762,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4039381153305204e-05,
+      "loss": 1.2078,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4067510548523206e-05,
+      "loss": 1.2467,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.409563994374121e-05,
+      "loss": 1.2361,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4123769338959213e-05,
+      "loss": 1.1892,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4151898734177215e-05,
+      "loss": 1.1837,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4180028129395218e-05,
+      "loss": 1.2333,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4208157524613222e-05,
+      "loss": 1.2832,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4236286919831224e-05,
+      "loss": 1.2077,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4264416315049228e-05,
+      "loss": 1.2341,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4292545710267229e-05,
+      "loss": 1.2107,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4320675105485233e-05,
+      "loss": 1.2006,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4348804500703235e-05,
+      "loss": 1.2046,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4376933895921239e-05,
+      "loss": 1.2037,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.440506329113924e-05,
+      "loss": 1.2076,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4433192686357244e-05,
+      "loss": 1.2879,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4461322081575246e-05,
+      "loss": 1.2212,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.448945147679325e-05,
+      "loss": 1.193,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4517580872011253e-05,
+      "loss": 1.2328,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4545710267229257e-05,
+      "loss": 1.1685,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4573839662447257e-05,
+      "loss": 1.2075,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4601969057665261e-05,
+      "loss": 1.211,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4630098452883264e-05,
+      "loss": 1.2127,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4658227848101268e-05,
+      "loss": 1.1411,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4686357243319268e-05,
+      "loss": 1.2752,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4714486638537272e-05,
+      "loss": 1.2524,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4742616033755275e-05,
+      "loss": 1.2364,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4770745428973279e-05,
+      "loss": 1.1975,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.479887482419128e-05,
+      "loss": 1.2057,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4827004219409284e-05,
+      "loss": 1.2362,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4855133614627286e-05,
+      "loss": 1.2457,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.488326300984529e-05,
+      "loss": 1.1731,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4911392405063292e-05,
+      "loss": 1.2027,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4939521800281296e-05,
+      "loss": 1.1769,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4967651195499297e-05,
+      "loss": 1.2367,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.49957805907173e-05,
+      "loss": 1.2338,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5023909985935303e-05,
+      "loss": 1.1758,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5052039381153304e-05,
+      "loss": 1.1712,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5080168776371308e-05,
+      "loss": 1.1593,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.510829817158931e-05,
+      "loss": 1.2047,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5136427566807315e-05,
+      "loss": 1.172,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5164556962025317e-05,
+      "loss": 1.211,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.519268635724332e-05,
+      "loss": 1.3282,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5220815752461322e-05,
+      "loss": 1.2521,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5248945147679326e-05,
+      "loss": 1.1736,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.527707454289733e-05,
+      "loss": 1.1656,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5305203938115334e-05,
+      "loss": 1.1817,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5333333333333334e-05,
+      "loss": 1.2386,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.536146272855134e-05,
+      "loss": 1.1947,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.538959212376934e-05,
+      "loss": 1.1798,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5417721518987343e-05,
+      "loss": 1.2103,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5445850914205344e-05,
+      "loss": 1.225,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5473980309423348e-05,
+      "loss": 1.1858,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5502109704641352e-05,
+      "loss": 1.0966,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5530239099859356e-05,
+      "loss": 1.2525,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5558368495077357e-05,
+      "loss": 1.2124,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.558649789029536e-05,
+      "loss": 1.1662,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.561462728551336e-05,
+      "loss": 1.1642,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5642756680731365e-05,
+      "loss": 1.1776,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.567088607594937e-05,
+      "loss": 1.2129,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5699015471167374e-05,
+      "loss": 1.2101,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5727144866385374e-05,
+      "loss": 1.2277,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5755274261603378e-05,
+      "loss": 1.1968,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.578340365682138e-05,
+      "loss": 1.2375,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5811533052039383e-05,
+      "loss": 1.2068,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5839662447257384e-05,
+      "loss": 1.1977,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5867791842475388e-05,
+      "loss": 1.2232,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5895921237693392e-05,
+      "loss": 1.2745,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5924050632911396e-05,
+      "loss": 1.3067,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5952180028129396e-05,
+      "loss": 1.2067,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.59803094233474e-05,
+      "loss": 1.2138,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.60084388185654e-05,
+      "loss": 1.2345,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6036568213783405e-05,
+      "loss": 1.2369,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.606469760900141e-05,
+      "loss": 1.2381,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6092827004219413e-05,
+      "loss": 1.1185,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6120956399437414e-05,
+      "loss": 1.223,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6149085794655415e-05,
+      "loss": 1.1658,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.617721518987342e-05,
+      "loss": 1.2058,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.620534458509142e-05,
+      "loss": 1.2548,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6233473980309423e-05,
+      "loss": 1.1756,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6261603375527427e-05,
+      "loss": 1.1279,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.628973277074543e-05,
+      "loss": 1.2045,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6317862165963432e-05,
+      "loss": 1.274,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6345991561181436e-05,
+      "loss": 1.2094,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6374120956399437e-05,
+      "loss": 1.1258,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.640225035161744e-05,
+      "loss": 1.2026,
+      "step": 512
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6430379746835445e-05,
+      "loss": 1.2064,
+      "step": 513
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.645850914205345e-05,
+      "loss": 1.1746,
+      "step": 514
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648663853727145e-05,
+      "loss": 1.1419,
+      "step": 515
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6514767932489454e-05,
+      "loss": 1.2334,
+      "step": 516
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6542897327707454e-05,
+      "loss": 1.2332,
+      "step": 517
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.657102672292546e-05,
+      "loss": 1.2326,
+      "step": 518
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.659915611814346e-05,
+      "loss": 1.2113,
+      "step": 519
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6627285513361463e-05,
+      "loss": 1.1487,
+      "step": 520
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6655414908579467e-05,
+      "loss": 1.2403,
+      "step": 521
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.668354430379747e-05,
+      "loss": 1.1828,
+      "step": 522
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6711673699015472e-05,
+      "loss": 1.2205,
+      "step": 523
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6739803094233476e-05,
+      "loss": 1.2251,
+      "step": 524
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6767932489451477e-05,
+      "loss": 1.2074,
+      "step": 525
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.679606188466948e-05,
+      "loss": 1.237,
+      "step": 526
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6824191279887485e-05,
+      "loss": 1.1709,
+      "step": 527
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.685232067510549e-05,
+      "loss": 1.151,
+      "step": 528
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.688045007032349e-05,
+      "loss": 1.2377,
+      "step": 529
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6908579465541493e-05,
+      "loss": 1.1149,
+      "step": 530
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6936708860759494e-05,
+      "loss": 1.2434,
+      "step": 531
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6964838255977498e-05,
+      "loss": 1.2551,
+      "step": 532
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6992967651195502e-05,
+      "loss": 1.2043,
+      "step": 533
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7021097046413503e-05,
+      "loss": 1.1923,
+      "step": 534
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7049226441631507e-05,
+      "loss": 1.3175,
+      "step": 535
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.707735583684951e-05,
+      "loss": 1.1757,
+      "step": 536
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.710548523206751e-05,
+      "loss": 1.1966,
+      "step": 537
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7133614627285516e-05,
+      "loss": 1.2231,
+      "step": 538
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7161744022503516e-05,
+      "loss": 1.2254,
+      "step": 539
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.718987341772152e-05,
+      "loss": 1.2002,
+      "step": 540
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7218002812939524e-05,
+      "loss": 1.2828,
+      "step": 541
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.724613220815753e-05,
+      "loss": 1.1746,
+      "step": 542
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.727426160337553e-05,
+      "loss": 1.0737,
+      "step": 543
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7302390998593533e-05,
+      "loss": 1.1748,
+      "step": 544
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7330520393811534e-05,
+      "loss": 1.0934,
+      "step": 545
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7358649789029538e-05,
+      "loss": 1.1801,
+      "step": 546
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7386779184247542e-05,
+      "loss": 1.1517,
+      "step": 547
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7414908579465543e-05,
+      "loss": 1.1653,
+      "step": 548
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7443037974683547e-05,
+      "loss": 1.2643,
+      "step": 549
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7471167369901547e-05,
+      "loss": 1.1251,
+      "step": 550
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.749929676511955e-05,
+      "loss": 1.1587,
+      "step": 551
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7527426160337555e-05,
+      "loss": 1.2191,
+      "step": 552
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.755555555555556e-05,
+      "loss": 1.2744,
+      "step": 553
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.758368495077356e-05,
+      "loss": 1.1878,
+      "step": 554
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7611814345991564e-05,
+      "loss": 1.1773,
+      "step": 555
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7639943741209565e-05,
+      "loss": 1.1479,
+      "step": 556
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.766807313642757e-05,
+      "loss": 1.2154,
+      "step": 557
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7696202531645573e-05,
+      "loss": 1.2788,
+      "step": 558
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7724331926863574e-05,
+      "loss": 1.2194,
+      "step": 559
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7752461322081578e-05,
+      "loss": 1.1739,
+      "step": 560
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7780590717299582e-05,
+      "loss": 1.227,
+      "step": 561
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7808720112517582e-05,
+      "loss": 1.2561,
+      "step": 562
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7836849507735586e-05,
+      "loss": 1.2027,
+      "step": 563
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7864978902953587e-05,
+      "loss": 1.1742,
+      "step": 564
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.789310829817159e-05,
+      "loss": 1.2356,
+      "step": 565
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7921237693389595e-05,
+      "loss": 1.0978,
+      "step": 566
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.79493670886076e-05,
+      "loss": 1.2479,
+      "step": 567
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.79774964838256e-05,
+      "loss": 1.1242,
+      "step": 568
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8005625879043604e-05,
+      "loss": 1.2042,
+      "step": 569
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8033755274261605e-05,
+      "loss": 1.1459,
+      "step": 570
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.806188466947961e-05,
+      "loss": 1.2426,
+      "step": 571
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8090014064697613e-05,
+      "loss": 1.154,
+      "step": 572
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8118143459915613e-05,
+      "loss": 1.1821,
+      "step": 573
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8146272855133617e-05,
+      "loss": 1.151,
+      "step": 574
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.817440225035162e-05,
+      "loss": 1.2206,
+      "step": 575
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8202531645569622e-05,
+      "loss": 1.2065,
+      "step": 576
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8230661040787626e-05,
+      "loss": 1.2702,
+      "step": 577
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8258790436005627e-05,
+      "loss": 1.2185,
+      "step": 578
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.828691983122363e-05,
+      "loss": 1.1821,
+      "step": 579
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8315049226441635e-05,
+      "loss": 1.2163,
+      "step": 580
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.834317862165964e-05,
+      "loss": 1.2312,
+      "step": 581
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.837130801687764e-05,
+      "loss": 1.3219,
+      "step": 582
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8399437412095644e-05,
+      "loss": 1.1961,
+      "step": 583
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8427566807313644e-05,
+      "loss": 1.2581,
+      "step": 584
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.845569620253165e-05,
+      "loss": 1.2009,
+      "step": 585
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8483825597749652e-05,
+      "loss": 1.2207,
+      "step": 586
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8511954992967653e-05,
+      "loss": 1.1577,
+      "step": 587
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8540084388185657e-05,
+      "loss": 1.1855,
+      "step": 588
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8568213783403658e-05,
+      "loss": 1.2373,
+      "step": 589
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8596343178621662e-05,
+      "loss": 1.2019,
+      "step": 590
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8624472573839663e-05,
+      "loss": 1.1734,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8652601969057667e-05,
+      "loss": 1.2075,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.868073136427567e-05,
+      "loss": 1.1884,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8708860759493675e-05,
+      "loss": 1.199,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8736990154711675e-05,
+      "loss": 1.2285,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.876511954992968e-05,
+      "loss": 1.2556,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.879324894514768e-05,
+      "loss": 1.2085,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8821378340365684e-05,
+      "loss": 1.1946,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8849507735583688e-05,
+      "loss": 1.258,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8877637130801692e-05,
+      "loss": 1.0542,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8905766526019693e-05,
+      "loss": 1.2004,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8933895921237697e-05,
+      "loss": 1.2088,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8962025316455698e-05,
+      "loss": 1.2035,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.89901547116737e-05,
+      "loss": 1.2464,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9018284106891702e-05,
+      "loss": 1.1954,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9046413502109706e-05,
+      "loss": 1.2435,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.907454289732771e-05,
+      "loss": 1.1363,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9102672292545714e-05,
+      "loss": 1.3111,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130801687763715e-05,
+      "loss": 1.1952,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.915893108298172e-05,
+      "loss": 1.2287,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918706047819972e-05,
+      "loss": 1.3014,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9215189873417724e-05,
+      "loss": 1.1963,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9243319268635728e-05,
+      "loss": 1.1524,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9271448663853732e-05,
+      "loss": 1.1859,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9299578059071733e-05,
+      "loss": 1.1875,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9327707454289737e-05,
+      "loss": 1.1537,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9355836849507737e-05,
+      "loss": 1.2278,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.938396624472574e-05,
+      "loss": 1.184,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9412095639943742e-05,
+      "loss": 1.1164,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9440225035161746e-05,
+      "loss": 1.2149,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.946835443037975e-05,
+      "loss": 1.2176,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9496483825597754e-05,
+      "loss": 1.176,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9524613220815755e-05,
+      "loss": 1.1403,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.955274261603376e-05,
+      "loss": 1.2759,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.958087201125176e-05,
+      "loss": 1.1784,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9609001406469764e-05,
+      "loss": 1.2194,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9637130801687768e-05,
+      "loss": 1.2149,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.966526019690577e-05,
+      "loss": 1.2106,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9693389592123772e-05,
+      "loss": 1.2784,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9721518987341773e-05,
+      "loss": 1.1774,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9749648382559777e-05,
+      "loss": 1.1997,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9777777777777778e-05,
+      "loss": 1.213,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9805907172995782e-05,
+      "loss": 1.2791,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9834036568213786e-05,
+      "loss": 1.2667,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986216596343179e-05,
+      "loss": 1.169,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.989029535864979e-05,
+      "loss": 1.2439,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9918424753867795e-05,
+      "loss": 1.2611,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9946554149085795e-05,
+      "loss": 1.2257,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99746835443038e-05,
+      "loss": 1.24,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0002812939521803e-05,
+      "loss": 1.1728,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0030942334739807e-05,
+      "loss": 1.1943,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0059071729957808e-05,
+      "loss": 1.2422,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0087201125175812e-05,
+      "loss": 1.1744,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0115330520393813e-05,
+      "loss": 1.2197,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0143459915611817e-05,
+      "loss": 1.1529,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0171589310829818e-05,
+      "loss": 1.2153,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.019971870604782e-05,
+      "loss": 1.1647,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0227848101265826e-05,
+      "loss": 1.1862,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.025597749648383e-05,
+      "loss": 1.2105,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.028410689170183e-05,
+      "loss": 1.1939,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0312236286919834e-05,
+      "loss": 1.2047,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0340365682137835e-05,
+      "loss": 1.2312,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.036849507735584e-05,
+      "loss": 1.2229,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0396624472573843e-05,
+      "loss": 1.2222,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0424753867791847e-05,
+      "loss": 1.2106,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0452883263009848e-05,
+      "loss": 1.17,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0481012658227852e-05,
+      "loss": 1.1897,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0509142053445853e-05,
+      "loss": 1.145,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0537271448663857e-05,
+      "loss": 1.1863,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.056540084388186e-05,
+      "loss": 1.2619,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.059353023909986e-05,
+      "loss": 1.2558,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0621659634317865e-05,
+      "loss": 1.226,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.064978902953587e-05,
+      "loss": 1.1768,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.067791842475387e-05,
+      "loss": 1.1911,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0706047819971874e-05,
+      "loss": 1.2352,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0734177215189875e-05,
+      "loss": 1.1733,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.076230661040788e-05,
+      "loss": 1.1538,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0790436005625883e-05,
+      "loss": 1.1037,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0818565400843887e-05,
+      "loss": 1.1819,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0846694796061888e-05,
+      "loss": 1.1994,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0874824191279888e-05,
+      "loss": 1.1938,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0902953586497892e-05,
+      "loss": 1.1987,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0931082981715896e-05,
+      "loss": 1.2208,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.09592123769339e-05,
+      "loss": 1.1696,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.09873417721519e-05,
+      "loss": 1.3434,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1015471167369905e-05,
+      "loss": 1.1991,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1043600562587906e-05,
+      "loss": 1.2216,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.107172995780591e-05,
+      "loss": 1.1238,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.109985935302391e-05,
+      "loss": 1.1799,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1127988748241915e-05,
+      "loss": 1.2602,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.115611814345992e-05,
+      "loss": 1.2103,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1184247538677923e-05,
+      "loss": 1.1088,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1212376933895923e-05,
+      "loss": 1.1414,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1240506329113927e-05,
+      "loss": 1.243,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1268635724331928e-05,
+      "loss": 1.208,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1296765119549932e-05,
+      "loss": 1.1666,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1324894514767936e-05,
+      "loss": 1.2497,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.135302390998594e-05,
+      "loss": 1.259,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.138115330520394e-05,
+      "loss": 1.1537,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1409282700421945e-05,
+      "loss": 1.2632,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1437412095639946e-05,
+      "loss": 1.1581,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.146554149085795e-05,
+      "loss": 1.2236,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.149367088607595e-05,
+      "loss": 1.1497,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1521800281293954e-05,
+      "loss": 1.2693,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.154992967651196e-05,
+      "loss": 1.2406,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1578059071729962e-05,
+      "loss": 1.2272,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1606188466947963e-05,
+      "loss": 1.1634,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1634317862165967e-05,
+      "loss": 1.1804,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1662447257383968e-05,
+      "loss": 1.2197,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1690576652601972e-05,
+      "loss": 1.095,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1718706047819976e-05,
+      "loss": 1.256,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.174683544303798e-05,
+      "loss": 1.2458,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.177496483825598e-05,
+      "loss": 1.158,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1803094233473985e-05,
+      "loss": 1.1551,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1831223628691985e-05,
+      "loss": 1.3003,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.185935302390999e-05,
+      "loss": 1.2751,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.188748241912799e-05,
+      "loss": 1.2192,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1915611814345994e-05,
+      "loss": 1.2106,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1943741209563998e-05,
+      "loss": 1.1488,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1971870604782002e-05,
+      "loss": 1.2184,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 1.1636,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999990636473e-05,
+      "loss": 1.1926,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999962545891e-05,
+      "loss": 1.1725,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999999157282556e-05,
+      "loss": 1.1963,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999850183567e-05,
+      "loss": 1.269,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999765911827e-05,
+      "loss": 1.237,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999996629130362e-05,
+      "loss": 1.177,
+      "step": 717
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999995411871975e-05,
+      "loss": 1.1844,
+      "step": 718
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999994007343125e-05,
+      "loss": 1.2005,
+      "step": 719
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999992415543845e-05,
+      "loss": 1.2356,
+      "step": 720
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199999063647416e-05,
+      "loss": 1.1436,
+      "step": 721
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999988670134104e-05,
+      "loss": 1.1822,
+      "step": 722
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999986516523715e-05,
+      "loss": 1.2045,
+      "step": 723
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199998417564303e-05,
+      "loss": 1.2261,
+      "step": 724
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999981647492098e-05,
+      "loss": 1.1754,
+      "step": 725
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999978932070965e-05,
+      "loss": 1.1938,
+      "step": 726
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999976029379677e-05,
+      "loss": 1.2612,
+      "step": 727
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999972939418295e-05,
+      "loss": 1.1664,
+      "step": 728
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999969662186873e-05,
+      "loss": 1.2261,
+      "step": 729
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999966197685472e-05,
+      "loss": 1.2467,
+      "step": 730
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199996254591416e-05,
+      "loss": 1.2376,
+      "step": 731
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999958706873004e-05,
+      "loss": 1.2003,
+      "step": 732
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999954680562076e-05,
+      "loss": 1.1952,
+      "step": 733
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199995046698145e-05,
+      "loss": 1.1748,
+      "step": 734
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999946066131205e-05,
+      "loss": 1.1874,
+      "step": 735
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999941478011427e-05,
+      "loss": 1.1923,
+      "step": 736
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999936702622196e-05,
+      "loss": 1.2262,
+      "step": 737
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999931739963606e-05,
+      "loss": 1.2161,
+      "step": 738
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199992659003575e-05,
+      "loss": 1.157,
+      "step": 739
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999921252838718e-05,
+      "loss": 1.2021,
+      "step": 740
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199991572837262e-05,
+      "loss": 1.2115,
+      "step": 741
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999910016637553e-05,
+      "loss": 1.2644,
+      "step": 742
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999904117633622e-05,
+      "loss": 1.2401,
+      "step": 743
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999898031360945e-05,
+      "loss": 1.2581,
+      "step": 744
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999891757819627e-05,
+      "loss": 1.2048,
+      "step": 745
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999885297009794e-05,
+      "loss": 1.1902,
+      "step": 746
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999878648931564e-05,
+      "loss": 1.1482,
+      "step": 747
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199987181358506e-05,
+      "loss": 1.1816,
+      "step": 748
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199986479097041e-05,
+      "loss": 1.2164,
+      "step": 749
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999857581087744e-05,
+      "loss": 1.1545,
+      "step": 750
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19998501839372e-05,
+      "loss": 1.2581,
+      "step": 751
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999842599518923e-05,
+      "loss": 1.2717,
+      "step": 752
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199983482783304e-05,
+      "loss": 1.1966,
+      "step": 753
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999826868879706e-05,
+      "loss": 1.143,
+      "step": 754
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199981872265907e-05,
+      "loss": 1.2056,
+      "step": 755
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999810389171283e-05,
+      "loss": 1.2475,
+      "step": 756
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19998018684165e-05,
+      "loss": 1.1629,
+      "step": 757
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199979316039488e-05,
+      "loss": 1.2456,
+      "step": 758
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199978426510659e-05,
+      "loss": 1.1585,
+      "step": 759
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999775182551793e-05,
+      "loss": 1.2949,
+      "step": 760
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999765912730662e-05,
+      "loss": 1.1958,
+      "step": 761
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999756455643368e-05,
+      "loss": 1.1873,
+      "step": 762
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199974681129009e-05,
+      "loss": 1.1885,
+      "step": 763
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999736979671006e-05,
+      "loss": 1.2057,
+      "step": 764
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19997269607863e-05,
+      "loss": 1.3298,
+      "step": 765
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999716754636167e-05,
+      "loss": 1.2069,
+      "step": 766
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199970636122079e-05,
+      "loss": 1.2245,
+      "step": 767
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999695780540365e-05,
+      "loss": 1.1619,
+      "step": 768
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999685012595093e-05,
+      "loss": 1.151,
+      "step": 769
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999674057385172e-05,
+      "loss": 1.2326,
+      "step": 770
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199966291491081e-05,
+      "loss": 1.1621,
+      "step": 771
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999651585172215e-05,
+      "loss": 1.2039,
+      "step": 772
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.19996400681696e-05,
+      "loss": 1.2215,
+      "step": 773
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999628363903178e-05,
+      "loss": 1.2007,
+      "step": 774
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999616472373167e-05,
+      "loss": 1.1734,
+      "step": 775
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999604393579798e-05,
+      "loss": 1.2435,
+      "step": 776
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199959212752329e-05,
+      "loss": 1.1755,
+      "step": 777
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999579674203875e-05,
+      "loss": 1.2103,
+      "step": 778
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999567033621782e-05,
+      "loss": 1.2079,
+      "step": 779
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999554205777255e-05,
+      "loss": 1.1943,
+      "step": 780
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199954119067053e-05,
+      "loss": 1.1857,
+      "step": 781
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199952798830185e-05,
+      "loss": 1.1211,
+      "step": 782
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999514598671466e-05,
+      "loss": 1.2137,
+      "step": 783
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999501021779624e-05,
+      "loss": 1.1784,
+      "step": 784
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199948725762658e-05,
+      "loss": 1.2786,
+      "step": 785
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999473306212596e-05,
+      "loss": 1.2382,
+      "step": 786
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999459167537926e-05,
+      "loss": 1.2293,
+      "step": 787
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199944484160284e-05,
+      "loss": 1.1226,
+      "step": 788
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999430328407606e-05,
+      "loss": 1.1744,
+      "step": 789
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999415627952493e-05,
+      "loss": 1.1711,
+      "step": 790
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999400740237778e-05,
+      "loss": 1.1243,
+      "step": 791
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199938566526374e-05,
+      "loss": 1.2326,
+      "step": 792
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199937040303066e-05,
+      "loss": 1.1567,
+      "step": 793
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999354953538825e-05,
+      "loss": 1.2233,
+      "step": 794
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199933931678852e-05,
+      "loss": 1.2292,
+      "step": 795
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199932349278005e-05,
+      "loss": 1.2662,
+      "step": 796
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999307481513694e-05,
+      "loss": 1.2343,
+      "step": 797
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199929128298977e-05,
+      "loss": 1.1457,
+      "step": 798
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999274897208567e-05,
+      "loss": 1.2051,
+      "step": 799
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999258324170398e-05,
+      "loss": 1.2503,
+      "step": 800
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999241563875576e-05,
+      "loss": 1.2798,
+      "step": 801
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999224616324408e-05,
+      "loss": 1.1424,
+      "step": 802
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199920748151721e-05,
+      "loss": 1.1604,
+      "step": 803
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199919015945432e-05,
+      "loss": 1.2186,
+      "step": 804
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199917265013604e-05,
+      "loss": 1.1809,
+      "step": 805
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199915495356271e-05,
+      "loss": 1.1714,
+      "step": 806
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999137069734663e-05,
+      "loss": 1.1532,
+      "step": 807
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999118998652228e-05,
+      "loss": 1.1746,
+      "step": 808
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999100740315745e-05,
+      "loss": 1.2566,
+      "step": 809
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999082294725556e-05,
+      "loss": 1.1489,
+      "step": 810
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199906366188201e-05,
+      "loss": 1.1134,
+      "step": 811
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199904484178545e-05,
+      "loss": 1.1731,
+      "step": 812
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999025834436235e-05,
+      "loss": 1.216,
+      "step": 813
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1999006639834713e-05,
+      "loss": 1.1668,
+      "step": 814
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998987257981253e-05,
+      "loss": 1.2231,
+      "step": 815
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199896768887621e-05,
+      "loss": 1.1888,
+      "step": 816
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998947932519953e-05,
+      "loss": 1.2148,
+      "step": 817
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998927988912854e-05,
+      "loss": 1.2336,
+      "step": 818
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199890785805528e-05,
+      "loss": 1.2132,
+      "step": 819
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.199888753994762e-05,
+      "loss": 1.2133,
+      "step": 820
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998867034590244e-05,
+      "loss": 1.1373,
+      "step": 821
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998846341983538e-05,
+      "loss": 1.1144,
+      "step": 822
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998825462127893e-05,
+      "loss": 1.1634,
+      "step": 823
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998804395023696e-05,
+      "loss": 1.2337,
+      "step": 824
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998783140671346e-05,
+      "loss": 1.2156,
+      "step": 825
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998761699071237e-05,
+      "loss": 1.1502,
+      "step": 826
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998740070223767e-05,
+      "loss": 1.2924,
+      "step": 827
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1998718254129355e-05,
+      "loss": 1.2245,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998696250788396e-05,
+      "loss": 1.1516,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998674060201304e-05,
+      "loss": 1.1819,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998651682368498e-05,
+      "loss": 1.1804,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19986291172904e-05,
+      "loss": 1.1562,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998606364967428e-05,
+      "loss": 1.1881,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998583425400007e-05,
+      "loss": 1.1932,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998560298588567e-05,
+      "loss": 1.1576,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998536984533545e-05,
+      "loss": 1.1344,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998513483235375e-05,
+      "loss": 1.1602,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998489794694496e-05,
+      "loss": 1.1968,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998465918911354e-05,
+      "loss": 1.2879,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998441855886395e-05,
+      "loss": 1.227,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199841760562007e-05,
+      "loss": 1.1849,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199839316811283e-05,
+      "loss": 1.2007,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199836854336514e-05,
+      "loss": 1.1607,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998343731377453e-05,
+      "loss": 1.1772,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199831873215024e-05,
+      "loss": 1.1842,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998293545683964e-05,
+      "loss": 1.1796,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19982681719791e-05,
+      "loss": 1.1496,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199824261103612e-05,
+      "loss": 1.2759,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199821686285551e-05,
+      "loss": 1.1934,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998190927437744e-05,
+      "loss": 1.1977,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998164804783314e-05,
+      "loss": 1.2378,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19981384948927e-05,
+      "loss": 1.1944,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998111997766403e-05,
+      "loss": 1.2553,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998085313404916e-05,
+      "loss": 1.1819,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998058441808744e-05,
+      "loss": 1.1676,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199803138297838e-05,
+      "loss": 1.1902,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1998004136914344e-05,
+      "loss": 1.2097,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997976703617133e-05,
+      "loss": 1.241,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997949083087267e-05,
+      "loss": 1.1864,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997921275325263e-05,
+      "loss": 1.2268,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199789328033164e-05,
+      "loss": 1.1956,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199786509810693e-05,
+      "loss": 1.2356,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199783672865165e-05,
+      "loss": 1.1694,
+      "step": 863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997808171966336e-05,
+      "loss": 1.1485,
+      "step": 864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997779428051522e-05,
+      "loss": 1.2177,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199775049690775e-05,
+      "loss": 1.1943,
+      "step": 866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997721378535557e-05,
+      "loss": 1.2613,
+      "step": 867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199769207293549e-05,
+      "loss": 1.1202,
+      "step": 868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19976625801081e-05,
+      "loss": 1.2411,
+      "step": 869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997632900053935e-05,
+      "loss": 1.117,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997603032773554e-05,
+      "loss": 1.174,
+      "step": 871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997572978267513e-05,
+      "loss": 1.2329,
+      "step": 872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997542736536377e-05,
+      "loss": 1.2473,
+      "step": 873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997512307580716e-05,
+      "loss": 1.2253,
+      "step": 874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997481691401098e-05,
+      "loss": 1.1632,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997450887998087e-05,
+      "loss": 1.1809,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997419897372274e-05,
+      "loss": 1.2167,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199738871952423e-05,
+      "loss": 1.2253,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997357354454543e-05,
+      "loss": 1.156,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997325802163798e-05,
+      "loss": 1.248,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997294062652588e-05,
+      "loss": 1.2372,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997262135921506e-05,
+      "loss": 1.1809,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997230021971153e-05,
+      "loss": 1.1226,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997197720802123e-05,
+      "loss": 1.1781,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997165232415027e-05,
+      "loss": 1.1649,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199713255681047e-05,
+      "loss": 1.2435,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997099693989073e-05,
+      "loss": 1.1884,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997066643951438e-05,
+      "loss": 1.2464,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1997033406698194e-05,
+      "loss": 1.1948,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996999982229957e-05,
+      "loss": 1.1754,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996966370547356e-05,
+      "loss": 1.1225,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199693257165102e-05,
+      "loss": 1.2276,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996898585541585e-05,
+      "loss": 1.1763,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199686441221968e-05,
+      "loss": 1.1743,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199683005168596e-05,
+      "loss": 1.216,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996795503941045e-05,
+      "loss": 1.2422,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996760768985603e-05,
+      "loss": 1.1751,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199672584682027e-05,
+      "loss": 1.1671,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199669073744571e-05,
+      "loss": 1.1949,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199665544086258e-05,
+      "loss": 1.1295,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996619957071534e-05,
+      "loss": 1.2344,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199658428607324e-05,
+      "loss": 1.2563,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199654842786837e-05,
+      "loss": 1.1076,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199651238245759e-05,
+      "loss": 1.2013,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996476149841575e-05,
+      "loss": 1.1576,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199643973002101e-05,
+      "loss": 1.2651,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996403122996567e-05,
+      "loss": 1.1696,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199636632876894e-05,
+      "loss": 1.229,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996329347338817e-05,
+      "loss": 1.2202,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996292178706885e-05,
+      "loss": 1.1566,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996254822873845e-05,
+      "loss": 1.2233,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996217279840395e-05,
+      "loss": 1.2173,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996179549607242e-05,
+      "loss": 1.2171,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996141632175085e-05,
+      "loss": 1.2006,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996103527544638e-05,
+      "loss": 1.1414,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199606523571662e-05,
+      "loss": 1.2082,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1996026756691735e-05,
+      "loss": 1.2326,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995988090470716e-05,
+      "loss": 1.1368,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199594923705428e-05,
+      "loss": 1.1915,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199591019644316e-05,
+      "loss": 1.1336,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995870968638085e-05,
+      "loss": 1.2475,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995831553639784e-05,
+      "loss": 1.1853,
+      "step": 922
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995791951449e-05,
+      "loss": 1.18,
+      "step": 923
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995752162066482e-05,
+      "loss": 1.2716,
+      "step": 924
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199571218549296e-05,
+      "loss": 1.1409,
+      "step": 925
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995672021729194e-05,
+      "loss": 1.173,
+      "step": 926
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995631670775933e-05,
+      "loss": 1.2063,
+      "step": 927
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995591132633927e-05,
+      "loss": 1.2238,
+      "step": 928
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995550407303944e-05,
+      "loss": 1.2206,
+      "step": 929
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995509494786745e-05,
+      "loss": 1.2156,
+      "step": 930
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199546839508309e-05,
+      "loss": 1.1898,
+      "step": 931
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995427108193755e-05,
+      "loss": 1.232,
+      "step": 932
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199538563411951e-05,
+      "loss": 1.2183,
+      "step": 933
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995343972861132e-05,
+      "loss": 1.192,
+      "step": 934
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995302124419407e-05,
+      "loss": 1.0992,
+      "step": 935
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995260088795107e-05,
+      "loss": 1.2218,
+      "step": 936
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199521786598903e-05,
+      "loss": 1.114,
+      "step": 937
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199517545600196e-05,
+      "loss": 1.2128,
+      "step": 938
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995132858834694e-05,
+      "loss": 1.2741,
+      "step": 939
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199509007448803e-05,
+      "loss": 1.2826,
+      "step": 940
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199504710296277e-05,
+      "loss": 1.2077,
+      "step": 941
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1995003944259716e-05,
+      "loss": 1.2378,
+      "step": 942
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199496059837968e-05,
+      "loss": 1.1749,
+      "step": 943
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994917065323467e-05,
+      "loss": 1.2127,
+      "step": 944
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994873345091897e-05,
+      "loss": 1.2324,
+      "step": 945
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199482943768579e-05,
+      "loss": 1.1866,
+      "step": 946
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199478534310597e-05,
+      "loss": 1.3118,
+      "step": 947
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994741061353253e-05,
+      "loss": 1.1937,
+      "step": 948
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994696592428478e-05,
+      "loss": 1.2545,
+      "step": 949
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994651936332474e-05,
+      "loss": 1.242,
+      "step": 950
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199460709306608e-05,
+      "loss": 1.1992,
+      "step": 951
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199456206263013e-05,
+      "loss": 1.1861,
+      "step": 952
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199451684502547e-05,
+      "loss": 1.2099,
+      "step": 953
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994471440252952e-05,
+      "loss": 1.2006,
+      "step": 954
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199442584831342e-05,
+      "loss": 1.2108,
+      "step": 955
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994380069207728e-05,
+      "loss": 1.1773,
+      "step": 956
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994334102936733e-05,
+      "loss": 1.2175,
+      "step": 957
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994287949501303e-05,
+      "loss": 1.1743,
+      "step": 958
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199424160890229e-05,
+      "loss": 1.2444,
+      "step": 959
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199419508114058e-05,
+      "loss": 1.1989,
+      "step": 960
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1994148366217024e-05,
+      "loss": 1.1649,
+      "step": 961
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199410146413251e-05,
+      "loss": 1.2588,
+      "step": 962
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199405437488791e-05,
+      "loss": 1.2389,
+      "step": 963
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199400709848411e-05,
+      "loss": 1.2015,
+      "step": 964
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993959634921993e-05,
+      "loss": 1.2628,
+      "step": 965
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993911984202445e-05,
+      "loss": 1.2099,
+      "step": 966
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199386414632637e-05,
+      "loss": 1.214,
+      "step": 967
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993816121294647e-05,
+      "loss": 1.1876,
+      "step": 968
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199376790910819e-05,
+      "loss": 1.1946,
+      "step": 969
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993719509767895e-05,
+      "loss": 1.1543,
+      "step": 970
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993670923274668e-05,
+      "loss": 1.1378,
+      "step": 971
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199362214962942e-05,
+      "loss": 1.2911,
+      "step": 972
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199357318883306e-05,
+      "loss": 1.1661,
+      "step": 973
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199352404088652e-05,
+      "loss": 1.1791,
+      "step": 974
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993474705790702e-05,
+      "loss": 1.1973,
+      "step": 975
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199342518354654e-05,
+      "loss": 1.2474,
+      "step": 976
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199337547415496e-05,
+      "loss": 1.1927,
+      "step": 977
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199332557761689e-05,
+      "loss": 1.1548,
+      "step": 978
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993275493933267e-05,
+      "loss": 1.1865,
+      "step": 979
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199322522310503e-05,
+      "loss": 1.1613,
+      "step": 980
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199317476513312e-05,
+      "loss": 1.2021,
+      "step": 981
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1993124120018474e-05,
+      "loss": 1.1346,
+      "step": 982
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199307328776205e-05,
+      "loss": 1.1169,
+      "step": 983
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19930222683648e-05,
+      "loss": 1.297,
+      "step": 984
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992971061827678e-05,
+      "loss": 1.1912,
+      "step": 985
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992919668151636e-05,
+      "loss": 1.1459,
+      "step": 986
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992868087337645e-05,
+      "loss": 1.1721,
+      "step": 987
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199281631938667e-05,
+      "loss": 1.1703,
+      "step": 988
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992764364299674e-05,
+      "loss": 1.2909,
+      "step": 989
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992712222077636e-05,
+      "loss": 1.2097,
+      "step": 990
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992659892721533e-05,
+      "loss": 1.2031,
+      "step": 991
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199260737623234e-05,
+      "loss": 1.2309,
+      "step": 992
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199255467261104e-05,
+      "loss": 1.1781,
+      "step": 993
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199250178185863e-05,
+      "loss": 1.1824,
+      "step": 994
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199244870397609e-05,
+      "loss": 1.2051,
+      "step": 995
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199239543896442e-05,
+      "loss": 1.2139,
+      "step": 996
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992341986824614e-05,
+      "loss": 1.2363,
+      "step": 997
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992288347557676e-05,
+      "loss": 1.1626,
+      "step": 998
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992234521164606e-05,
+      "loss": 1.126,
+      "step": 999
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992180507646416e-05,
+      "loss": 1.1941,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_text_loss": 0.9967461824417114,
+      "eval_text_runtime": 1.5746,
+      "eval_text_samples_per_second": 46.996,
+      "eval_text_steps_per_second": 1.27,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "eval_svit_loss": 1.2622443437576294,
+      "eval_svit_runtime": 5.7936,
+      "eval_svit_samples_per_second": 117.543,
+      "eval_svit_steps_per_second": 1.899,
+      "step": 1000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992126307004114e-05,
+      "loss": 1.1938,
+      "step": 1001
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1992071919238718e-05,
+      "loss": 1.1847,
+      "step": 1002
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199201734435125e-05,
+      "loss": 1.1628,
+      "step": 1003
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991962582342725e-05,
+      "loss": 1.1662,
+      "step": 1004
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991907633214172e-05,
+      "loss": 1.1636,
+      "step": 1005
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991852496966618e-05,
+      "loss": 1.1747,
+      "step": 1006
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991797173601097e-05,
+      "loss": 1.1404,
+      "step": 1007
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991741663118642e-05,
+      "loss": 1.3253,
+      "step": 1008
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19916859655203e-05,
+      "loss": 1.1835,
+      "step": 1009
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199163008080711e-05,
+      "loss": 1.1589,
+      "step": 1010
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991574008980115e-05,
+      "loss": 1.0729,
+      "step": 1011
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199151775004037e-05,
+      "loss": 1.1989,
+      "step": 1012
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991461303988928e-05,
+      "loss": 1.1801,
+      "step": 1013
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991404670826843e-05,
+      "loss": 1.1897,
+      "step": 1014
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991347850555177e-05,
+      "loss": 1.1681,
+      "step": 1015
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991290843174993e-05,
+      "loss": 1.2415,
+      "step": 1016
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991233648687363e-05,
+      "loss": 1.2083,
+      "step": 1017
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991176267093356e-05,
+      "loss": 1.1696,
+      "step": 1018
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991118698394043e-05,
+      "loss": 1.1675,
+      "step": 1019
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199106094259051e-05,
+      "loss": 1.2177,
+      "step": 1020
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1991002999683828e-05,
+      "loss": 1.2141,
+      "step": 1021
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990944869675087e-05,
+      "loss": 1.2337,
+      "step": 1022
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990886552565377e-05,
+      "loss": 1.1968,
+      "step": 1023
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990828048355792e-05,
+      "loss": 1.2175,
+      "step": 1024
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199076935704742e-05,
+      "loss": 1.1815,
+      "step": 1025
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199071047864137e-05,
+      "loss": 1.1245,
+      "step": 1026
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199065141313874e-05,
+      "loss": 1.2213,
+      "step": 1027
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990592160540632e-05,
+      "loss": 1.1348,
+      "step": 1028
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990532720848162e-05,
+      "loss": 1.1957,
+      "step": 1029
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990473094062435e-05,
+      "loss": 1.1474,
+      "step": 1030
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990413280184578e-05,
+      "loss": 1.211,
+      "step": 1031
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990353279215707e-05,
+      "loss": 1.1868,
+      "step": 1032
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990293091156943e-05,
+      "loss": 1.2526,
+      "step": 1033
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990232716009414e-05,
+      "loss": 1.1497,
+      "step": 1034
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990172153774253e-05,
+      "loss": 1.1767,
+      "step": 1035
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1990111404452593e-05,
+      "loss": 1.1939,
+      "step": 1036
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.199005046804557e-05,
+      "loss": 1.214,
+      "step": 1037
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198998934455433e-05,
+      "loss": 1.2414,
+      "step": 1038
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989928033980012e-05,
+      "loss": 1.2018,
+      "step": 1039
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989866536323764e-05,
+      "loss": 1.1452,
+      "step": 1040
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989804851586744e-05,
+      "loss": 1.194,
+      "step": 1041
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19897429797701e-05,
+      "loss": 1.1718,
+      "step": 1042
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989680920875e-05,
+      "loss": 1.2719,
+      "step": 1043
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198961867490259e-05,
+      "loss": 1.1668,
+      "step": 1044
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989556241854056e-05,
+      "loss": 1.2006,
+      "step": 1045
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198949362173055e-05,
+      "loss": 1.1541,
+      "step": 1046
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989430814533257e-05,
+      "loss": 1.1984,
+      "step": 1047
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198936782026334e-05,
+      "loss": 1.1201,
+      "step": 1048
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989304638921998e-05,
+      "loss": 1.1942,
+      "step": 1049
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989241270510395e-05,
+      "loss": 1.175,
+      "step": 1050
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989177715029727e-05,
+      "loss": 1.1647,
+      "step": 1051
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989113972481185e-05,
+      "loss": 1.1017,
+      "step": 1052
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1989050042865955e-05,
+      "loss": 1.1724,
+      "step": 1053
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988985926185243e-05,
+      "loss": 1.2058,
+      "step": 1054
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198892162244025e-05,
+      "loss": 1.134,
+      "step": 1055
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988857131632174e-05,
+      "loss": 1.243,
+      "step": 1056
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198879245376222e-05,
+      "loss": 1.199,
+      "step": 1057
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198872758883161e-05,
+      "loss": 1.1447,
+      "step": 1058
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988662536841555e-05,
+      "loss": 1.1958,
+      "step": 1059
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988597297793264e-05,
+      "loss": 1.1996,
+      "step": 1060
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.1988531871687972e-05,
+      "loss": 1.1589,
+      "step": 1061
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.19884662585269e-05,
+      "loss": 1.1982,
+      "step": 1062
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198840045831127e-05,
+      "loss": 1.1465,
+      "step": 1063
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198833447104232e-05,
+      "loss": 1.1635,
+      "step": 1064
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.198826829672129e-05,
+      "loss": 1.166,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198820193534941e-05,
+      "loss": 1.1505,
+      "step": 1066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198813538692793e-05,
+      "loss": 1.1824,
+      "step": 1067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198806865145809e-05,
+      "loss": 1.1844,
+      "step": 1068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1988001728941145e-05,
+      "loss": 1.1923,
+      "step": 1069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987934619378348e-05,
+      "loss": 1.2296,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198786732277095e-05,
+      "loss": 1.1459,
+      "step": 1071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987799839120223e-05,
+      "loss": 1.1898,
+      "step": 1072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987732168427416e-05,
+      "loss": 1.2184,
+      "step": 1073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987664310693807e-05,
+      "loss": 1.199,
+      "step": 1074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987596265920663e-05,
+      "loss": 1.1532,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987528034109257e-05,
+      "loss": 1.1246,
+      "step": 1076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198745961526087e-05,
+      "loss": 1.2403,
+      "step": 1077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198739100937678e-05,
+      "loss": 1.2231,
+      "step": 1078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987322216458276e-05,
+      "loss": 1.1482,
+      "step": 1079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987253236506642e-05,
+      "loss": 1.1712,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198718406952317e-05,
+      "loss": 1.2005,
+      "step": 1081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198711471550916e-05,
+      "loss": 1.1562,
+      "step": 1082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1987045174465908e-05,
+      "loss": 1.2046,
+      "step": 1083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986975446394715e-05,
+      "loss": 1.1489,
+      "step": 1084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986905531296888e-05,
+      "loss": 1.0809,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986835429173732e-05,
+      "loss": 1.1862,
+      "step": 1086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198676514002657e-05,
+      "loss": 1.1864,
+      "step": 1087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198669466385671e-05,
+      "loss": 1.2054,
+      "step": 1088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198662400066547e-05,
+      "loss": 1.1656,
+      "step": 1089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986553150454182e-05,
+      "loss": 1.1569,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986482113224166e-05,
+      "loss": 1.2422,
+      "step": 1091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986410888976752e-05,
+      "loss": 1.1595,
+      "step": 1092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198633947771328e-05,
+      "loss": 1.1554,
+      "step": 1093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1986267879435084e-05,
+      "loss": 1.1611,
+      "step": 1094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19861960941435e-05,
+      "loss": 1.1644,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198612412183988e-05,
+      "loss": 1.1737,
+      "step": 1096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198605196252557e-05,
+      "loss": 1.1811,
+      "step": 1097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985979616201916e-05,
+      "loss": 1.1683,
+      "step": 1098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198590708287028e-05,
+      "loss": 1.1932,
+      "step": 1099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985834362532014e-05,
+      "loss": 1.1872,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985761455188488e-05,
+      "loss": 1.1777,
+      "step": 1101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985688360841056e-05,
+      "loss": 1.1755,
+      "step": 1102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985615079491097e-05,
+      "loss": 1.0788,
+      "step": 1103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985541611139977e-05,
+      "loss": 1.2657,
+      "step": 1104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985467955789075e-05,
+      "loss": 1.1737,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198539411343977e-05,
+      "loss": 1.1958,
+      "step": 1106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985320084093447e-05,
+      "loss": 1.2469,
+      "step": 1107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985245867751486e-05,
+      "loss": 1.183,
+      "step": 1108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985171464415278e-05,
+      "loss": 1.1877,
+      "step": 1109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985096874086227e-05,
+      "loss": 1.1674,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1985022096765715e-05,
+      "loss": 1.1054,
+      "step": 1111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984947132455152e-05,
+      "loss": 1.2279,
+      "step": 1112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198487198115594e-05,
+      "loss": 1.1771,
+      "step": 1113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198479664286948e-05,
+      "loss": 1.1551,
+      "step": 1114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984721117597196e-05,
+      "loss": 1.1832,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198464540534049e-05,
+      "loss": 1.1597,
+      "step": 1116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984569506100785e-05,
+      "loss": 1.2167,
+      "step": 1117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984493419879504e-05,
+      "loss": 1.1276,
+      "step": 1118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984417146678065e-05,
+      "loss": 1.153,
+      "step": 1119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984340686497905e-05,
+      "loss": 1.1893,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984264039340456e-05,
+      "loss": 1.1653,
+      "step": 1121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198418720520714e-05,
+      "loss": 1.1702,
+      "step": 1122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984110184099412e-05,
+      "loss": 1.2332,
+      "step": 1123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1984032976018707e-05,
+      "loss": 1.1047,
+      "step": 1124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198395558096647e-05,
+      "loss": 1.1373,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198387799894415e-05,
+      "loss": 1.178,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983800229953207e-05,
+      "loss": 1.2157,
+      "step": 1127
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983722273995085e-05,
+      "loss": 1.2078,
+      "step": 1128
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198364413107126e-05,
+      "loss": 1.192,
+      "step": 1129
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983565801183177e-05,
+      "loss": 1.2303,
+      "step": 1130
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198348728433232e-05,
+      "loss": 1.1556,
+      "step": 1131
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983408580520148e-05,
+      "loss": 1.1841,
+      "step": 1132
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198332968974814e-05,
+      "loss": 1.1644,
+      "step": 1133
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198325061201777e-05,
+      "loss": 1.1588,
+      "step": 1134
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1983171347330523e-05,
+      "loss": 1.1564,
+      "step": 1135
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198309189568788e-05,
+      "loss": 1.2139,
+      "step": 1136
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198301225709133e-05,
+      "loss": 1.1478,
+      "step": 1137
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198293243154237e-05,
+      "loss": 1.1989,
+      "step": 1138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198285241904248e-05,
+      "loss": 1.2333,
+      "step": 1139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982772219593172e-05,
+      "loss": 1.2058,
+      "step": 1140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982691833195947e-05,
+      "loss": 1.2174,
+      "step": 1141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982611259852305e-05,
+      "loss": 1.2536,
+      "step": 1142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982530499563753e-05,
+      "loss": 1.166,
+      "step": 1143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982449552331813e-05,
+      "loss": 1.2117,
+      "step": 1144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982368418157995e-05,
+      "loss": 1.1491,
+      "step": 1145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982287097043815e-05,
+      "loss": 1.1538,
+      "step": 1146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1982205588990802e-05,
+      "loss": 1.2035,
+      "step": 1147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198212389400048e-05,
+      "loss": 1.1294,
+      "step": 1148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198204201207438e-05,
+      "loss": 1.1994,
+      "step": 1149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198195994321403e-05,
+      "loss": 1.1969,
+      "step": 1150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981877687420977e-05,
+      "loss": 1.2282,
+      "step": 1151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981795244696756e-05,
+      "loss": 1.1891,
+      "step": 1152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981712615042908e-05,
+      "loss": 1.1903,
+      "step": 1153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198162979846098e-05,
+      "loss": 1.1441,
+      "step": 1154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981546794952532e-05,
+      "loss": 1.2054,
+      "step": 1155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198146360451911e-05,
+      "loss": 1.1977,
+      "step": 1156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198138022716227e-05,
+      "loss": 1.1893,
+      "step": 1157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981296662883585e-05,
+      "loss": 1.1384,
+      "step": 1158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198121291168461e-05,
+      "loss": 1.1077,
+      "step": 1159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981128973566912e-05,
+      "loss": 1.1428,
+      "step": 1160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1981044848532072e-05,
+      "loss": 1.1792,
+      "step": 1161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980960536581656e-05,
+      "loss": 1.1937,
+      "step": 1162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.198087603771725e-05,
+      "loss": 1.1011,
+      "step": 1163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980791351940435e-05,
+      "loss": 1.1396,
+      "step": 1164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980706479252794e-05,
+      "loss": 1.1835,
+      "step": 1165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980621419655916e-05,
+      "loss": 1.2147,
+      "step": 1166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19805361731514e-05,
+      "loss": 1.1129,
+      "step": 1167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980450739740833e-05,
+      "loss": 1.2395,
+      "step": 1168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980365119425826e-05,
+      "loss": 1.2573,
+      "step": 1169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980279312207972e-05,
+      "loss": 1.162,
+      "step": 1170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980193318088886e-05,
+      "loss": 1.1139,
+      "step": 1171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980107137070174e-05,
+      "loss": 1.2639,
+      "step": 1172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1980020769153453e-05,
+      "loss": 1.1649,
+      "step": 1173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979934214340335e-05,
+      "loss": 1.158,
+      "step": 1174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979847472632443e-05,
+      "loss": 1.1687,
+      "step": 1175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979760544031406e-05,
+      "loss": 1.2827,
+      "step": 1176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197967342853885e-05,
+      "loss": 1.1575,
+      "step": 1177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19795861261564e-05,
+      "loss": 1.199,
+      "step": 1178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979498636885696e-05,
+      "loss": 1.1272,
+      "step": 1179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197941096072838e-05,
+      "loss": 1.1889,
+      "step": 1180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979323097686088e-05,
+      "loss": 1.2393,
+      "step": 1181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979235047760468e-05,
+      "loss": 1.1957,
+      "step": 1182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197914681095317e-05,
+      "loss": 1.2589,
+      "step": 1183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1979058387265845e-05,
+      "loss": 1.1887,
+      "step": 1184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197896977670015e-05,
+      "loss": 1.2295,
+      "step": 1185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197888097925774e-05,
+      "loss": 1.219,
+      "step": 1186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197879199494028e-05,
+      "loss": 1.225,
+      "step": 1187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978702823749444e-05,
+      "loss": 1.1769,
+      "step": 1188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197861346568689e-05,
+      "loss": 1.1318,
+      "step": 1189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19785239207543e-05,
+      "loss": 1.1637,
+      "step": 1190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978434188953348e-05,
+      "loss": 1.1447,
+      "step": 1191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978344270285713e-05,
+      "loss": 1.1743,
+      "step": 1192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197825416475308e-05,
+      "loss": 1.2502,
+      "step": 1193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978163872357136e-05,
+      "loss": 1.2552,
+      "step": 1194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1978073393099573e-05,
+      "loss": 1.2377,
+      "step": 1195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977982726982086e-05,
+      "loss": 1.172,
+      "step": 1196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197789187400637e-05,
+      "loss": 1.1547,
+      "step": 1197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977800834174128e-05,
+      "loss": 1.1531,
+      "step": 1198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197770960748707e-05,
+      "loss": 1.1901,
+      "step": 1199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977618193946896e-05,
+      "loss": 1.2037,
+      "step": 1200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197752659355532e-05,
+      "loss": 1.1408,
+      "step": 1201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977434806314065e-05,
+      "loss": 1.1948,
+      "step": 1202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977342832224838e-05,
+      "loss": 1.167,
+      "step": 1203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197725067128937e-05,
+      "loss": 1.1748,
+      "step": 1204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977158323509383e-05,
+      "loss": 1.1752,
+      "step": 1205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1977065788886603e-05,
+      "loss": 1.1659,
+      "step": 1206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197697306742277e-05,
+      "loss": 1.2346,
+      "step": 1207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976880159119623e-05,
+      "loss": 1.1976,
+      "step": 1208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976787063978893e-05,
+      "loss": 1.1896,
+      "step": 1209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976693782002328e-05,
+      "loss": 1.1386,
+      "step": 1210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976600313191675e-05,
+      "loss": 1.2234,
+      "step": 1211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197650665754868e-05,
+      "loss": 1.1803,
+      "step": 1212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976412815075104e-05,
+      "loss": 1.1034,
+      "step": 1213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19763187857727e-05,
+      "loss": 1.233,
+      "step": 1214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976224569643227e-05,
+      "loss": 1.1872,
+      "step": 1215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197613016668845e-05,
+      "loss": 1.1671,
+      "step": 1216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1976035576910142e-05,
+      "loss": 1.2576,
+      "step": 1217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197594080031007e-05,
+      "loss": 1.2393,
+      "step": 1218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975845836890016e-05,
+      "loss": 1.1924,
+      "step": 1219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197575068665175e-05,
+      "loss": 1.1988,
+      "step": 1220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975655349597054e-05,
+      "loss": 1.2457,
+      "step": 1221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975559825727715e-05,
+      "loss": 1.1238,
+      "step": 1222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975464115045524e-05,
+      "loss": 1.1791,
+      "step": 1223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975368217552273e-05,
+      "loss": 1.275,
+      "step": 1224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975272133249754e-05,
+      "loss": 1.2147,
+      "step": 1225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975175862139774e-05,
+      "loss": 1.2013,
+      "step": 1226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1975079404224128e-05,
+      "loss": 1.2037,
+      "step": 1227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974982759504624e-05,
+      "loss": 1.1537,
+      "step": 1228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974885927983077e-05,
+      "loss": 1.2638,
+      "step": 1229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974788909661295e-05,
+      "loss": 1.2126,
+      "step": 1230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974691704541098e-05,
+      "loss": 1.2074,
+      "step": 1231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19745943126243e-05,
+      "loss": 1.2588,
+      "step": 1232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974496733912737e-05,
+      "loss": 1.2153,
+      "step": 1233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974398968408224e-05,
+      "loss": 1.1268,
+      "step": 1234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974301016112603e-05,
+      "loss": 1.1436,
+      "step": 1235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974202877027698e-05,
+      "loss": 1.1698,
+      "step": 1236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197410455115535e-05,
+      "loss": 1.108,
+      "step": 1237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1974006038497407e-05,
+      "loss": 1.2085,
+      "step": 1238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19739073390557e-05,
+      "loss": 1.1921,
+      "step": 1239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973808452832092e-05,
+      "loss": 1.1986,
+      "step": 1240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197370937982843e-05,
+      "loss": 1.1633,
+      "step": 1241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973610120046564e-05,
+      "loss": 1.1936,
+      "step": 1242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197351067348836e-05,
+      "loss": 1.1956,
+      "step": 1243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973411040155676e-05,
+      "loss": 1.1216,
+      "step": 1244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197331122005038e-05,
+      "loss": 1.246,
+      "step": 1245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197321121317434e-05,
+      "loss": 1.232,
+      "step": 1246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1973111019529427e-05,
+      "loss": 1.2818,
+      "step": 1247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197301063911752e-05,
+      "loss": 1.2635,
+      "step": 1248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972910071940502e-05,
+      "loss": 1.2607,
+      "step": 1249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197280931800025e-05,
+      "loss": 1.1837,
+      "step": 1250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972708377298655e-05,
+      "loss": 1.2122,
+      "step": 1251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972607249837602e-05,
+      "loss": 1.2594,
+      "step": 1252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197250593561899e-05,
+      "loss": 1.2052,
+      "step": 1253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197240443464472e-05,
+      "loss": 1.2195,
+      "step": 1254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972302746916684e-05,
+      "loss": 1.1727,
+      "step": 1255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972200872436793e-05,
+      "loss": 1.1951,
+      "step": 1256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1972098811206947e-05,
+      "loss": 1.2083,
+      "step": 1257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971996563229068e-05,
+      "loss": 1.1557,
+      "step": 1258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197189412850506e-05,
+      "loss": 1.2075,
+      "step": 1259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197179150703685e-05,
+      "loss": 1.218,
+      "step": 1260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971688698826355e-05,
+      "loss": 1.137,
+      "step": 1261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971585703875503e-05,
+      "loss": 1.2234,
+      "step": 1262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971482522186215e-05,
+      "loss": 1.2938,
+      "step": 1263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971379153760438e-05,
+      "loss": 1.162,
+      "step": 1264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197127559860009e-05,
+      "loss": 1.278,
+      "step": 1265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1971171856707125e-05,
+      "loss": 1.2052,
+      "step": 1266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197106792808348e-05,
+      "loss": 1.1764,
+      "step": 1267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970963812731104e-05,
+      "loss": 1.1861,
+      "step": 1268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970859510651943e-05,
+      "loss": 1.1781,
+      "step": 1269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197075502184795e-05,
+      "loss": 1.2445,
+      "step": 1270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970650346321085e-05,
+      "loss": 1.2008,
+      "step": 1271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970545484073308e-05,
+      "loss": 1.248,
+      "step": 1272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197044043510658e-05,
+      "loss": 1.1651,
+      "step": 1273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197033519942287e-05,
+      "loss": 1.089,
+      "step": 1274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1970229777024147e-05,
+      "loss": 1.2164,
+      "step": 1275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197012416791239e-05,
+      "loss": 1.2198,
+      "step": 1276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.197001837208957e-05,
+      "loss": 1.1011,
+      "step": 1277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969912389557673e-05,
+      "loss": 1.1833,
+      "step": 1278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196980622031868e-05,
+      "loss": 1.1336,
+      "step": 1279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969699864374586e-05,
+      "loss": 1.2356,
+      "step": 1280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196959332172738e-05,
+      "loss": 1.1385,
+      "step": 1281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969486592379053e-05,
+      "loss": 1.2347,
+      "step": 1282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969379676331606e-05,
+      "loss": 1.1184,
+      "step": 1283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969272573587042e-05,
+      "loss": 1.1661,
+      "step": 1284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969165284147364e-05,
+      "loss": 1.1507,
+      "step": 1285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1969057808014586e-05,
+      "loss": 1.1605,
+      "step": 1286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196895014519072e-05,
+      "loss": 1.2259,
+      "step": 1287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968842295677776e-05,
+      "loss": 1.1629,
+      "step": 1288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196873425947778e-05,
+      "loss": 1.1877,
+      "step": 1289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968626036592758e-05,
+      "loss": 1.2069,
+      "step": 1290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968517627024726e-05,
+      "loss": 1.1227,
+      "step": 1291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196840903077572e-05,
+      "loss": 1.1876,
+      "step": 1292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196830024784778e-05,
+      "loss": 1.1736,
+      "step": 1293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1968191278242937e-05,
+      "loss": 1.1626,
+      "step": 1294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196808212196323e-05,
+      "loss": 1.1246,
+      "step": 1295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.19679727790107e-05,
+      "loss": 1.1778,
+      "step": 1296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967863249387403e-05,
+      "loss": 1.1584,
+      "step": 1297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967753533095386e-05,
+      "loss": 1.1575,
+      "step": 1298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967643630136707e-05,
+      "loss": 1.2301,
+      "step": 1299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.1967533540513424e-05,
+      "loss": 1.1659,
+      "step": 1300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.196742326422759e-05,
+      "loss": 1.2078,
+      "step": 1301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196731280128128e-05,
+      "loss": 1.2,
+      "step": 1302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1967202151676558e-05,
+      "loss": 1.1322,
+      "step": 1303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19670913154155e-05,
+      "loss": 1.1184,
+      "step": 1304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966980292500175e-05,
+      "loss": 1.206,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966869082932667e-05,
+      "loss": 1.2611,
+      "step": 1306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196675768671506e-05,
+      "loss": 1.2787,
+      "step": 1307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966646103849438e-05,
+      "loss": 1.2535,
+      "step": 1308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196653433433789e-05,
+      "loss": 1.2277,
+      "step": 1309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196642237818251e-05,
+      "loss": 1.1736,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966310235385388e-05,
+      "loss": 1.2016,
+      "step": 1311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1966197905948634e-05,
+      "loss": 1.2015,
+      "step": 1312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196608538987435e-05,
+      "loss": 1.165,
+      "step": 1313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196597268716464e-05,
+      "loss": 1.1402,
+      "step": 1314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965859797821614e-05,
+      "loss": 1.1401,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965746721847388e-05,
+      "loss": 1.1641,
+      "step": 1316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196563345924408e-05,
+      "loss": 1.216,
+      "step": 1317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965520010013813e-05,
+      "loss": 1.1216,
+      "step": 1318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965406374158706e-05,
+      "loss": 1.1447,
+      "step": 1319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196529255168089e-05,
+      "loss": 1.1922,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1965178542582497e-05,
+      "loss": 1.2044,
+      "step": 1321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196506434686566e-05,
+      "loss": 1.2585,
+      "step": 1322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196494996453252e-05,
+      "loss": 1.1433,
+      "step": 1323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964835395585216e-05,
+      "loss": 1.2325,
+      "step": 1324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19647206400259e-05,
+      "loss": 1.2057,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964605697856713e-05,
+      "loss": 1.1778,
+      "step": 1326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964490569079813e-05,
+      "loss": 1.2534,
+      "step": 1327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964375253697353e-05,
+      "loss": 1.2051,
+      "step": 1328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964259751711495e-05,
+      "loss": 1.2072,
+      "step": 1329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19641440631244e-05,
+      "loss": 1.0901,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1964028187938236e-05,
+      "loss": 1.1531,
+      "step": 1331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963912126155174e-05,
+      "loss": 1.2235,
+      "step": 1332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196379587777738e-05,
+      "loss": 1.2222,
+      "step": 1333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963679442807043e-05,
+      "loss": 1.098,
+      "step": 1334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963562821246334e-05,
+      "loss": 1.1693,
+      "step": 1335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963446013097443e-05,
+      "loss": 1.1263,
+      "step": 1336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196332901836255e-05,
+      "loss": 1.2332,
+      "step": 1337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963211837043854e-05,
+      "loss": 1.1949,
+      "step": 1338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1963094469143547e-05,
+      "loss": 1.2144,
+      "step": 1339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962976914663823e-05,
+      "loss": 1.0597,
+      "step": 1340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962859173606888e-05,
+      "loss": 1.1411,
+      "step": 1341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962741245974944e-05,
+      "loss": 1.1891,
+      "step": 1342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962623131770203e-05,
+      "loss": 1.136,
+      "step": 1343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962504830994876e-05,
+      "loss": 1.1075,
+      "step": 1344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962386343651177e-05,
+      "loss": 1.1364,
+      "step": 1345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1962267669741323e-05,
+      "loss": 1.1828,
+      "step": 1346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196214880926754e-05,
+      "loss": 1.1238,
+      "step": 1347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196202976223205e-05,
+      "loss": 1.1451,
+      "step": 1348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196191052863709e-05,
+      "loss": 1.122,
+      "step": 1349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961791108484882e-05,
+      "loss": 1.2076,
+      "step": 1350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196167150177767e-05,
+      "loss": 1.1866,
+      "step": 1351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961551708517693e-05,
+      "loss": 1.1612,
+      "step": 1352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961431728707192e-05,
+      "loss": 1.1275,
+      "step": 1353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196131156234842e-05,
+      "loss": 1.1247,
+      "step": 1354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1961191209443615e-05,
+      "loss": 1.1685,
+      "step": 1355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196107066999505e-05,
+      "loss": 1.1308,
+      "step": 1356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960949944004958e-05,
+      "loss": 1.1794,
+      "step": 1357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196082903147562e-05,
+      "loss": 1.1071,
+      "step": 1358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196070793240929e-05,
+      "loss": 1.198,
+      "step": 1359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960586646808243e-05,
+      "loss": 1.2115,
+      "step": 1360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960465174674743e-05,
+      "loss": 1.2263,
+      "step": 1361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960343516011067e-05,
+      "loss": 1.2192,
+      "step": 1362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1960221670819497e-05,
+      "loss": 1.1593,
+      "step": 1363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.196009963910231e-05,
+      "loss": 1.1338,
+      "step": 1364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959977420861795e-05,
+      "loss": 1.2047,
+      "step": 1365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195985501610024e-05,
+      "loss": 1.1838,
+      "step": 1366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959732424819935e-05,
+      "loss": 1.1626,
+      "step": 1367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959609647023176e-05,
+      "loss": 1.1545,
+      "step": 1368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959486682712266e-05,
+      "loss": 1.1925,
+      "step": 1369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19593635318895e-05,
+      "loss": 1.1952,
+      "step": 1370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959240194557197e-05,
+      "loss": 1.2065,
+      "step": 1371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1959116670717657e-05,
+      "loss": 1.1316,
+      "step": 1372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958992960373195e-05,
+      "loss": 1.1291,
+      "step": 1373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195886906352613e-05,
+      "loss": 1.1833,
+      "step": 1374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958744980178775e-05,
+      "loss": 1.1591,
+      "step": 1375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958620710333463e-05,
+      "loss": 1.2072,
+      "step": 1376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958496253992518e-05,
+      "loss": 1.2345,
+      "step": 1377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195837161115827e-05,
+      "loss": 1.2603,
+      "step": 1378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958246781833052e-05,
+      "loss": 1.141,
+      "step": 1379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1958121766019203e-05,
+      "loss": 1.1796,
+      "step": 1380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957996563719062e-05,
+      "loss": 1.1151,
+      "step": 1381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195787117493498e-05,
+      "loss": 1.1797,
+      "step": 1382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19577455996693e-05,
+      "loss": 1.2355,
+      "step": 1383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957619837924375e-05,
+      "loss": 1.1994,
+      "step": 1384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195749388970256e-05,
+      "loss": 1.2626,
+      "step": 1385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195736775500621e-05,
+      "loss": 1.1383,
+      "step": 1386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195724143383769e-05,
+      "loss": 1.2383,
+      "step": 1387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1957114926199372e-05,
+      "loss": 1.1229,
+      "step": 1388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956988232093618e-05,
+      "loss": 1.2417,
+      "step": 1389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19568613515228e-05,
+      "loss": 1.1367,
+      "step": 1390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956734284489296e-05,
+      "loss": 1.1881,
+      "step": 1391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956607030995485e-05,
+      "loss": 1.2604,
+      "step": 1392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956479591043753e-05,
+      "loss": 1.198,
+      "step": 1393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195635196463648e-05,
+      "loss": 1.1449,
+      "step": 1394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956224151776063e-05,
+      "loss": 1.208,
+      "step": 1395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1956096152464894e-05,
+      "loss": 1.0989,
+      "step": 1396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955967966705368e-05,
+      "loss": 1.1976,
+      "step": 1397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955839594499887e-05,
+      "loss": 1.1365,
+      "step": 1398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955711035850853e-05,
+      "loss": 1.2364,
+      "step": 1399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955582290760676e-05,
+      "loss": 1.2036,
+      "step": 1400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955453359231767e-05,
+      "loss": 1.2148,
+      "step": 1401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195532424126654e-05,
+      "loss": 1.1248,
+      "step": 1402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955194936867408e-05,
+      "loss": 1.1976,
+      "step": 1403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1955065446036803e-05,
+      "loss": 1.1821,
+      "step": 1404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195493576877714e-05,
+      "loss": 1.1236,
+      "step": 1405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954805905090854e-05,
+      "loss": 1.2448,
+      "step": 1406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954675854980373e-05,
+      "loss": 1.2147,
+      "step": 1407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954545618448134e-05,
+      "loss": 1.1676,
+      "step": 1408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954415195496575e-05,
+      "loss": 1.187,
+      "step": 1409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195428458612814e-05,
+      "loss": 1.1716,
+      "step": 1410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1954153790345276e-05,
+      "loss": 1.1468,
+      "step": 1411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195402280815043e-05,
+      "loss": 1.137,
+      "step": 1412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953891639546056e-05,
+      "loss": 1.1922,
+      "step": 1413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953760284534607e-05,
+      "loss": 1.2402,
+      "step": 1414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953628743118548e-05,
+      "loss": 1.1919,
+      "step": 1415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195349701530034e-05,
+      "loss": 1.1319,
+      "step": 1416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953365101082452e-05,
+      "loss": 1.2229,
+      "step": 1417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1953233000467347e-05,
+      "loss": 1.2068,
+      "step": 1418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195310071345751e-05,
+      "loss": 1.1265,
+      "step": 1419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952968240055413e-05,
+      "loss": 1.1553,
+      "step": 1420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952835580263532e-05,
+      "loss": 1.203,
+      "step": 1421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195270273408436e-05,
+      "loss": 1.1911,
+      "step": 1422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952569701520377e-05,
+      "loss": 1.2116,
+      "step": 1423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195243648257408e-05,
+      "loss": 1.1444,
+      "step": 1424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195230307724796e-05,
+      "loss": 1.1618,
+      "step": 1425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195216948554452e-05,
+      "loss": 1.2275,
+      "step": 1426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1952035707466256e-05,
+      "loss": 1.1743,
+      "step": 1427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951901743015677e-05,
+      "loss": 1.1531,
+      "step": 1428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195176759219529e-05,
+      "loss": 1.178,
+      "step": 1429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951633255007612e-05,
+      "loss": 1.147,
+      "step": 1430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951498731455152e-05,
+      "loss": 1.1333,
+      "step": 1431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951364021540434e-05,
+      "loss": 1.1128,
+      "step": 1432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951229125265975e-05,
+      "loss": 1.182,
+      "step": 1433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1951094042634307e-05,
+      "loss": 1.2197,
+      "step": 1434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195095877364796e-05,
+      "loss": 1.1529,
+      "step": 1435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950823318309462e-05,
+      "loss": 1.137,
+      "step": 1436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950687676621354e-05,
+      "loss": 1.1798,
+      "step": 1437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950551848586174e-05,
+      "loss": 1.1878,
+      "step": 1438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950415834206466e-05,
+      "loss": 1.1598,
+      "step": 1439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195027963348478e-05,
+      "loss": 1.2303,
+      "step": 1440
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.195014324642366e-05,
+      "loss": 1.0965,
+      "step": 1441
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1950006673025666e-05,
+      "loss": 1.186,
+      "step": 1442
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949869913293356e-05,
+      "loss": 1.2005,
+      "step": 1443
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949732967229286e-05,
+      "loss": 1.1708,
+      "step": 1444
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194959583483603e-05,
+      "loss": 1.1654,
+      "step": 1445
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194945851611614e-05,
+      "loss": 1.2669,
+      "step": 1446
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949321011072205e-05,
+      "loss": 1.189,
+      "step": 1447
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194918331970679e-05,
+      "loss": 1.152,
+      "step": 1448
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1949045442022476e-05,
+      "loss": 1.1946,
+      "step": 1449
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948907378021844e-05,
+      "loss": 1.2106,
+      "step": 1450
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948769127707484e-05,
+      "loss": 1.2603,
+      "step": 1451
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948630691081978e-05,
+      "loss": 1.1245,
+      "step": 1452
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194849206814792e-05,
+      "loss": 1.2026,
+      "step": 1453
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948353258907912e-05,
+      "loss": 1.19,
+      "step": 1454
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194821426336455e-05,
+      "loss": 1.1963,
+      "step": 1455
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1948075081520432e-05,
+      "loss": 1.2295,
+      "step": 1456
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947935713378175e-05,
+      "loss": 1.194,
+      "step": 1457
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194779615894038e-05,
+      "loss": 1.182,
+      "step": 1458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947656418209663e-05,
+      "loss": 1.1774,
+      "step": 1459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194751649118864e-05,
+      "loss": 1.2277,
+      "step": 1460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947376377879935e-05,
+      "loss": 1.2588,
+      "step": 1461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947236078286167e-05,
+      "loss": 1.1077,
+      "step": 1462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1947095592409966e-05,
+      "loss": 1.1319,
+      "step": 1463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946954920253968e-05,
+      "loss": 1.1582,
+      "step": 1464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946814061820798e-05,
+      "loss": 1.1554,
+      "step": 1465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19466730171131e-05,
+      "loss": 1.2524,
+      "step": 1466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946531786133513e-05,
+      "loss": 1.1926,
+      "step": 1467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194639036888468e-05,
+      "loss": 1.2468,
+      "step": 1468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946248765369255e-05,
+      "loss": 1.1626,
+      "step": 1469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1946106975589885e-05,
+      "loss": 1.2552,
+      "step": 1470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194596499954923e-05,
+      "loss": 1.0931,
+      "step": 1471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194582283724994e-05,
+      "loss": 1.2012,
+      "step": 1472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945680488694687e-05,
+      "loss": 1.1748,
+      "step": 1473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194553795388613e-05,
+      "loss": 1.1373,
+      "step": 1474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945395232826943e-05,
+      "loss": 1.1361,
+      "step": 1475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945252325519797e-05,
+      "loss": 1.187,
+      "step": 1476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1945109231967368e-05,
+      "loss": 1.2316,
+      "step": 1477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944965952172333e-05,
+      "loss": 1.1899,
+      "step": 1478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944822486137376e-05,
+      "loss": 1.1756,
+      "step": 1479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194467883386519e-05,
+      "loss": 1.1649,
+      "step": 1480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944534995358458e-05,
+      "loss": 1.1916,
+      "step": 1481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944390970619878e-05,
+      "loss": 1.1173,
+      "step": 1482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944246759652146e-05,
+      "loss": 1.1718,
+      "step": 1483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1944102362457962e-05,
+      "loss": 1.1966,
+      "step": 1484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194395777904003e-05,
+      "loss": 1.1688,
+      "step": 1485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943813009401056e-05,
+      "loss": 1.1749,
+      "step": 1486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943668053543755e-05,
+      "loss": 1.1177,
+      "step": 1487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194352291147084e-05,
+      "loss": 1.0877,
+      "step": 1488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943377583185028e-05,
+      "loss": 1.2204,
+      "step": 1489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194323206868904e-05,
+      "loss": 1.1899,
+      "step": 1490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1943086367985605e-05,
+      "loss": 1.157,
+      "step": 1491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942940481077447e-05,
+      "loss": 1.1277,
+      "step": 1492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19427944079673e-05,
+      "loss": 1.1383,
+      "step": 1493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19426481486579e-05,
+      "loss": 1.13,
+      "step": 1494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942501703151985e-05,
+      "loss": 1.197,
+      "step": 1495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19423550714523e-05,
+      "loss": 1.1731,
+      "step": 1496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942208253561586e-05,
+      "loss": 1.1387,
+      "step": 1497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1942061249482595e-05,
+      "loss": 1.1096,
+      "step": 1498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194191405921808e-05,
+      "loss": 1.152,
+      "step": 1499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19417666827708e-05,
+      "loss": 1.2324,
+      "step": 1500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941619120143515e-05,
+      "loss": 1.2119,
+      "step": 1501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194147137133898e-05,
+      "loss": 1.1892,
+      "step": 1502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941323436359972e-05,
+      "loss": 1.1827,
+      "step": 1503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941175315209258e-05,
+      "loss": 1.158,
+      "step": 1504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1941027007889613e-05,
+      "loss": 1.1691,
+      "step": 1505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1940878514403805e-05,
+      "loss": 1.2485,
+      "step": 1506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194072983475463e-05,
+      "loss": 1.1375,
+      "step": 1507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194058096894486e-05,
+      "loss": 1.1928,
+      "step": 1508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194043191697729e-05,
+      "loss": 1.1961,
+      "step": 1509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194028267885471e-05,
+      "loss": 1.202,
+      "step": 1510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.194013325457991e-05,
+      "loss": 1.1866,
+      "step": 1511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939983644155695e-05,
+      "loss": 1.2191,
+      "step": 1512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939833847584864e-05,
+      "loss": 1.1658,
+      "step": 1513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193968386487022e-05,
+      "loss": 1.2073,
+      "step": 1514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939533696014575e-05,
+      "loss": 1.1824,
+      "step": 1515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939383341020742e-05,
+      "loss": 1.1863,
+      "step": 1516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193923279989153e-05,
+      "loss": 1.2097,
+      "step": 1517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1939082072629767e-05,
+      "loss": 1.2399,
+      "step": 1518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938931159238266e-05,
+      "loss": 1.2291,
+      "step": 1519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938780059719864e-05,
+      "loss": 1.2207,
+      "step": 1520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938628774077383e-05,
+      "loss": 1.2516,
+      "step": 1521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193847730231366e-05,
+      "loss": 1.2248,
+      "step": 1522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193832564443153e-05,
+      "loss": 1.2048,
+      "step": 1523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1938173800433832e-05,
+      "loss": 1.1523,
+      "step": 1524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193802177032341e-05,
+      "loss": 1.1099,
+      "step": 1525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937869554103117e-05,
+      "loss": 1.1915,
+      "step": 1526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937717151775793e-05,
+      "loss": 1.1464,
+      "step": 1527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.19375645633443e-05,
+      "loss": 1.1932,
+      "step": 1528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193741178881149e-05,
+      "loss": 1.1495,
+      "step": 1529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937258828180233e-05,
+      "loss": 1.2202,
+      "step": 1530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1937105681453382e-05,
+      "loss": 1.1378,
+      "step": 1531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936952348633813e-05,
+      "loss": 1.1548,
+      "step": 1532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936798829724396e-05,
+      "loss": 1.2149,
+      "step": 1533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936645124728003e-05,
+      "loss": 1.1879,
+      "step": 1534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.1936491233647514e-05,
+      "loss": 1.2809,
+      "step": 1535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193633715648581e-05,
+      "loss": 1.1505,
+      "step": 1536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193618289324578e-05,
+      "loss": 1.1588,
+      "step": 1537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.193602844393031e-05,
+      "loss": 1.2327,
+      "step": 1538
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935873808542297e-05,
+      "loss": 1.2236,
+      "step": 1539
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935718987084627e-05,
+      "loss": 1.1922,
+      "step": 1540
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935563979560206e-05,
+      "loss": 1.2184,
+      "step": 1541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193540878597194e-05,
+      "loss": 1.1893,
+      "step": 1542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193525340632273e-05,
+      "loss": 1.1939,
+      "step": 1543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1935097840615485e-05,
+      "loss": 1.2075,
+      "step": 1544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193494208885312e-05,
+      "loss": 1.1794,
+      "step": 1545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934786151038555e-05,
+      "loss": 1.2007,
+      "step": 1546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934630027174708e-05,
+      "loss": 1.1168,
+      "step": 1547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934473717264502e-05,
+      "loss": 1.1856,
+      "step": 1548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934317221310862e-05,
+      "loss": 1.2309,
+      "step": 1549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193416053931672e-05,
+      "loss": 1.1691,
+      "step": 1550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1934003671285014e-05,
+      "loss": 1.1665,
+      "step": 1551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193384661721868e-05,
+      "loss": 1.1124,
+      "step": 1552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933689377120657e-05,
+      "loss": 1.2246,
+      "step": 1553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193353195099389e-05,
+      "loss": 1.1758,
+      "step": 1554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193337433884133e-05,
+      "loss": 1.1962,
+      "step": 1555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1933216540665924e-05,
+      "loss": 1.1656,
+      "step": 1556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193305855647063e-05,
+      "loss": 1.1288,
+      "step": 1557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193290038625841e-05,
+      "loss": 1.1655,
+      "step": 1558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932742030032216e-05,
+      "loss": 1.2341,
+      "step": 1559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932583487795023e-05,
+      "loss": 1.1849,
+      "step": 1560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932424759549798e-05,
+      "loss": 1.2083,
+      "step": 1561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1932265845299512e-05,
+      "loss": 1.1948,
+      "step": 1562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193210674504714e-05,
+      "loss": 1.2228,
+      "step": 1563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931947458795668e-05,
+      "loss": 1.2255,
+      "step": 1564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193178798654807e-05,
+      "loss": 1.1405,
+      "step": 1565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931628328307333e-05,
+      "loss": 1.2032,
+      "step": 1566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931468484076454e-05,
+      "loss": 1.227,
+      "step": 1567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931308453858423e-05,
+      "loss": 1.19,
+      "step": 1568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1931148237656234e-05,
+      "loss": 1.1841,
+      "step": 1569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930987835472894e-05,
+      "loss": 1.1971,
+      "step": 1570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19308272473114e-05,
+      "loss": 1.1722,
+      "step": 1571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193066647317476e-05,
+      "loss": 1.1847,
+      "step": 1572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193050551306599e-05,
+      "loss": 1.1121,
+      "step": 1573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19303443669881e-05,
+      "loss": 1.147,
+      "step": 1574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.193018303494411e-05,
+      "loss": 1.1743,
+      "step": 1575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1930021516937038e-05,
+      "loss": 1.1612,
+      "step": 1576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929859812969916e-05,
+      "loss": 1.1781,
+      "step": 1577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192969792304576e-05,
+      "loss": 1.1378,
+      "step": 1578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929535847167615e-05,
+      "loss": 1.1467,
+      "step": 1579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192937358533851e-05,
+      "loss": 1.1978,
+      "step": 1580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192921113756148e-05,
+      "loss": 1.1698,
+      "step": 1581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1929048503839576e-05,
+      "loss": 1.1428,
+      "step": 1582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928885684175835e-05,
+      "loss": 1.169,
+      "step": 1583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192872267857331e-05,
+      "loss": 1.2232,
+      "step": 1584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928559487035054e-05,
+      "loss": 1.1902,
+      "step": 1585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928396109564126e-05,
+      "loss": 1.2033,
+      "step": 1586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928232546163577e-05,
+      "loss": 1.1854,
+      "step": 1587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1928068796836477e-05,
+      "loss": 1.252,
+      "step": 1588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192790486158589e-05,
+      "loss": 1.2124,
+      "step": 1589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192774074041489e-05,
+      "loss": 1.1304,
+      "step": 1590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927576433326546e-05,
+      "loss": 1.0819,
+      "step": 1591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927411940323933e-05,
+      "loss": 1.189,
+      "step": 1592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927247261410137e-05,
+      "loss": 1.0968,
+      "step": 1593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1927082396588242e-05,
+      "loss": 1.1203,
+      "step": 1594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192691734586133e-05,
+      "loss": 1.1198,
+      "step": 1595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926752109232495e-05,
+      "loss": 1.1008,
+      "step": 1596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926586686704833e-05,
+      "loss": 1.2054,
+      "step": 1597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192642107828144e-05,
+      "loss": 1.1714,
+      "step": 1598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926255283965418e-05,
+      "loss": 1.185,
+      "step": 1599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1926089303759872e-05,
+      "loss": 1.1733,
+      "step": 1600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192592313766791e-05,
+      "loss": 1.1739,
+      "step": 1601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925756785692643e-05,
+      "loss": 1.1717,
+      "step": 1602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925590247837183e-05,
+      "loss": 1.1528,
+      "step": 1603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925423524104656e-05,
+      "loss": 1.1703,
+      "step": 1604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925256614498183e-05,
+      "loss": 1.2258,
+      "step": 1605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1925089519020886e-05,
+      "loss": 1.178,
+      "step": 1606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924922237675894e-05,
+      "loss": 1.2443,
+      "step": 1607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924754770466344e-05,
+      "loss": 1.1834,
+      "step": 1608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192458711739537e-05,
+      "loss": 1.2072,
+      "step": 1609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924419278466108e-05,
+      "loss": 1.257,
+      "step": 1610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924251253681708e-05,
+      "loss": 1.1391,
+      "step": 1611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1924083043045312e-05,
+      "loss": 1.1286,
+      "step": 1612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192391464656007e-05,
+      "loss": 1.1022,
+      "step": 1613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923746064229137e-05,
+      "loss": 1.1473,
+      "step": 1614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192357729605567e-05,
+      "loss": 1.2406,
+      "step": 1615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923408342042834e-05,
+      "loss": 1.2164,
+      "step": 1616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1923239202193782e-05,
+      "loss": 1.1624,
+      "step": 1617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192306987651169e-05,
+      "loss": 1.1959,
+      "step": 1618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922900364999727e-05,
+      "loss": 1.1458,
+      "step": 1619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922730667661067e-05,
+      "loss": 1.1842,
+      "step": 1620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192256078449889e-05,
+      "loss": 1.1485,
+      "step": 1621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922390715516372e-05,
+      "loss": 1.1406,
+      "step": 1622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1922220460716703e-05,
+      "loss": 1.2576,
+      "step": 1623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192205002010307e-05,
+      "loss": 1.1897,
+      "step": 1624
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921879393678668e-05,
+      "loss": 1.198,
+      "step": 1625
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921708581446684e-05,
+      "loss": 1.2022,
+      "step": 1626
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921537583410325e-05,
+      "loss": 1.1648,
+      "step": 1627
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.192136639957279e-05,
+      "loss": 1.1242,
+      "step": 1628
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921195029937284e-05,
+      "loss": 1.1597,
+      "step": 1629
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1921023474507015e-05,
+      "loss": 1.2055,
+      "step": 1630
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920851733285203e-05,
+      "loss": 1.1789,
+      "step": 1631
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920679806275054e-05,
+      "loss": 1.2101,
+      "step": 1632
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920507693479796e-05,
+      "loss": 1.1868,
+      "step": 1633
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920335394902643e-05,
+      "loss": 1.1816,
+      "step": 1634
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1920162910546835e-05,
+      "loss": 1.2133,
+      "step": 1635
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919990240415594e-05,
+      "loss": 1.1932,
+      "step": 1636
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191981738451215e-05,
+      "loss": 1.094,
+      "step": 1637
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919644342839746e-05,
+      "loss": 1.1162,
+      "step": 1638
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919471115401625e-05,
+      "loss": 1.141,
+      "step": 1639
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1919297702201024e-05,
+      "loss": 1.1263,
+      "step": 1640
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191912410324119e-05,
+      "loss": 1.2152,
+      "step": 1641
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918950318525382e-05,
+      "loss": 1.2214,
+      "step": 1642
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191877634805685e-05,
+      "loss": 1.0669,
+      "step": 1643
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191860219183885e-05,
+      "loss": 1.1491,
+      "step": 1644
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191842784987465e-05,
+      "loss": 1.2326,
+      "step": 1645
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918253322167502e-05,
+      "loss": 1.1384,
+      "step": 1646
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1918078608720688e-05,
+      "loss": 1.2258,
+      "step": 1647
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917903709537474e-05,
+      "loss": 1.1345,
+      "step": 1648
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917728624621135e-05,
+      "loss": 1.2075,
+      "step": 1649
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917553353974953e-05,
+      "loss": 1.2282,
+      "step": 1650
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191737789760221e-05,
+      "loss": 1.1361,
+      "step": 1651
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1917202255506184e-05,
+      "loss": 1.1923,
+      "step": 1652
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191702642769017e-05,
+      "loss": 1.1252,
+      "step": 1653
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916850414157464e-05,
+      "loss": 1.1496,
+      "step": 1654
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191667421491136e-05,
+      "loss": 1.0949,
+      "step": 1655
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916497829955155e-05,
+      "loss": 1.1822,
+      "step": 1656
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916321259292153e-05,
+      "loss": 1.2125,
+      "step": 1657
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1916144502925662e-05,
+      "loss": 1.1124,
+      "step": 1658
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191596756085899e-05,
+      "loss": 1.2074,
+      "step": 1659
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915790433095458e-05,
+      "loss": 1.1675,
+      "step": 1660
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915613119638372e-05,
+      "loss": 1.1937,
+      "step": 1661
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915435620491057e-05,
+      "loss": 1.1978,
+      "step": 1662
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915257935656845e-05,
+      "loss": 1.1355,
+      "step": 1663
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1915080065139048e-05,
+      "loss": 1.2068,
+      "step": 1664
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191490200894101e-05,
+      "loss": 1.1502,
+      "step": 1665
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914723767066063e-05,
+      "loss": 1.1988,
+      "step": 1666
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191454533951754e-05,
+      "loss": 1.2294,
+      "step": 1667
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914366726298784e-05,
+      "loss": 1.1168,
+      "step": 1668
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191418792741314e-05,
+      "loss": 1.1802,
+      "step": 1669
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1914008942863964e-05,
+      "loss": 1.1615,
+      "step": 1670
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913829772654597e-05,
+      "loss": 1.1017,
+      "step": 1671
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19136504167884e-05,
+      "loss": 1.1511,
+      "step": 1672
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191347087526873e-05,
+      "loss": 1.1478,
+      "step": 1673
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913291148098948e-05,
+      "loss": 1.1645,
+      "step": 1674
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1913111235282424e-05,
+      "loss": 1.185,
+      "step": 1675
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912931136822526e-05,
+      "loss": 1.2512,
+      "step": 1676
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191275085272262e-05,
+      "loss": 1.1612,
+      "step": 1677
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191257038298609e-05,
+      "loss": 1.2491,
+      "step": 1678
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912389727616315e-05,
+      "loss": 1.0958,
+      "step": 1679
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191220888661668e-05,
+      "loss": 1.198,
+      "step": 1680
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1912027859990563e-05,
+      "loss": 1.1727,
+      "step": 1681
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191184664774136e-05,
+      "loss": 1.199,
+      "step": 1682
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911665249872466e-05,
+      "loss": 1.1935,
+      "step": 1683
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911483666387276e-05,
+      "loss": 1.1562,
+      "step": 1684
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911301897289188e-05,
+      "loss": 1.2083,
+      "step": 1685
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1911119942581607e-05,
+      "loss": 1.1654,
+      "step": 1686
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910937802267946e-05,
+      "loss": 1.1951,
+      "step": 1687
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191075547635161e-05,
+      "loss": 1.1554,
+      "step": 1688
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910572964836016e-05,
+      "loss": 1.2024,
+      "step": 1689
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.191039026772458e-05,
+      "loss": 1.1354,
+      "step": 1690
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910207385020725e-05,
+      "loss": 1.2502,
+      "step": 1691
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1910024316727877e-05,
+      "loss": 1.0988,
+      "step": 1692
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190984106284946e-05,
+      "loss": 1.191,
+      "step": 1693
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909657623388907e-05,
+      "loss": 1.1649,
+      "step": 1694
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1909473998349658e-05,
+      "loss": 1.1893,
+      "step": 1695
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190929018773515e-05,
+      "loss": 1.1851,
+      "step": 1696
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190910619154882e-05,
+      "loss": 1.1572,
+      "step": 1697
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190892200979412e-05,
+      "loss": 1.1578,
+      "step": 1698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908737642474494e-05,
+      "loss": 1.1538,
+      "step": 1699
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.19085530895934e-05,
+      "loss": 1.1993,
+      "step": 1700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190836835115429e-05,
+      "loss": 1.2323,
+      "step": 1701
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1908183427160628e-05,
+      "loss": 1.2343,
+      "step": 1702
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190799831761587e-05,
+      "loss": 1.2138,
+      "step": 1703
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190781302252349e-05,
+      "loss": 1.1878,
+      "step": 1704
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907627541886954e-05,
+      "loss": 1.1747,
+      "step": 1705
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907441875709734e-05,
+      "loss": 1.1842,
+      "step": 1706
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907256023995314e-05,
+      "loss": 1.1859,
+      "step": 1707
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1907069986747164e-05,
+      "loss": 1.2139,
+      "step": 1708
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906883763968775e-05,
+      "loss": 1.1638,
+      "step": 1709
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906697355663636e-05,
+      "loss": 1.1587,
+      "step": 1710
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1906510761835234e-05,
+      "loss": 1.2093,
+      "step": 1711
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190632398248706e-05,
+      "loss": 1.1628,
+      "step": 1712
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190613701762262e-05,
+      "loss": 1.148,
+      "step": 1713
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905949867245414e-05,
+      "loss": 1.163,
+      "step": 1714
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190576253135894e-05,
+      "loss": 1.1376,
+      "step": 1715
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190557500996671e-05,
+      "loss": 1.2186,
+      "step": 1716
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905387303072237e-05,
+      "loss": 1.1642,
+      "step": 1717
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905199410679038e-05,
+      "loss": 1.09,
+      "step": 1718
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1905011332790627e-05,
+      "loss": 1.2128,
+      "step": 1719
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190482306941053e-05,
+      "loss": 1.102,
+      "step": 1720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190463462054227e-05,
+      "loss": 1.1768,
+      "step": 1721
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904445986189374e-05,
+      "loss": 1.1223,
+      "step": 1722
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190425716635538e-05,
+      "loss": 1.1408,
+      "step": 1723
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1904068161043818e-05,
+      "loss": 1.1221,
+      "step": 1724
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903878970258235e-05,
+      "loss": 1.1724,
+      "step": 1725
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903689594002164e-05,
+      "loss": 1.1184,
+      "step": 1726
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1903500032279167e-05,
+      "loss": 1.0797,
+      "step": 1727
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190331028509278e-05,
+      "loss": 1.1388,
+      "step": 1728
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190312035244656e-05,
+      "loss": 1.1777,
+      "step": 1729
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190293023434406e-05,
+      "loss": 1.2011,
+      "step": 1730
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902739930788852e-05,
+      "loss": 1.1319,
+      "step": 1731
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190254944178449e-05,
+      "loss": 1.1642,
+      "step": 1732
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902358767334544e-05,
+      "loss": 1.2455,
+      "step": 1733
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1902167907442586e-05,
+      "loss": 1.2346,
+      "step": 1734
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190197686211219e-05,
+      "loss": 1.1335,
+      "step": 1735
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190178563134693e-05,
+      "loss": 1.1589,
+      "step": 1736
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190159421515039e-05,
+      "loss": 1.1938,
+      "step": 1737
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190140261352616e-05,
+      "loss": 1.2195,
+      "step": 1738
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1901210826477816e-05,
+      "loss": 1.2059,
+      "step": 1739
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190101885400896e-05,
+      "loss": 1.1827,
+      "step": 1740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900826696123186e-05,
+      "loss": 1.1895,
+      "step": 1741
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900634352824084e-05,
+      "loss": 1.1601,
+      "step": 1742
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900441824115265e-05,
+      "loss": 1.137,
+      "step": 1743
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.190024911000033e-05,
+      "loss": 1.1472,
+      "step": 1744
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1900056210482893e-05,
+      "loss": 1.1773,
+      "step": 1745
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899863125566562e-05,
+      "loss": 1.1157,
+      "step": 1746
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899669855254952e-05,
+      "loss": 1.166,
+      "step": 1747
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899476399551686e-05,
+      "loss": 1.2602,
+      "step": 1748
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899282758460385e-05,
+      "loss": 1.1501,
+      "step": 1749
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1899088931984676e-05,
+      "loss": 1.181,
+      "step": 1750
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1898894920128186e-05,
+      "loss": 1.2181,
+      "step": 1751
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189870072289455e-05,
+      "loss": 1.2318,
+      "step": 1752
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189850634028741e-05,
+      "loss": 1.182,
+      "step": 1753
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1898311772310398e-05,
+      "loss": 1.2185,
+      "step": 1754
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189811701896716e-05,
+      "loss": 1.1761,
+      "step": 1755
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897922080261348e-05,
+      "loss": 1.168,
+      "step": 1756
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189772695619661e-05,
+      "loss": 1.2197,
+      "step": 1757
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189753164677659e-05,
+      "loss": 1.1607,
+      "step": 1758
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1897336152004965e-05,
+      "loss": 1.1159,
+      "step": 1759
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189714047188538e-05,
+      "loss": 1.1654,
+      "step": 1760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896944606421506e-05,
+      "loss": 1.1181,
+      "step": 1761
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896748555617013e-05,
+      "loss": 1.1353,
+      "step": 1762
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896552319475564e-05,
+      "loss": 1.209,
+      "step": 1763
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896355898000844e-05,
+      "loss": 1.1265,
+      "step": 1764
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1896159291196527e-05,
+      "loss": 1.1441,
+      "step": 1765
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895962499066294e-05,
+      "loss": 1.1402,
+      "step": 1766
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189576552161383e-05,
+      "loss": 1.2242,
+      "step": 1767
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895568358842828e-05,
+      "loss": 1.1951,
+      "step": 1768
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1895371010756974e-05,
+      "loss": 1.1539,
+      "step": 1769
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189517347735997e-05,
+      "loss": 1.1518,
+      "step": 1770
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894975758655506e-05,
+      "loss": 1.2379,
+      "step": 1771
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.18947778546473e-05,
+      "loss": 1.231,
+      "step": 1772
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189457976533904e-05,
+      "loss": 1.1553,
+      "step": 1773
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.1894381490734453e-05,
+      "loss": 1.0962,
+      "step": 1774
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.189418303083724e-05,
+      "loss": 1.1176,
+      "step": 1775
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189398438565112e-05,
+      "loss": 1.3001,
+      "step": 1776
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893785555179816e-05,
+      "loss": 1.2221,
+      "step": 1777
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189358653942705e-05,
+      "loss": 1.1627,
+      "step": 1778
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1893387338396548e-05,
+      "loss": 1.1952,
+      "step": 1779
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189318795209204e-05,
+      "loss": 1.1957,
+      "step": 1780
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892988380517262e-05,
+      "loss": 1.1973,
+      "step": 1781
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189278862367595e-05,
+      "loss": 1.1793,
+      "step": 1782
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189258868157185e-05,
+      "loss": 1.1321,
+      "step": 1783
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18923885542087e-05,
+      "loss": 1.1466,
+      "step": 1784
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1892188241590247e-05,
+      "loss": 1.1066,
+      "step": 1785
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891987743720245e-05,
+      "loss": 1.1841,
+      "step": 1786
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189178706060245e-05,
+      "loss": 1.2758,
+      "step": 1787
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189158619224062e-05,
+      "loss": 1.152,
+      "step": 1788
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189138513863851e-05,
+      "loss": 1.2268,
+      "step": 1789
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1891183899799896e-05,
+      "loss": 1.1618,
+      "step": 1790
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890982475728542e-05,
+      "loss": 1.1662,
+      "step": 1791
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.189078086642822e-05,
+      "loss": 1.1106,
+      "step": 1792
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18905790719027e-05,
+      "loss": 1.1531,
+      "step": 1793
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890377092155765e-05,
+      "loss": 1.1867,
+      "step": 1794
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1890174927191204e-05,
+      "loss": 1.1508,
+      "step": 1795
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889972577012795e-05,
+      "loss": 1.0666,
+      "step": 1796
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889770041624327e-05,
+      "loss": 1.2188,
+      "step": 1797
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889567321029598e-05,
+      "loss": 1.2109,
+      "step": 1798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1889364415232402e-05,
+      "loss": 1.0988,
+      "step": 1799
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188916132423654e-05,
+      "loss": 1.2094,
+      "step": 1800
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888958048045815e-05,
+      "loss": 1.2134,
+      "step": 1801
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188875458666403e-05,
+      "loss": 1.1106,
+      "step": 1802
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888550940095e-05,
+      "loss": 1.164,
+      "step": 1803
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888347108342535e-05,
+      "loss": 1.0968,
+      "step": 1804
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1888143091410453e-05,
+      "loss": 1.1789,
+      "step": 1805
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188793888930258e-05,
+      "loss": 1.226,
+      "step": 1806
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887734502022733e-05,
+      "loss": 1.1548,
+      "step": 1807
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887529929574742e-05,
+      "loss": 1.1294,
+      "step": 1808
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887325171962435e-05,
+      "loss": 1.1079,
+      "step": 1809
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1887120229189657e-05,
+      "loss": 1.1299,
+      "step": 1810
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886915101260234e-05,
+      "loss": 1.1257,
+      "step": 1811
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1886709788178015e-05,
+      "loss": 1.1799,
+      "step": 1812
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188650428994684e-05,
+      "loss": 1.2966,
+      "step": 1813
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188629860657056e-05,
+      "loss": 1.2115,
+      "step": 1814
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188609273805303e-05,
+      "loss": 1.1962,
+      "step": 1815
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885886684398097e-05,
+      "loss": 1.1641,
+      "step": 1816
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885680445609625e-05,
+      "loss": 1.1631,
+      "step": 1817
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885474021691475e-05,
+      "loss": 1.1535,
+      "step": 1818
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885267412647518e-05,
+      "loss": 1.1501,
+      "step": 1819
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1885060618481617e-05,
+      "loss": 1.1674,
+      "step": 1820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188485363919764e-05,
+      "loss": 1.1666,
+      "step": 1821
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884646474799477e-05,
+      "loss": 1.1809,
+      "step": 1822
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884439125291e-05,
+      "loss": 1.2185,
+      "step": 1823
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188423159067609e-05,
+      "loss": 1.2268,
+      "step": 1824
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1884023870958635e-05,
+      "loss": 1.153,
+      "step": 1825
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883815966142524e-05,
+      "loss": 1.2195,
+      "step": 1826
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883607876231654e-05,
+      "loss": 1.1124,
+      "step": 1827
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188339960122992e-05,
+      "loss": 1.1814,
+      "step": 1828
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1883191141141218e-05,
+      "loss": 1.2023,
+      "step": 1829
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188298249596946e-05,
+      "loss": 1.1681,
+      "step": 1830
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188277366571855e-05,
+      "loss": 1.2133,
+      "step": 1831
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882564650392396e-05,
+      "loss": 1.1822,
+      "step": 1832
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882355449994916e-05,
+      "loss": 1.1221,
+      "step": 1833
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1882146064530024e-05,
+      "loss": 1.0871,
+      "step": 1834
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881936494001646e-05,
+      "loss": 1.1684,
+      "step": 1835
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881726738413703e-05,
+      "loss": 1.1447,
+      "step": 1836
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881516797770125e-05,
+      "loss": 1.1926,
+      "step": 1837
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881306672074843e-05,
+      "loss": 1.1186,
+      "step": 1838
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1881096361331786e-05,
+      "loss": 1.2443,
+      "step": 1839
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880885865544905e-05,
+      "loss": 1.2109,
+      "step": 1840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188067518471813e-05,
+      "loss": 1.1952,
+      "step": 1841
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880464318855412e-05,
+      "loss": 1.1476,
+      "step": 1842
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1880253267960705e-05,
+      "loss": 1.1645,
+      "step": 1843
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.188004203203795e-05,
+      "loss": 1.1787,
+      "step": 1844
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187983061109111e-05,
+      "loss": 1.161,
+      "step": 1845
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879619005124144e-05,
+      "loss": 1.2044,
+      "step": 1846
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879407214141013e-05,
+      "loss": 1.1087,
+      "step": 1847
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1879195238145683e-05,
+      "loss": 1.1786,
+      "step": 1848
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878983077142125e-05,
+      "loss": 1.1741,
+      "step": 1849
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878770731134313e-05,
+      "loss": 1.1828,
+      "step": 1850
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878558200126224e-05,
+      "loss": 1.1208,
+      "step": 1851
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187834548412183e-05,
+      "loss": 1.174,
+      "step": 1852
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1878132583125125e-05,
+      "loss": 1.1116,
+      "step": 1853
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187791949714009e-05,
+      "loss": 1.2248,
+      "step": 1854
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187770622617072e-05,
+      "loss": 1.1961,
+      "step": 1855
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877492770221008e-05,
+      "loss": 1.1958,
+      "step": 1856
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1877279129294945e-05,
+      "loss": 1.1321,
+      "step": 1857
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187706530339654e-05,
+      "loss": 1.1611,
+      "step": 1858
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876851292529794e-05,
+      "loss": 1.16,
+      "step": 1859
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876637096698712e-05,
+      "loss": 1.1123,
+      "step": 1860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1876422715907314e-05,
+      "loss": 1.124,
+      "step": 1861
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18762081501596e-05,
+      "loss": 1.1699,
+      "step": 1862
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875993399459606e-05,
+      "loss": 1.1492,
+      "step": 1863
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187577846381134e-05,
+      "loss": 1.1202,
+      "step": 1864
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187556334321883e-05,
+      "loss": 1.1362,
+      "step": 1865
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875348037686107e-05,
+      "loss": 1.1481,
+      "step": 1866
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1875132547217203e-05,
+      "loss": 1.0899,
+      "step": 1867
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874916871816152e-05,
+      "loss": 1.103,
+      "step": 1868
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874701011486995e-05,
+      "loss": 1.1392,
+      "step": 1869
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874484966233775e-05,
+      "loss": 1.1597,
+      "step": 1870
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1874268736060532e-05,
+      "loss": 1.1625,
+      "step": 1871
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187405232097132e-05,
+      "loss": 1.2049,
+      "step": 1872
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873835720970192e-05,
+      "loss": 1.2165,
+      "step": 1873
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873618936061205e-05,
+      "loss": 1.27,
+      "step": 1874
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873401966248416e-05,
+      "loss": 1.2017,
+      "step": 1875
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1873184811535892e-05,
+      "loss": 1.1091,
+      "step": 1876
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872967471927695e-05,
+      "loss": 1.1019,
+      "step": 1877
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18727499474279e-05,
+      "loss": 1.1255,
+      "step": 1878
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872532238040573e-05,
+      "loss": 1.1528,
+      "step": 1879
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1872314343769798e-05,
+      "loss": 1.1497,
+      "step": 1880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187209626461965e-05,
+      "loss": 1.0957,
+      "step": 1881
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871878000594224e-05,
+      "loss": 1.0571,
+      "step": 1882
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1871659551697597e-05,
+      "loss": 1.1387,
+      "step": 1883
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187144091793386e-05,
+      "loss": 1.1279,
+      "step": 1884
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187122209930711e-05,
+      "loss": 1.1429,
+      "step": 1885
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187100309582145e-05,
+      "loss": 1.1374,
+      "step": 1886
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870783907480973e-05,
+      "loss": 1.1059,
+      "step": 1887
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870564534289784e-05,
+      "loss": 1.168,
+      "step": 1888
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1870344976252e-05,
+      "loss": 1.1834,
+      "step": 1889
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.187012523337172e-05,
+      "loss": 1.1007,
+      "step": 1890
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869905305653075e-05,
+      "loss": 1.2279,
+      "step": 1891
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869685193100165e-05,
+      "loss": 1.141,
+      "step": 1892
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869464895717128e-05,
+      "loss": 1.1491,
+      "step": 1893
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869244413508082e-05,
+      "loss": 1.2535,
+      "step": 1894
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1869023746477156e-05,
+      "loss": 1.1269,
+      "step": 1895
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868802894628486e-05,
+      "loss": 1.1642,
+      "step": 1896
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1868581857966205e-05,
+      "loss": 1.0676,
+      "step": 1897
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186836063649445e-05,
+      "loss": 1.1565,
+      "step": 1898
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186813923021737e-05,
+      "loss": 1.1531,
+      "step": 1899
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186791763913911e-05,
+      "loss": 1.187,
+      "step": 1900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867695863263813e-05,
+      "loss": 1.1356,
+      "step": 1901
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867473902595635e-05,
+      "loss": 1.107,
+      "step": 1902
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186725175713874e-05,
+      "loss": 1.1769,
+      "step": 1903
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1867029426897282e-05,
+      "loss": 1.1757,
+      "step": 1904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866806911875425e-05,
+      "loss": 1.1199,
+      "step": 1905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866584212077336e-05,
+      "loss": 1.1968,
+      "step": 1906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866361327507184e-05,
+      "loss": 1.1386,
+      "step": 1907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1866138258169145e-05,
+      "loss": 1.1522,
+      "step": 1908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865915004067403e-05,
+      "loss": 1.2153,
+      "step": 1909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865691565206124e-05,
+      "loss": 1.1839,
+      "step": 1910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865467941589504e-05,
+      "loss": 1.1242,
+      "step": 1911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865244133221724e-05,
+      "loss": 1.2174,
+      "step": 1912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1865020140106984e-05,
+      "loss": 1.1817,
+      "step": 1913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186479596224947e-05,
+      "loss": 1.1182,
+      "step": 1914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186457159965339e-05,
+      "loss": 1.1217,
+      "step": 1915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864347052322928e-05,
+      "loss": 1.1161,
+      "step": 1916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1864122320262307e-05,
+      "loss": 1.1145,
+      "step": 1917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186389740347573e-05,
+      "loss": 1.1226,
+      "step": 1918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863672301967407e-05,
+      "loss": 1.1705,
+      "step": 1919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863447015741555e-05,
+      "loss": 1.2369,
+      "step": 1920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1863221544802388e-05,
+      "loss": 1.183,
+      "step": 1921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862995889154138e-05,
+      "loss": 1.2242,
+      "step": 1922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862770048801022e-05,
+      "loss": 1.1308,
+      "step": 1923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862544023747273e-05,
+      "loss": 1.2075,
+      "step": 1924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862317813997126e-05,
+      "loss": 1.2067,
+      "step": 1925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1862091419554817e-05,
+      "loss": 1.1439,
+      "step": 1926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186186484042458e-05,
+      "loss": 1.15,
+      "step": 1927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1861638076610662e-05,
+      "loss": 1.1824,
+      "step": 1928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1861411128117308e-05,
+      "loss": 1.0531,
+      "step": 1929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1861183994948773e-05,
+      "loss": 1.1693,
+      "step": 1930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860956677109304e-05,
+      "loss": 1.0782,
+      "step": 1931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860729174603165e-05,
+      "loss": 1.1869,
+      "step": 1932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186050148743461e-05,
+      "loss": 1.1371,
+      "step": 1933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1860273615607907e-05,
+      "loss": 1.132,
+      "step": 1934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.186004555912732e-05,
+      "loss": 1.0901,
+      "step": 1935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185981731799712e-05,
+      "loss": 1.1622,
+      "step": 1936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859588892221586e-05,
+      "loss": 1.1613,
+      "step": 1937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859360281804992e-05,
+      "loss": 1.2386,
+      "step": 1938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1859131486751624e-05,
+      "loss": 1.2293,
+      "step": 1939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858902507065758e-05,
+      "loss": 1.1956,
+      "step": 1940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858673342751686e-05,
+      "loss": 1.1753,
+      "step": 1941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858443993813703e-05,
+      "loss": 1.2463,
+      "step": 1942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1858214460256096e-05,
+      "loss": 1.1803,
+      "step": 1943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857984742083172e-05,
+      "loss": 1.1684,
+      "step": 1944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857754839299232e-05,
+      "loss": 1.1676,
+      "step": 1945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857524751908577e-05,
+      "loss": 1.2054,
+      "step": 1946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857294479915517e-05,
+      "loss": 1.1434,
+      "step": 1947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1857064023324366e-05,
+      "loss": 1.1991,
+      "step": 1948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185683338213944e-05,
+      "loss": 1.1458,
+      "step": 1949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856602556365056e-05,
+      "loss": 1.1362,
+      "step": 1950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185637154600554e-05,
+      "loss": 1.1329,
+      "step": 1951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1856140351065214e-05,
+      "loss": 1.1145,
+      "step": 1952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185590897154841e-05,
+      "loss": 1.2306,
+      "step": 1953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855677407459462e-05,
+      "loss": 1.1187,
+      "step": 1954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18554456588027e-05,
+      "loss": 1.1056,
+      "step": 1955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1855213725582476e-05,
+      "loss": 1.1428,
+      "step": 1956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854981607803123e-05,
+      "loss": 1.1562,
+      "step": 1957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854749305468992e-05,
+      "loss": 1.204,
+      "step": 1958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854516818584437e-05,
+      "loss": 1.127,
+      "step": 1959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1854284147153802e-05,
+      "loss": 1.2297,
+      "step": 1960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185405129118145e-05,
+      "loss": 1.1659,
+      "step": 1961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853818250671747e-05,
+      "loss": 1.1512,
+      "step": 1962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853585025629048e-05,
+      "loss": 1.1072,
+      "step": 1963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1853351616057724e-05,
+      "loss": 1.1111,
+      "step": 1964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185311802196215e-05,
+      "loss": 1.1692,
+      "step": 1965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852884243346692e-05,
+      "loss": 1.2219,
+      "step": 1966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852650280215733e-05,
+      "loss": 1.1443,
+      "step": 1967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1852416132573658e-05,
+      "loss": 1.2072,
+      "step": 1968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185218180042485e-05,
+      "loss": 1.1673,
+      "step": 1969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185194728377369e-05,
+      "loss": 1.2046,
+      "step": 1970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851712582624578e-05,
+      "loss": 1.1578,
+      "step": 1971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851477696981907e-05,
+      "loss": 1.2302,
+      "step": 1972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851242626850077e-05,
+      "loss": 1.1305,
+      "step": 1973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1851007372233488e-05,
+      "loss": 1.1572,
+      "step": 1974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850771933136545e-05,
+      "loss": 1.1054,
+      "step": 1975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185053630956366e-05,
+      "loss": 1.0874,
+      "step": 1976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1850300501519243e-05,
+      "loss": 1.176,
+      "step": 1977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.185006450900771e-05,
+      "loss": 1.1332,
+      "step": 1978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184982833203348e-05,
+      "loss": 1.1387,
+      "step": 1979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184959197060098e-05,
+      "loss": 1.1338,
+      "step": 1980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1849355424714634e-05,
+      "loss": 1.1795,
+      "step": 1981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1849118694378867e-05,
+      "loss": 1.1378,
+      "step": 1982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184888177959812e-05,
+      "loss": 1.1507,
+      "step": 1983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848644680376826e-05,
+      "loss": 1.1152,
+      "step": 1984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184840739671943e-05,
+      "loss": 1.2089,
+      "step": 1985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1848169928630366e-05,
+      "loss": 1.2178,
+      "step": 1986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847932276114085e-05,
+      "loss": 1.1364,
+      "step": 1987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184769443917504e-05,
+      "loss": 1.1857,
+      "step": 1988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847456417817683e-05,
+      "loss": 1.1676,
+      "step": 1989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1847218212046478e-05,
+      "loss": 1.1334,
+      "step": 1990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846979821865872e-05,
+      "loss": 1.1727,
+      "step": 1991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184674124728034e-05,
+      "loss": 1.1304,
+      "step": 1992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184650248829435e-05,
+      "loss": 1.1506,
+      "step": 1993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846263544912368e-05,
+      "loss": 1.1496,
+      "step": 1994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1846024417138874e-05,
+      "loss": 1.205,
+      "step": 1995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184578510497834e-05,
+      "loss": 1.1389,
+      "step": 1996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184554560843525e-05,
+      "loss": 1.2392,
+      "step": 1997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845305927514094e-05,
+      "loss": 1.1936,
+      "step": 1998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1845066062219356e-05,
+      "loss": 1.1706,
+      "step": 1999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844826012555524e-05,
+      "loss": 1.1613,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_text_loss": 1.0020034313201904,
+      "eval_text_runtime": 1.5966,
+      "eval_text_samples_per_second": 46.348,
+      "eval_text_steps_per_second": 1.253,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "eval_svit_loss": 1.225849986076355,
+      "eval_svit_runtime": 4.2014,
+      "eval_svit_samples_per_second": 162.09,
+      "eval_svit_steps_per_second": 2.618,
+      "step": 2000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844585778527104e-05,
+      "loss": 1.0994,
+      "step": 2001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844345360138585e-05,
+      "loss": 1.1968,
+      "step": 2002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1844104757394473e-05,
+      "loss": 1.208,
+      "step": 2003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843863970299275e-05,
+      "loss": 1.1671,
+      "step": 2004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.18436229988575e-05,
+      "loss": 1.1391,
+      "step": 2005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843381843073657e-05,
+      "loss": 1.1549,
+      "step": 2006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1843140502952267e-05,
+      "loss": 1.1317,
+      "step": 2007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184289897849785e-05,
+      "loss": 1.1157,
+      "step": 2008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.1842657269714924e-05,
+      "loss": 1.1171,
+      "step": 2009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184241537660802e-05,
+      "loss": 1.1584,
+      "step": 2010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.184217329918166e-05,
+      "loss": 1.1554,
+      "step": 2011
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1841931037440388e-05,
+      "loss": 1.1672,
+      "step": 2012
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184168859138874e-05,
+      "loss": 1.089,
+      "step": 2013
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1841445961031246e-05,
+      "loss": 1.1982,
+      "step": 2014
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184120314637246e-05,
+      "loss": 1.1618,
+      "step": 2015
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1840960147416923e-05,
+      "loss": 1.1751,
+      "step": 2016
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184071696416919e-05,
+      "loss": 1.066,
+      "step": 2017
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184047359663381e-05,
+      "loss": 1.2056,
+      "step": 2018
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.184023004481535e-05,
+      "loss": 1.1121,
+      "step": 2019
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839986308718362e-05,
+      "loss": 1.1132,
+      "step": 2020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183974238834741e-05,
+      "loss": 1.141,
+      "step": 2021
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183949828370707e-05,
+      "loss": 1.1014,
+      "step": 2022
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1839253994801908e-05,
+      "loss": 1.1961,
+      "step": 2023
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18390095216365e-05,
+      "loss": 1.1282,
+      "step": 2024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183876486421542e-05,
+      "loss": 1.1315,
+      "step": 2025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838520022543252e-05,
+      "loss": 1.1774,
+      "step": 2026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1838274996624586e-05,
+      "loss": 1.1495,
+      "step": 2027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183802978646401e-05,
+      "loss": 1.2145,
+      "step": 2028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183778439206611e-05,
+      "loss": 1.2418,
+      "step": 2029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837538813435483e-05,
+      "loss": 1.2033,
+      "step": 2030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183729305057673e-05,
+      "loss": 1.1285,
+      "step": 2031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1837047103494456e-05,
+      "loss": 1.1456,
+      "step": 2032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836800972193264e-05,
+      "loss": 1.1154,
+      "step": 2033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183655465667776e-05,
+      "loss": 1.1446,
+      "step": 2034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836308156952564e-05,
+      "loss": 1.1116,
+      "step": 2035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1836061473022286e-05,
+      "loss": 1.1317,
+      "step": 2036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835814604891547e-05,
+      "loss": 1.1458,
+      "step": 2037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835567552564973e-05,
+      "loss": 1.1655,
+      "step": 2038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1835320316047185e-05,
+      "loss": 1.2045,
+      "step": 2039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183507289534282e-05,
+      "loss": 1.0699,
+      "step": 2040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183482529045651e-05,
+      "loss": 1.132,
+      "step": 2041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183457750139289e-05,
+      "loss": 1.1879,
+      "step": 2042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834329528156594e-05,
+      "loss": 1.1601,
+      "step": 2043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1834081370752277e-05,
+      "loss": 1.1639,
+      "step": 2044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183383302918458e-05,
+      "loss": 1.1426,
+      "step": 2045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1833584503458156e-05,
+      "loss": 1.1891,
+      "step": 2046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1833335793577656e-05,
+      "loss": 1.1201,
+      "step": 2047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183308689954774e-05,
+      "loss": 1.1752,
+      "step": 2048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832837821373073e-05,
+      "loss": 1.1954,
+      "step": 2049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832588559058312e-05,
+      "loss": 1.302,
+      "step": 2050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832339112608132e-05,
+      "loss": 1.2247,
+      "step": 2051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1832089482027196e-05,
+      "loss": 1.2862,
+      "step": 2052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831839667320186e-05,
+      "loss": 1.0829,
+      "step": 2053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831589668491778e-05,
+      "loss": 1.2087,
+      "step": 2054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1831339485546654e-05,
+      "loss": 1.1794,
+      "step": 2055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18310891184895e-05,
+      "loss": 1.2224,
+      "step": 2056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830838567325e-05,
+      "loss": 1.1998,
+      "step": 2057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183058783205785e-05,
+      "loss": 1.1816,
+      "step": 2058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1830336912692748e-05,
+      "loss": 1.156,
+      "step": 2059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.183008580923439e-05,
+      "loss": 1.1291,
+      "step": 2060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829834521687476e-05,
+      "loss": 1.2153,
+      "step": 2061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829583050056715e-05,
+      "loss": 1.1672,
+      "step": 2062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1829331394346813e-05,
+      "loss": 1.0662,
+      "step": 2063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182907955456249e-05,
+      "loss": 1.2448,
+      "step": 2064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828827530708456e-05,
+      "loss": 1.2377,
+      "step": 2065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828575322789434e-05,
+      "loss": 1.2086,
+      "step": 2066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828322930810146e-05,
+      "loss": 1.1485,
+      "step": 2067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1828070354775315e-05,
+      "loss": 1.2952,
+      "step": 2068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827817594689677e-05,
+      "loss": 1.1842,
+      "step": 2069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182756465055796e-05,
+      "loss": 1.1373,
+      "step": 2070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827311522384904e-05,
+      "loss": 1.1364,
+      "step": 2071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1827058210175246e-05,
+      "loss": 1.1195,
+      "step": 2072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826804713933737e-05,
+      "loss": 1.1522,
+      "step": 2073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182655103366512e-05,
+      "loss": 1.1617,
+      "step": 2074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826297169374145e-05,
+      "loss": 1.1316,
+      "step": 2075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1826043121065564e-05,
+      "loss": 1.1231,
+      "step": 2076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825788888744137e-05,
+      "loss": 1.1444,
+      "step": 2077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825534472414625e-05,
+      "loss": 1.1308,
+      "step": 2078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1825279872081797e-05,
+      "loss": 1.204,
+      "step": 2079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182502508775041e-05,
+      "loss": 1.2502,
+      "step": 2080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824770119425248e-05,
+      "loss": 1.2066,
+      "step": 2081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824514967111078e-05,
+      "loss": 1.2278,
+      "step": 2082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182425963081268e-05,
+      "loss": 1.1383,
+      "step": 2083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1824004110534832e-05,
+      "loss": 1.1614,
+      "step": 2084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182374840628233e-05,
+      "loss": 1.1772,
+      "step": 2085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182349251805995e-05,
+      "loss": 1.1371,
+      "step": 2086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182323644587249e-05,
+      "loss": 1.1621,
+      "step": 2087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822980189724748e-05,
+      "loss": 1.1439,
+      "step": 2088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822723749621515e-05,
+      "loss": 1.1657,
+      "step": 2089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18224671255676e-05,
+      "loss": 1.1603,
+      "step": 2090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1822210317567807e-05,
+      "loss": 1.1186,
+      "step": 2091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821953325626946e-05,
+      "loss": 1.1085,
+      "step": 2092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821696149749832e-05,
+      "loss": 1.1765,
+      "step": 2093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821438789941275e-05,
+      "loss": 1.1703,
+      "step": 2094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1821181246206096e-05,
+      "loss": 1.0603,
+      "step": 2095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.182092351854912e-05,
+      "loss": 1.1385,
+      "step": 2096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820665606975177e-05,
+      "loss": 1.196,
+      "step": 2097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820407511489093e-05,
+      "loss": 1.172,
+      "step": 2098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1820149232095696e-05,
+      "loss": 1.1722,
+      "step": 2099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1819890768799833e-05,
+      "loss": 1.2005,
+      "step": 2100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1819632121606336e-05,
+      "loss": 1.1297,
+      "step": 2101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181937329052006e-05,
+      "loss": 1.2159,
+      "step": 2102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1819114275545836e-05,
+      "loss": 1.0967,
+      "step": 2103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818855076688526e-05,
+      "loss": 1.1941,
+      "step": 2104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181859569395298e-05,
+      "loss": 1.1019,
+      "step": 2105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1818336127344057e-05,
+      "loss": 1.1564,
+      "step": 2106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181807637686662e-05,
+      "loss": 1.1749,
+      "step": 2107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1817816442525526e-05,
+      "loss": 1.1189,
+      "step": 2108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1817556324325655e-05,
+      "loss": 1.1893,
+      "step": 2109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1817296022271865e-05,
+      "loss": 1.1581,
+      "step": 2110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181703553636904e-05,
+      "loss": 1.1921,
+      "step": 2111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181677486662205e-05,
+      "loss": 1.1851,
+      "step": 2112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816514013035785e-05,
+      "loss": 1.125,
+      "step": 2113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1816252975615122e-05,
+      "loss": 1.1448,
+      "step": 2114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181599175436496e-05,
+      "loss": 1.1139,
+      "step": 2115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181573034929018e-05,
+      "loss": 1.1462,
+      "step": 2116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1815468760395684e-05,
+      "loss": 1.177,
+      "step": 2117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181520698768637e-05,
+      "loss": 1.1925,
+      "step": 2118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814945031167136e-05,
+      "loss": 1.2142,
+      "step": 2119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181468289084289e-05,
+      "loss": 1.1607,
+      "step": 2120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1814420566718548e-05,
+      "loss": 1.1297,
+      "step": 2121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181415805879901e-05,
+      "loss": 1.1556,
+      "step": 2122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813895367089203e-05,
+      "loss": 1.1438,
+      "step": 2123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813632491594038e-05,
+      "loss": 1.1618,
+      "step": 2124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813369432318442e-05,
+      "loss": 1.1048,
+      "step": 2125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1813106189267343e-05,
+      "loss": 1.1377,
+      "step": 2126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181284276244567e-05,
+      "loss": 1.1589,
+      "step": 2127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181257915185835e-05,
+      "loss": 1.1357,
+      "step": 2128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181231535751033e-05,
+      "loss": 1.102,
+      "step": 2129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1812051379406542e-05,
+      "loss": 1.2633,
+      "step": 2130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181178721755193e-05,
+      "loss": 1.158,
+      "step": 2131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181152287195145e-05,
+      "loss": 1.1258,
+      "step": 2132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1811258342610042e-05,
+      "loss": 1.099,
+      "step": 2133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181099362953266e-05,
+      "loss": 1.1844,
+      "step": 2134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181072873272427e-05,
+      "loss": 1.1834,
+      "step": 2135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1810463652189826e-05,
+      "loss": 1.1566,
+      "step": 2136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.181019838793429e-05,
+      "loss": 1.1574,
+      "step": 2137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180993293996264e-05,
+      "loss": 1.1681,
+      "step": 2138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809667308279835e-05,
+      "loss": 1.0923,
+      "step": 2139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809401492890858e-05,
+      "loss": 1.1767,
+      "step": 2140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1809135493800682e-05,
+      "loss": 1.1582,
+      "step": 2141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1808869311014292e-05,
+      "loss": 1.0804,
+      "step": 2142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180860294453667e-05,
+      "loss": 1.1523,
+      "step": 2143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1808336394372805e-05,
+      "loss": 1.117,
+      "step": 2144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1808069660527687e-05,
+      "loss": 1.1568,
+      "step": 2145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807802743006317e-05,
+      "loss": 1.1635,
+      "step": 2146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807535641813686e-05,
+      "loss": 1.1302,
+      "step": 2147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18072683569548e-05,
+      "loss": 1.2313,
+      "step": 2148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1807000888434667e-05,
+      "loss": 1.1322,
+      "step": 2149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1806733236258295e-05,
+      "loss": 1.1399,
+      "step": 2150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180646540043069e-05,
+      "loss": 1.1711,
+      "step": 2151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1806197380956873e-05,
+      "loss": 1.1437,
+      "step": 2152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805929177841865e-05,
+      "loss": 1.2141,
+      "step": 2153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180566079109068e-05,
+      "loss": 1.1521,
+      "step": 2154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805392220708357e-05,
+      "loss": 1.1532,
+      "step": 2155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1805123466699916e-05,
+      "loss": 1.1523,
+      "step": 2156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804854529070392e-05,
+      "loss": 1.098,
+      "step": 2157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804585407824823e-05,
+      "loss": 1.1225,
+      "step": 2158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1804316102968247e-05,
+      "loss": 1.1781,
+      "step": 2159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180404661450571e-05,
+      "loss": 1.2267,
+      "step": 2160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803776942442254e-05,
+      "loss": 1.12,
+      "step": 2161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1803507086782935e-05,
+      "loss": 1.1292,
+      "step": 2162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.18032370475328e-05,
+      "loss": 1.2376,
+      "step": 2163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802966824696914e-05,
+      "loss": 1.138,
+      "step": 2164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180269641828033e-05,
+      "loss": 1.1288,
+      "step": 2165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1802425828288116e-05,
+      "loss": 1.1793,
+      "step": 2166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.180215505472534e-05,
+      "loss": 1.1355,
+      "step": 2167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801884097597072e-05,
+      "loss": 1.141,
+      "step": 2168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801612956908385e-05,
+      "loss": 1.1241,
+      "step": 2169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801341632664354e-05,
+      "loss": 1.2398,
+      "step": 2170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1801070124870066e-05,
+      "loss": 1.157,
+      "step": 2171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800798433530603e-05,
+      "loss": 1.1262,
+      "step": 2172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800526558651053e-05,
+      "loss": 1.2155,
+      "step": 2173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1800254500236506e-05,
+      "loss": 1.1985,
+      "step": 2174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799982258292057e-05,
+      "loss": 1.1754,
+      "step": 2175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799709832822806e-05,
+      "loss": 1.171,
+      "step": 2176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799437223833855e-05,
+      "loss": 1.1803,
+      "step": 2177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1799164431330306e-05,
+      "loss": 1.1737,
+      "step": 2178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798891455317273e-05,
+      "loss": 1.2106,
+      "step": 2179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798618295799864e-05,
+      "loss": 1.1782,
+      "step": 2180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798344952783196e-05,
+      "loss": 1.1323,
+      "step": 2181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1798071426272385e-05,
+      "loss": 1.2166,
+      "step": 2182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179779771627256e-05,
+      "loss": 1.1569,
+      "step": 2183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179752382278884e-05,
+      "loss": 1.2452,
+      "step": 2184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1797249745826353e-05,
+      "loss": 1.1566,
+      "step": 2185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796975485390238e-05,
+      "loss": 1.1611,
+      "step": 2186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179670104148563e-05,
+      "loss": 1.1395,
+      "step": 2187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796426414117662e-05,
+      "loss": 1.0789,
+      "step": 2188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1796151603291487e-05,
+      "loss": 1.19,
+      "step": 2189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795876609012247e-05,
+      "loss": 1.1927,
+      "step": 2190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795601431285085e-05,
+      "loss": 1.1561,
+      "step": 2191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1795326070115167e-05,
+      "loss": 1.1154,
+      "step": 2192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179505052550764e-05,
+      "loss": 1.1489,
+      "step": 2193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1794774797467668e-05,
+      "loss": 1.2008,
+      "step": 2194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179449888600041e-05,
+      "loss": 1.1404,
+      "step": 2195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179422279111104e-05,
+      "loss": 1.2634,
+      "step": 2196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1793946512804724e-05,
+      "loss": 1.1214,
+      "step": 2197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179367005108664e-05,
+      "loss": 1.2333,
+      "step": 2198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1793393405961958e-05,
+      "loss": 1.1339,
+      "step": 2199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179311657743587e-05,
+      "loss": 1.1617,
+      "step": 2200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792839565513547e-05,
+      "loss": 1.1572,
+      "step": 2201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792562370200186e-05,
+      "loss": 1.1792,
+      "step": 2202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179228499150097e-05,
+      "loss": 1.154,
+      "step": 2203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1792007429421103e-05,
+      "loss": 1.1618,
+      "step": 2204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179172968396578e-05,
+      "loss": 1.1336,
+      "step": 2205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.17914517551402e-05,
+      "loss": 1.2336,
+      "step": 2206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1791173642949566e-05,
+      "loss": 1.1707,
+      "step": 2207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179089534739909e-05,
+      "loss": 1.1137,
+      "step": 2208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790616868493982e-05,
+      "loss": 1.0908,
+      "step": 2209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.179033820623946e-05,
+      "loss": 1.2229,
+      "step": 2210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1790059360640735e-05,
+      "loss": 1.1279,
+      "step": 2211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789780331703037e-05,
+      "loss": 1.1524,
+      "step": 2212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178950111943159e-05,
+      "loss": 1.166,
+      "step": 2213
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1789221723831618e-05,
+      "loss": 1.1476,
+      "step": 2214
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1788942144908356e-05,
+      "loss": 1.138,
+      "step": 2215
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1788662382667046e-05,
+      "loss": 1.1762,
+      "step": 2216
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1788382437112915e-05,
+      "loss": 1.1757,
+      "step": 2217
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1788102308251215e-05,
+      "loss": 1.1602,
+      "step": 2218
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787821996087187e-05,
+      "loss": 1.1122,
+      "step": 2219
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787541500626084e-05,
+      "loss": 1.1836,
+      "step": 2220
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1787260821873156e-05,
+      "loss": 1.1721,
+      "step": 2221
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786979959833662e-05,
+      "loss": 1.1618,
+      "step": 2222
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786698914512857e-05,
+      "loss": 1.1584,
+      "step": 2223
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178641768591601e-05,
+      "loss": 1.2471,
+      "step": 2224
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1786136274048382e-05,
+      "loss": 1.1942,
+      "step": 2225
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178585467891525e-05,
+      "loss": 1.1128,
+      "step": 2226
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785572900521882e-05,
+      "loss": 1.1332,
+      "step": 2227
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1785290938873554e-05,
+      "loss": 1.0992,
+      "step": 2228
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178500879397555e-05,
+      "loss": 1.1509,
+      "step": 2229
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178472646583315e-05,
+      "loss": 1.1213,
+      "step": 2230
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1784443954451648e-05,
+      "loss": 1.1447,
+      "step": 2231
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178416125983633e-05,
+      "loss": 1.1984,
+      "step": 2232
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1783878381992486e-05,
+      "loss": 1.1694,
+      "step": 2233
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1783595320925417e-05,
+      "loss": 1.1801,
+      "step": 2234
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178331207664043e-05,
+      "loss": 1.2242,
+      "step": 2235
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178302864914282e-05,
+      "loss": 1.151,
+      "step": 2236
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782745038437898e-05,
+      "loss": 1.159,
+      "step": 2237
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1782461244530975e-05,
+      "loss": 1.1338,
+      "step": 2238
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178217726742737e-05,
+      "loss": 1.1833,
+      "step": 2239
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781893107132395e-05,
+      "loss": 1.1219,
+      "step": 2240
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781608763651373e-05,
+      "loss": 1.1168,
+      "step": 2241
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178132423698963e-05,
+      "loss": 1.1758,
+      "step": 2242
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1781039527152497e-05,
+      "loss": 1.1198,
+      "step": 2243
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.17807546341453e-05,
+      "loss": 1.1349,
+      "step": 2244
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.178046955797338e-05,
+      "loss": 1.1587,
+      "step": 2245
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1780184298642067e-05,
+      "loss": 1.1411,
+      "step": 2246
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1779898856156714e-05,
+      "loss": 1.1576,
+      "step": 2247
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.1779613230522662e-05,
+      "loss": 1.2243,
+      "step": 2248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1779327421745257e-05,
+      "loss": 1.1483,
+      "step": 2249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1779041429829856e-05,
+      "loss": 1.2028,
+      "step": 2250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1778755254781807e-05,
+      "loss": 1.2333,
+      "step": 2251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177846889660648e-05,
+      "loss": 1.1797,
+      "step": 2252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177818235530923e-05,
+      "loss": 1.1712,
+      "step": 2253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777895630895425e-05,
+      "loss": 1.1702,
+      "step": 2254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177760872337043e-05,
+      "loss": 1.1231,
+      "step": 2255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777321632739632e-05,
+      "loss": 1.1362,
+      "step": 2256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1777034359008394e-05,
+      "loss": 1.1353,
+      "step": 2257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17767469021821e-05,
+      "loss": 1.1805,
+      "step": 2258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1776459262266132e-05,
+      "loss": 1.216,
+      "step": 2259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1776171439265883e-05,
+      "loss": 1.1478,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775883433186733e-05,
+      "loss": 1.1509,
+      "step": 2261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177559524403408e-05,
+      "loss": 1.1125,
+      "step": 2262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1775306871813324e-05,
+      "loss": 1.1668,
+      "step": 2263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177501831652986e-05,
+      "loss": 1.1672,
+      "step": 2264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774729578189094e-05,
+      "loss": 1.171,
+      "step": 2265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774440656796437e-05,
+      "loss": 1.1244,
+      "step": 2266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1774151552357295e-05,
+      "loss": 1.1965,
+      "step": 2267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1773862264877083e-05,
+      "loss": 1.1951,
+      "step": 2268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177357279436122e-05,
+      "loss": 1.1671,
+      "step": 2269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1773283140815125e-05,
+      "loss": 1.1627,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772993304244223e-05,
+      "loss": 1.2397,
+      "step": 2271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177270328465394e-05,
+      "loss": 1.1968,
+      "step": 2272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772413082049714e-05,
+      "loss": 1.1686,
+      "step": 2273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1772122696436972e-05,
+      "loss": 1.1702,
+      "step": 2274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771832127821156e-05,
+      "loss": 1.1628,
+      "step": 2275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771541376207703e-05,
+      "loss": 1.1839,
+      "step": 2276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1771250441602063e-05,
+      "loss": 1.1461,
+      "step": 2277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770959324009686e-05,
+      "loss": 1.2479,
+      "step": 2278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770668023436016e-05,
+      "loss": 1.1577,
+      "step": 2279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1770376539886514e-05,
+      "loss": 1.1217,
+      "step": 2280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.177008487336664e-05,
+      "loss": 1.1939,
+      "step": 2281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769793023881847e-05,
+      "loss": 1.2133,
+      "step": 2282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1769500991437613e-05,
+      "loss": 1.142,
+      "step": 2283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17692087760394e-05,
+      "loss": 1.2192,
+      "step": 2284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768916377692677e-05,
+      "loss": 1.1707,
+      "step": 2285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768623796402925e-05,
+      "loss": 1.1989,
+      "step": 2286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768331032175625e-05,
+      "loss": 1.1243,
+      "step": 2287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1768038085016253e-05,
+      "loss": 1.1275,
+      "step": 2288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767744954930303e-05,
+      "loss": 1.1848,
+      "step": 2289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767451641923256e-05,
+      "loss": 1.1594,
+      "step": 2290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1767158146000612e-05,
+      "loss": 1.1414,
+      "step": 2291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176686446716786e-05,
+      "loss": 1.2264,
+      "step": 2292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766570605430508e-05,
+      "loss": 1.1163,
+      "step": 2293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1766276560794056e-05,
+      "loss": 1.1984,
+      "step": 2294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765982333264007e-05,
+      "loss": 1.2106,
+      "step": 2295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765687922845874e-05,
+      "loss": 1.1741,
+      "step": 2296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176539332954517e-05,
+      "loss": 1.107,
+      "step": 2297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1765098553367414e-05,
+      "loss": 1.1713,
+      "step": 2298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764803594318123e-05,
+      "loss": 1.2137,
+      "step": 2299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1764508452402825e-05,
+      "loss": 1.1076,
+      "step": 2300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176421312762704e-05,
+      "loss": 1.1879,
+      "step": 2301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763917619996303e-05,
+      "loss": 1.1985,
+      "step": 2302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763621929516152e-05,
+      "loss": 1.2121,
+      "step": 2303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1763326056192118e-05,
+      "loss": 1.143,
+      "step": 2304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176303000002974e-05,
+      "loss": 1.1703,
+      "step": 2305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762733761034574e-05,
+      "loss": 1.1172,
+      "step": 2306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762437339212153e-05,
+      "loss": 1.1718,
+      "step": 2307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1762140734568038e-05,
+      "loss": 1.2062,
+      "step": 2308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1761843947107782e-05,
+      "loss": 1.2034,
+      "step": 2309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176154697683694e-05,
+      "loss": 1.2028,
+      "step": 2310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1761249823761078e-05,
+      "loss": 1.0804,
+      "step": 2311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760952487885752e-05,
+      "loss": 1.1896,
+      "step": 2312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.176065496921654e-05,
+      "loss": 1.0669,
+      "step": 2313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760357267759012e-05,
+      "loss": 1.1823,
+      "step": 2314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1760059383518737e-05,
+      "loss": 1.1154,
+      "step": 2315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759761316501296e-05,
+      "loss": 1.1247,
+      "step": 2316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759463066712274e-05,
+      "loss": 1.2172,
+      "step": 2317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1759164634157255e-05,
+      "loss": 1.1975,
+      "step": 2318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758866018841824e-05,
+      "loss": 1.1574,
+      "step": 2319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175856722077158e-05,
+      "loss": 1.2421,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1758268239952112e-05,
+      "loss": 1.2084,
+      "step": 2321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757969076389025e-05,
+      "loss": 1.1537,
+      "step": 2322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757669730087918e-05,
+      "loss": 1.132,
+      "step": 2323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1757370201054397e-05,
+      "loss": 1.1571,
+      "step": 2324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175707048929407e-05,
+      "loss": 1.1305,
+      "step": 2325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756770594812556e-05,
+      "loss": 1.1663,
+      "step": 2326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756470517615463e-05,
+      "loss": 1.1982,
+      "step": 2327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1756170257708414e-05,
+      "loss": 1.1934,
+      "step": 2328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175586981509703e-05,
+      "loss": 1.1691,
+      "step": 2329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1755569189786945e-05,
+      "loss": 1.0856,
+      "step": 2330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175526838178378e-05,
+      "loss": 1.1337,
+      "step": 2331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175496739109317e-05,
+      "loss": 1.0707,
+      "step": 2332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175466621772076e-05,
+      "loss": 1.1549,
+      "step": 2333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1754364861672175e-05,
+      "loss": 1.1061,
+      "step": 2334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175406332295307e-05,
+      "loss": 1.0801,
+      "step": 2335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753761601569093e-05,
+      "loss": 1.151,
+      "step": 2336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753459697525885e-05,
+      "loss": 1.283,
+      "step": 2337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1753157610829105e-05,
+      "loss": 1.1116,
+      "step": 2338
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752855341484412e-05,
+      "loss": 1.1324,
+      "step": 2339
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752552889497465e-05,
+      "loss": 1.1127,
+      "step": 2340
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1752250254873927e-05,
+      "loss": 1.082,
+      "step": 2341
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1751947437619463e-05,
+      "loss": 1.2018,
+      "step": 2342
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175164443773975e-05,
+      "loss": 1.1744,
+      "step": 2343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175134125524046e-05,
+      "loss": 1.199,
+      "step": 2344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175103789012727e-05,
+      "loss": 1.1216,
+      "step": 2345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.175073434240586e-05,
+      "loss": 1.2822,
+      "step": 2346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750430612081915e-05,
+      "loss": 1.0488,
+      "step": 2347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1750126699161125e-05,
+      "loss": 1.1462,
+      "step": 2348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174982260364918e-05,
+      "loss": 1.1493,
+      "step": 2349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174951832555178e-05,
+      "loss": 1.1526,
+      "step": 2350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1749213864874613e-05,
+      "loss": 1.1254,
+      "step": 2351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748909221623385e-05,
+      "loss": 1.2008,
+      "step": 2352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748604395803803e-05,
+      "loss": 1.1834,
+      "step": 2353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1748299387421574e-05,
+      "loss": 1.217,
+      "step": 2354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174799419648241e-05,
+      "loss": 1.1274,
+      "step": 2355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174768882299203e-05,
+      "loss": 1.151,
+      "step": 2356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1747383266956146e-05,
+      "loss": 1.1315,
+      "step": 2357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174707752838048e-05,
+      "loss": 1.1466,
+      "step": 2358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174677160727077e-05,
+      "loss": 1.0985,
+      "step": 2359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174646550363273e-05,
+      "loss": 1.2085,
+      "step": 2360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17461592174721e-05,
+      "loss": 1.1732,
+      "step": 2361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745852748794615e-05,
+      "loss": 1.1333,
+      "step": 2362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174554609760601e-05,
+      "loss": 1.1566,
+      "step": 2363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1745239263912037e-05,
+      "loss": 1.1571,
+      "step": 2364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744932247718434e-05,
+      "loss": 1.1186,
+      "step": 2365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744625049030953e-05,
+      "loss": 1.1073,
+      "step": 2366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744317667855346e-05,
+      "loss": 1.2216,
+      "step": 2367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1744010104197373e-05,
+      "loss": 1.13,
+      "step": 2368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1743702358062787e-05,
+      "loss": 1.1068,
+      "step": 2369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174339442945736e-05,
+      "loss": 1.1332,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174308631838685e-05,
+      "loss": 1.1787,
+      "step": 2371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174277802485703e-05,
+      "loss": 1.1794,
+      "step": 2372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1742469548873676e-05,
+      "loss": 1.1403,
+      "step": 2373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174216089044256e-05,
+      "loss": 1.1497,
+      "step": 2374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741852049569467e-05,
+      "loss": 1.1906,
+      "step": 2375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741543026260182e-05,
+      "loss": 1.2354,
+      "step": 2376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1741233820520485e-05,
+      "loss": 1.2463,
+      "step": 2377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740924432356173e-05,
+      "loss": 1.1362,
+      "step": 2378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.174061486177304e-05,
+      "loss": 1.1058,
+      "step": 2379
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1740305108776875e-05,
+      "loss": 1.0892,
+      "step": 2380
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739995173373487e-05,
+      "loss": 1.1451,
+      "step": 2381
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173968505556868e-05,
+      "loss": 1.1037,
+      "step": 2382
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1739374755368258e-05,
+      "loss": 1.1056,
+      "step": 2383
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173906427277803e-05,
+      "loss": 1.1301,
+      "step": 2384
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1738753607803817e-05,
+      "loss": 1.1734,
+      "step": 2385
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173844276045143e-05,
+      "loss": 1.0703,
+      "step": 2386
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17381317307267e-05,
+      "loss": 1.1267,
+      "step": 2387
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1737820518635443e-05,
+      "loss": 1.2035,
+      "step": 2388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1737509124183488e-05,
+      "loss": 1.1772,
+      "step": 2389
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173719754737667e-05,
+      "loss": 1.1218,
+      "step": 2390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173688578822082e-05,
+      "loss": 1.1645,
+      "step": 2391
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1736573846721782e-05,
+      "loss": 1.2138,
+      "step": 2392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1736261722885392e-05,
+      "loss": 1.1492,
+      "step": 2393
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735949416717496e-05,
+      "loss": 1.1664,
+      "step": 2394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735636928223946e-05,
+      "loss": 1.16,
+      "step": 2395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735324257410593e-05,
+      "loss": 1.1709,
+      "step": 2396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1735011404283287e-05,
+      "loss": 1.1035,
+      "step": 2397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1734698368847895e-05,
+      "loss": 1.1561,
+      "step": 2398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1734385151110273e-05,
+      "loss": 1.1621,
+      "step": 2399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173407175107629e-05,
+      "loss": 1.1683,
+      "step": 2400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733758168751813e-05,
+      "loss": 1.2659,
+      "step": 2401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733444404142714e-05,
+      "loss": 1.1717,
+      "step": 2402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1733130457254872e-05,
+      "loss": 1.0904,
+      "step": 2403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1732816328094167e-05,
+      "loss": 1.1143,
+      "step": 2404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173250201666648e-05,
+      "loss": 1.1271,
+      "step": 2405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1732187522977694e-05,
+      "loss": 1.1668,
+      "step": 2406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731872847033697e-05,
+      "loss": 1.1651,
+      "step": 2407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1731557988840394e-05,
+      "loss": 1.1468,
+      "step": 2408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173124294840367e-05,
+      "loss": 1.1218,
+      "step": 2409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730927725729426e-05,
+      "loss": 1.1403,
+      "step": 2410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.173061232082357e-05,
+      "loss": 1.1589,
+      "step": 2411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1730296733692007e-05,
+      "loss": 1.1673,
+      "step": 2412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729980964340643e-05,
+      "loss": 1.0976,
+      "step": 2413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729665012775392e-05,
+      "loss": 1.1582,
+      "step": 2414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172934887900218e-05,
+      "loss": 1.1697,
+      "step": 2415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1729032563026916e-05,
+      "loss": 1.1375,
+      "step": 2416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172871606485553e-05,
+      "loss": 1.1212,
+      "step": 2417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1728399384493944e-05,
+      "loss": 1.1594,
+      "step": 2418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1728082521948096e-05,
+      "loss": 1.2012,
+      "step": 2419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727765477223912e-05,
+      "loss": 1.191,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172744825032734e-05,
+      "loss": 1.1957,
+      "step": 2421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1727130841264307e-05,
+      "loss": 1.1693,
+      "step": 2422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726813250040764e-05,
+      "loss": 1.2185,
+      "step": 2423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172649547666266e-05,
+      "loss": 1.1794,
+      "step": 2424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1726177521135943e-05,
+      "loss": 1.1524,
+      "step": 2425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172585938346657e-05,
+      "loss": 1.1238,
+      "step": 2426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.17255410636605e-05,
+      "loss": 1.2535,
+      "step": 2427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172522256172369e-05,
+      "loss": 1.1492,
+      "step": 2428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1724903877662103e-05,
+      "loss": 1.0818,
+      "step": 2429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1724585011481712e-05,
+      "loss": 1.1907,
+      "step": 2430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172426596318849e-05,
+      "loss": 1.2112,
+      "step": 2431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1723946732788406e-05,
+      "loss": 1.2317,
+      "step": 2432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172362732028744e-05,
+      "loss": 1.1709,
+      "step": 2433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1723307725691576e-05,
+      "loss": 1.1247,
+      "step": 2434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1722987949006796e-05,
+      "loss": 1.1681,
+      "step": 2435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172266799023909e-05,
+      "loss": 1.0959,
+      "step": 2436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172234784939445e-05,
+      "loss": 1.1509,
+      "step": 2437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.172202752647887e-05,
+      "loss": 1.1632,
+      "step": 2438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721707021498355e-05,
+      "loss": 1.1053,
+      "step": 2439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721386334458898e-05,
+      "loss": 1.2185,
+      "step": 2440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1721065465366507e-05,
+      "loss": 1.1505,
+      "step": 2441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720744414227196e-05,
+      "loss": 1.1787,
+      "step": 2442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720423181046973e-05,
+      "loss": 1.1891,
+      "step": 2443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1720101765831854e-05,
+      "loss": 1.152,
+      "step": 2444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1719780168587858e-05,
+      "loss": 1.2052,
+      "step": 2445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1719458389321008e-05,
+      "loss": 1.1549,
+      "step": 2446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171913642803733e-05,
+      "loss": 1.2083,
+      "step": 2447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718814284742856e-05,
+      "loss": 1.07,
+      "step": 2448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718491959443616e-05,
+      "loss": 1.166,
+      "step": 2449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1718169452145645e-05,
+      "loss": 1.174,
+      "step": 2450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717846762854986e-05,
+      "loss": 1.2367,
+      "step": 2451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717523891577678e-05,
+      "loss": 1.1519,
+      "step": 2452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1717200838319773e-05,
+      "loss": 1.2323,
+      "step": 2453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1716877603087312e-05,
+      "loss": 1.0855,
+      "step": 2454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1716554185886356e-05,
+      "loss": 1.1655,
+      "step": 2455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1716230586722964e-05,
+      "loss": 1.1963,
+      "step": 2456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715906805603186e-05,
+      "loss": 1.1414,
+      "step": 2457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1715582842533095e-05,
+      "loss": 1.1478,
+      "step": 2458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171525869751875e-05,
+      "loss": 1.1982,
+      "step": 2459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171493437056623e-05,
+      "loss": 1.2116,
+      "step": 2460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714609861681598e-05,
+      "loss": 1.1755,
+      "step": 2461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1714285170870942e-05,
+      "loss": 1.1336,
+      "step": 2462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713960298140334e-05,
+      "loss": 1.2356,
+      "step": 2463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171363524349586e-05,
+      "loss": 1.0781,
+      "step": 2464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1713310006943615e-05,
+      "loss": 1.1706,
+      "step": 2465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171298458848968e-05,
+      "loss": 1.0509,
+      "step": 2466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1712658988140147e-05,
+      "loss": 1.1656,
+      "step": 2467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.171233320590113e-05,
+      "loss": 1.0542,
+      "step": 2468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1712007241778708e-05,
+      "loss": 1.1509,
+      "step": 2469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711681095779e-05,
+      "loss": 1.1207,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711354767908114e-05,
+      "loss": 1.2081,
+      "step": 2471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1711028258172155e-05,
+      "loss": 1.1903,
+      "step": 2472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710701566577236e-05,
+      "loss": 1.0872,
+      "step": 2473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710374693129483e-05,
+      "loss": 1.172,
+      "step": 2474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1710047637835012e-05,
+      "loss": 1.1753,
+      "step": 2475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170972040069995e-05,
+      "loss": 1.1703,
+      "step": 2476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170939298173042e-05,
+      "loss": 1.1501,
+      "step": 2477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1709065380932563e-05,
+      "loss": 1.1692,
+      "step": 2478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170873759831251e-05,
+      "loss": 1.1155,
+      "step": 2479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708409633876393e-05,
+      "loss": 1.1062,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1708081487630357e-05,
+      "loss": 1.1167,
+      "step": 2481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707753159580554e-05,
+      "loss": 1.1619,
+      "step": 2482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707424649733126e-05,
+      "loss": 1.1662,
+      "step": 2483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.1707095958094224e-05,
+      "loss": 1.1685,
+      "step": 2484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.170676708467001e-05,
+      "loss": 1.1603,
+      "step": 2485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170643802946664e-05,
+      "loss": 1.1892,
+      "step": 2486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170610879249027e-05,
+      "loss": 1.148,
+      "step": 2487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705779373747075e-05,
+      "loss": 1.18,
+      "step": 2488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170544977324322e-05,
+      "loss": 1.1201,
+      "step": 2489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1705119990984877e-05,
+      "loss": 1.1639,
+      "step": 2490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170479002697822e-05,
+      "loss": 1.1875,
+      "step": 2491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704459881229432e-05,
+      "loss": 1.0732,
+      "step": 2492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1704129553744696e-05,
+      "loss": 1.1136,
+      "step": 2493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703799044530193e-05,
+      "loss": 1.2108,
+      "step": 2494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1703468353592117e-05,
+      "loss": 1.1487,
+      "step": 2495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170313748093666e-05,
+      "loss": 1.161,
+      "step": 2496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170280642657002e-05,
+      "loss": 1.184,
+      "step": 2497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170247519049839e-05,
+      "loss": 1.1315,
+      "step": 2498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.170214377272798e-05,
+      "loss": 1.0991,
+      "step": 2499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701812173264993e-05,
+      "loss": 1.1211,
+      "step": 2500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701480392115642e-05,
+      "loss": 1.1115,
+      "step": 2501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1701148429286136e-05,
+      "loss": 1.0956,
+      "step": 2502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700816284782693e-05,
+      "loss": 1.2112,
+      "step": 2503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700483958611534e-05,
+      "loss": 1.1964,
+      "step": 2504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1700151450778885e-05,
+      "loss": 1.1024,
+      "step": 2505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1699818761290968e-05,
+      "loss": 1.1831,
+      "step": 2506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1699485890154016e-05,
+      "loss": 1.1554,
+      "step": 2507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1699152837374264e-05,
+      "loss": 1.2519,
+      "step": 2508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169881960295795e-05,
+      "loss": 1.1165,
+      "step": 2509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1698486186911305e-05,
+      "loss": 1.1866,
+      "step": 2510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1698152589240584e-05,
+      "loss": 1.2095,
+      "step": 2511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1697818809952027e-05,
+      "loss": 1.0974,
+      "step": 2512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169748484905189e-05,
+      "loss": 1.1298,
+      "step": 2513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1697150706546427e-05,
+      "loss": 1.2347,
+      "step": 2514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169681638244189e-05,
+      "loss": 1.1068,
+      "step": 2515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696481876744545e-05,
+      "loss": 1.1971,
+      "step": 2516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1696147189460656e-05,
+      "loss": 1.0735,
+      "step": 2517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695812320596488e-05,
+      "loss": 1.1044,
+      "step": 2518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695477270158312e-05,
+      "loss": 1.1185,
+      "step": 2519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1695142038152407e-05,
+      "loss": 1.1501,
+      "step": 2520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1694806624585047e-05,
+      "loss": 1.1112,
+      "step": 2521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169447102946251e-05,
+      "loss": 1.1821,
+      "step": 2522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169413525279109e-05,
+      "loss": 1.183,
+      "step": 2523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169379929457707e-05,
+      "loss": 1.1637,
+      "step": 2524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693463154826737e-05,
+      "loss": 1.1564,
+      "step": 2525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1693126833546392e-05,
+      "loss": 1.1403,
+      "step": 2526
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1692790330742337e-05,
+      "loss": 1.0964,
+      "step": 2527
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1692453646420862e-05,
+      "loss": 1.1717,
+      "step": 2528
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169211678058828e-05,
+      "loss": 1.1567,
+      "step": 2529
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16917797332509e-05,
+      "loss": 1.144,
+      "step": 2530
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169144250441503e-05,
+      "loss": 1.1352,
+      "step": 2531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169110509408699e-05,
+      "loss": 1.1202,
+      "step": 2532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169076750227309e-05,
+      "loss": 1.2564,
+      "step": 2533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1690429728979666e-05,
+      "loss": 1.1896,
+      "step": 2534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.169009177421303e-05,
+      "loss": 1.0379,
+      "step": 2535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168975363797952e-05,
+      "loss": 1.1309,
+      "step": 2536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1689415320285462e-05,
+      "loss": 1.1901,
+      "step": 2537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1689076821137195e-05,
+      "loss": 1.1421,
+      "step": 2538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1688738140541057e-05,
+      "loss": 1.2099,
+      "step": 2539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1688399278503394e-05,
+      "loss": 1.1799,
+      "step": 2540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168806023503055e-05,
+      "loss": 1.1836,
+      "step": 2541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687721010128867e-05,
+      "loss": 1.1032,
+      "step": 2542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687381603804705e-05,
+      "loss": 1.1251,
+      "step": 2543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1687042016064423e-05,
+      "loss": 1.1882,
+      "step": 2544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1686702246914372e-05,
+      "loss": 1.1891,
+      "step": 2545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1686362296360923e-05,
+      "loss": 1.1261,
+      "step": 2546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168602216441044e-05,
+      "loss": 1.1328,
+      "step": 2547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1685681851069286e-05,
+      "loss": 1.1531,
+      "step": 2548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168534135634384e-05,
+      "loss": 1.0954,
+      "step": 2549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168500068024048e-05,
+      "loss": 1.1732,
+      "step": 2550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1684659822765583e-05,
+      "loss": 1.1293,
+      "step": 2551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1684318783925534e-05,
+      "loss": 1.1482,
+      "step": 2552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168397756372672e-05,
+      "loss": 1.1774,
+      "step": 2553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1683636162175525e-05,
+      "loss": 1.0917,
+      "step": 2554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1683294579278348e-05,
+      "loss": 1.1177,
+      "step": 2555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1682952815041588e-05,
+      "loss": 1.1803,
+      "step": 2556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168261086947164e-05,
+      "loss": 1.1587,
+      "step": 2557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168226874257491e-05,
+      "loss": 1.1487,
+      "step": 2558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1681926434357805e-05,
+      "loss": 1.1033,
+      "step": 2559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1681583944826732e-05,
+      "loss": 1.1323,
+      "step": 2560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168124127398811e-05,
+      "loss": 1.1363,
+      "step": 2561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168089842184836e-05,
+      "loss": 1.1434,
+      "step": 2562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.168055538841389e-05,
+      "loss": 1.118,
+      "step": 2563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1680212173691135e-05,
+      "loss": 1.1748,
+      "step": 2564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679868777686517e-05,
+      "loss": 1.1419,
+      "step": 2565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1679525200406467e-05,
+      "loss": 1.1616,
+      "step": 2566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167918144185742e-05,
+      "loss": 1.0674,
+      "step": 2567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1678837502045814e-05,
+      "loss": 1.1915,
+      "step": 2568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167849338097809e-05,
+      "loss": 1.1198,
+      "step": 2569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1678149078660693e-05,
+      "loss": 1.2221,
+      "step": 2570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677804595100072e-05,
+      "loss": 1.1699,
+      "step": 2571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167745993030267e-05,
+      "loss": 1.1639,
+      "step": 2572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1677115084274954e-05,
+      "loss": 1.1892,
+      "step": 2573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676770057023373e-05,
+      "loss": 1.0786,
+      "step": 2574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1676424848554388e-05,
+      "loss": 1.1585,
+      "step": 2575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167607945887447e-05,
+      "loss": 1.1571,
+      "step": 2576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675733887990087e-05,
+      "loss": 1.1101,
+      "step": 2577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16753881359077e-05,
+      "loss": 1.1751,
+      "step": 2578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1675042202633798e-05,
+      "loss": 1.1568,
+      "step": 2579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167469608817485e-05,
+      "loss": 1.1534,
+      "step": 2580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167434979253734e-05,
+      "loss": 1.1784,
+      "step": 2581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1674003315727754e-05,
+      "loss": 1.1663,
+      "step": 2582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1673656657752578e-05,
+      "loss": 1.1573,
+      "step": 2583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1673309818618305e-05,
+      "loss": 1.1,
+      "step": 2584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672962798331433e-05,
+      "loss": 1.0577,
+      "step": 2585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167261559689846e-05,
+      "loss": 1.1345,
+      "step": 2586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1672268214325887e-05,
+      "loss": 1.1987,
+      "step": 2587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671920650620216e-05,
+      "loss": 1.1214,
+      "step": 2588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167157290578796e-05,
+      "loss": 1.1716,
+      "step": 2589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1671224979835634e-05,
+      "loss": 1.1401,
+      "step": 2590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1670876872769746e-05,
+      "loss": 1.1799,
+      "step": 2591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.167052858459682e-05,
+      "loss": 1.1086,
+      "step": 2592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1670180115323376e-05,
+      "loss": 1.102,
+      "step": 2593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1669831464955946e-05,
+      "loss": 1.1143,
+      "step": 2594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166948263350105e-05,
+      "loss": 1.0718,
+      "step": 2595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1669133620965226e-05,
+      "loss": 1.1291,
+      "step": 2596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668784427355008e-05,
+      "loss": 1.2326,
+      "step": 2597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668435052676935e-05,
+      "loss": 1.1129,
+      "step": 2598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1668085496937556e-05,
+      "loss": 1.1247,
+      "step": 2599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667735760143408e-05,
+      "loss": 1.18,
+      "step": 2600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667385842301046e-05,
+      "loss": 1.1557,
+      "step": 2601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1667035743417023e-05,
+      "loss": 1.1494,
+      "step": 2602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666685463497892e-05,
+      "loss": 1.1894,
+      "step": 2603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1666335002550215e-05,
+      "loss": 1.0774,
+      "step": 2604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665984360580555e-05,
+      "loss": 1.1143,
+      "step": 2605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166563353759548e-05,
+      "loss": 1.1775,
+      "step": 2606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1665282533601556e-05,
+      "loss": 1.1528,
+      "step": 2607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664931348605362e-05,
+      "loss": 1.1628,
+      "step": 2608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664579982613468e-05,
+      "loss": 1.1884,
+      "step": 2609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1664228435632455e-05,
+      "loss": 1.1422,
+      "step": 2610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1663876707668914e-05,
+      "loss": 1.1618,
+      "step": 2611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1663524798729422e-05,
+      "loss": 1.0839,
+      "step": 2612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1663172708820575e-05,
+      "loss": 1.1113,
+      "step": 2613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1662820437948966e-05,
+      "loss": 1.2331,
+      "step": 2614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1662467986121193e-05,
+      "loss": 1.1582,
+      "step": 2615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166211535334385e-05,
+      "loss": 1.1438,
+      "step": 2616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1661762539623548e-05,
+      "loss": 1.1496,
+      "step": 2617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1661409544966894e-05,
+      "loss": 1.2059,
+      "step": 2618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1661056369380492e-05,
+      "loss": 1.1988,
+      "step": 2619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1660703012870963e-05,
+      "loss": 1.1966,
+      "step": 2620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.166034947544492e-05,
+      "loss": 1.1639,
+      "step": 2621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659995757108984e-05,
+      "loss": 1.1298,
+      "step": 2622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659641857869782e-05,
+      "loss": 1.1644,
+      "step": 2623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1659287777733938e-05,
+      "loss": 1.2031,
+      "step": 2624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1658933516708086e-05,
+      "loss": 1.1096,
+      "step": 2625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165857907479886e-05,
+      "loss": 1.1115,
+      "step": 2626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1658224452012892e-05,
+      "loss": 1.2023,
+      "step": 2627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1657869648356833e-05,
+      "loss": 1.1947,
+      "step": 2628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165751466383732e-05,
+      "loss": 1.1326,
+      "step": 2629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1657159498461e-05,
+      "loss": 1.1893,
+      "step": 2630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656804152234528e-05,
+      "loss": 1.0671,
+      "step": 2631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1656448625164562e-05,
+      "loss": 1.0948,
+      "step": 2632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165609291725775e-05,
+      "loss": 1.133,
+      "step": 2633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655737028520763e-05,
+      "loss": 1.1959,
+      "step": 2634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165538095896026e-05,
+      "loss": 1.1815,
+      "step": 2635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1655024708582906e-05,
+      "loss": 1.1899,
+      "step": 2636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1654668277395386e-05,
+      "loss": 1.1897,
+      "step": 2637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165431166540436e-05,
+      "loss": 1.1136,
+      "step": 2638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1653954872616514e-05,
+      "loss": 1.1248,
+      "step": 2639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165359789903853e-05,
+      "loss": 1.1507,
+      "step": 2640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165324074467709e-05,
+      "loss": 1.1651,
+      "step": 2641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652883409538886e-05,
+      "loss": 1.0863,
+      "step": 2642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652525893630603e-05,
+      "loss": 1.1638,
+      "step": 2643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1652168196958943e-05,
+      "loss": 1.0723,
+      "step": 2644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16518103195306e-05,
+      "loss": 1.1869,
+      "step": 2645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1651452261352283e-05,
+      "loss": 1.1235,
+      "step": 2646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.165109402243069e-05,
+      "loss": 1.1405,
+      "step": 2647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650735602772533e-05,
+      "loss": 1.1115,
+      "step": 2648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650377002384525e-05,
+      "loss": 1.1949,
+      "step": 2649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1650018221273375e-05,
+      "loss": 1.2016,
+      "step": 2650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649659259445814e-05,
+      "loss": 1.1725,
+      "step": 2651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1649300116908554e-05,
+      "loss": 1.1924,
+      "step": 2652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164894079366832e-05,
+      "loss": 1.1245,
+      "step": 2653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1648581289731854e-05,
+      "loss": 1.1688,
+      "step": 2654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164822160510587e-05,
+      "loss": 1.1394,
+      "step": 2655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1647861739797122e-05,
+      "loss": 1.1568,
+      "step": 2656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164750169381234e-05,
+      "loss": 1.1472,
+      "step": 2657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164714146715826e-05,
+      "loss": 1.1654,
+      "step": 2658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164678105984164e-05,
+      "loss": 1.1116,
+      "step": 2659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646420471869222e-05,
+      "loss": 1.1984,
+      "step": 2660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1646059703247767e-05,
+      "loss": 1.2126,
+      "step": 2661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164569875398402e-05,
+      "loss": 1.1739,
+      "step": 2662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1645337624084747e-05,
+      "loss": 1.1992,
+      "step": 2663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1644976313556708e-05,
+      "loss": 1.181,
+      "step": 2664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1644614822406675e-05,
+      "loss": 1.1143,
+      "step": 2665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164425315064141e-05,
+      "loss": 1.1235,
+      "step": 2666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643891298267694e-05,
+      "loss": 1.1618,
+      "step": 2667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643529265292298e-05,
+      "loss": 1.154,
+      "step": 2668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1643167051722003e-05,
+      "loss": 1.1108,
+      "step": 2669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1642804657563593e-05,
+      "loss": 1.0849,
+      "step": 2670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164244208282385e-05,
+      "loss": 1.1725,
+      "step": 2671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164207932750957e-05,
+      "loss": 1.1431,
+      "step": 2672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641716391627544e-05,
+      "loss": 1.1057,
+      "step": 2673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1641353275184573e-05,
+      "loss": 1.1432,
+      "step": 2674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1640989978187446e-05,
+      "loss": 1.1078,
+      "step": 2675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164062650064298e-05,
+      "loss": 1.0979,
+      "step": 2676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.164026284255797e-05,
+      "loss": 1.1858,
+      "step": 2677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639899003939235e-05,
+      "loss": 1.1283,
+      "step": 2678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639534984793584e-05,
+      "loss": 1.1519,
+      "step": 2679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1639170785127835e-05,
+      "loss": 1.2108,
+      "step": 2680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163880640494881e-05,
+      "loss": 1.1922,
+      "step": 2681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163844184426333e-05,
+      "loss": 1.1577,
+      "step": 2682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1638077103078224e-05,
+      "loss": 1.1782,
+      "step": 2683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163771218140032e-05,
+      "loss": 1.1943,
+      "step": 2684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1637347079236455e-05,
+      "loss": 1.1539,
+      "step": 2685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1636981796593468e-05,
+      "loss": 1.1216,
+      "step": 2686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1636616333478195e-05,
+      "loss": 1.1484,
+      "step": 2687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163625068989748e-05,
+      "loss": 1.1524,
+      "step": 2688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1635884865858173e-05,
+      "loss": 1.1568,
+      "step": 2689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163551886136713e-05,
+      "loss": 1.0532,
+      "step": 2690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1635152676431192e-05,
+      "loss": 1.1804,
+      "step": 2691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634786311057227e-05,
+      "loss": 1.12,
+      "step": 2692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634419765252095e-05,
+      "loss": 1.1166,
+      "step": 2693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1634053039022654e-05,
+      "loss": 1.154,
+      "step": 2694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633686132375777e-05,
+      "loss": 1.0971,
+      "step": 2695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1633319045318333e-05,
+      "loss": 1.0552,
+      "step": 2696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632951777857203e-05,
+      "loss": 1.2008,
+      "step": 2697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632584329999252e-05,
+      "loss": 1.1076,
+      "step": 2698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1632216701751374e-05,
+      "loss": 1.1624,
+      "step": 2699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1631848893120443e-05,
+      "loss": 1.1982,
+      "step": 2700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1631480904113356e-05,
+      "loss": 1.0964,
+      "step": 2701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1631112734737002e-05,
+      "loss": 1.1821,
+      "step": 2702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.163074438499827e-05,
+      "loss": 1.155,
+      "step": 2703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1630375854904062e-05,
+      "loss": 1.0843,
+      "step": 2704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1630007144461283e-05,
+      "loss": 1.1444,
+      "step": 2705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162963825367683e-05,
+      "loss": 1.1341,
+      "step": 2706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1629269182557625e-05,
+      "loss": 1.1185,
+      "step": 2707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162889993111056e-05,
+      "loss": 1.1943,
+      "step": 2708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628530499342565e-05,
+      "loss": 1.1022,
+      "step": 2709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1628160887260552e-05,
+      "loss": 1.2086,
+      "step": 2710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162779109487145e-05,
+      "loss": 1.1654,
+      "step": 2711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.162742112218217e-05,
+      "loss": 1.0751,
+      "step": 2712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1627050969199657e-05,
+      "loss": 1.0843,
+      "step": 2713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1626680635930835e-05,
+      "loss": 1.1466,
+      "step": 2714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1626310122382634e-05,
+      "loss": 1.1331,
+      "step": 2715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1625939428562e-05,
+      "loss": 1.1493,
+      "step": 2716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1625568554475876e-05,
+      "loss": 1.2177,
+      "step": 2717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.16251975001312e-05,
+      "loss": 1.1794,
+      "step": 2718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1624826265534927e-05,
+      "loss": 1.1432,
+      "step": 2719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1624454850694007e-05,
+      "loss": 1.0831,
+      "step": 2720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.1624083255615392e-05,
+      "loss": 1.1782,
+      "step": 2721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162371148030605e-05,
+      "loss": 1.1759,
+      "step": 2722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1623339524772934e-05,
+      "loss": 1.1486,
+      "step": 2723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622967389023018e-05,
+      "loss": 1.1002,
+      "step": 2724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622595073063265e-05,
+      "loss": 1.1585,
+      "step": 2725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1622222576900647e-05,
+      "loss": 1.0941,
+      "step": 2726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162184990054214e-05,
+      "loss": 1.1113,
+      "step": 2727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1621477043994725e-05,
+      "loss": 1.1411,
+      "step": 2728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1621104007265387e-05,
+      "loss": 1.1321,
+      "step": 2729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.162073079036111e-05,
+      "loss": 1.1401,
+      "step": 2730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1620357393288883e-05,
+      "loss": 1.1214,
+      "step": 2731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1619983816055695e-05,
+      "loss": 1.1824,
+      "step": 2732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1619610058668544e-05,
+      "loss": 1.17,
+      "step": 2733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1619236121134436e-05,
+      "loss": 1.1249,
+      "step": 2734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618862003460363e-05,
+      "loss": 1.1604,
+      "step": 2735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618487705653337e-05,
+      "loss": 1.1449,
+      "step": 2736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1618113227720367e-05,
+      "loss": 1.1229,
+      "step": 2737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1617738569668467e-05,
+      "loss": 1.1712,
+      "step": 2738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161736373150465e-05,
+      "loss": 1.1278,
+      "step": 2739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161698871323594e-05,
+      "loss": 1.1538,
+      "step": 2740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1616613514869354e-05,
+      "loss": 1.1172,
+      "step": 2741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1616238136411924e-05,
+      "loss": 1.1238,
+      "step": 2742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615862577870675e-05,
+      "loss": 1.156,
+      "step": 2743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1615486839252646e-05,
+      "loss": 1.1916,
+      "step": 2744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161511092056487e-05,
+      "loss": 1.1476,
+      "step": 2745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614734821814386e-05,
+      "loss": 1.1126,
+      "step": 2746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1614358543008238e-05,
+      "loss": 1.1415,
+      "step": 2747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161398208415347e-05,
+      "loss": 1.1401,
+      "step": 2748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161360544525714e-05,
+      "loss": 1.0926,
+      "step": 2749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161322862632629e-05,
+      "loss": 1.1232,
+      "step": 2750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161285162736799e-05,
+      "loss": 1.0893,
+      "step": 2751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1612474448389287e-05,
+      "loss": 1.1078,
+      "step": 2752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1612097089397258e-05,
+      "loss": 1.1703,
+      "step": 2753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1611719550398953e-05,
+      "loss": 1.1546,
+      "step": 2754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1611341831401455e-05,
+      "loss": 1.1127,
+      "step": 2755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1610963932411838e-05,
+      "loss": 1.1854,
+      "step": 2756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161058585343717e-05,
+      "loss": 1.1089,
+      "step": 2757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.161020759448454e-05,
+      "loss": 1.0673,
+      "step": 2758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609829155561023e-05,
+      "loss": 1.1642,
+      "step": 2759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1609450536673713e-05,
+      "loss": 1.0947,
+      "step": 2760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16090717378297e-05,
+      "loss": 1.1267,
+      "step": 2761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160869275903607e-05,
+      "loss": 1.2198,
+      "step": 2762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160831360029993e-05,
+      "loss": 1.1962,
+      "step": 2763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160793426162838e-05,
+      "loss": 1.1776,
+      "step": 2764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1607554743028516e-05,
+      "loss": 1.1501,
+      "step": 2765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160717504450745e-05,
+      "loss": 1.0932,
+      "step": 2766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16067951660723e-05,
+      "loss": 1.1694,
+      "step": 2767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606415107730168e-05,
+      "loss": 1.1463,
+      "step": 2768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1606034869488173e-05,
+      "loss": 1.1644,
+      "step": 2769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605654451353442e-05,
+      "loss": 1.182,
+      "step": 2770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1605273853333094e-05,
+      "loss": 1.0898,
+      "step": 2771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604893075434262e-05,
+      "loss": 1.1081,
+      "step": 2772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160451211766407e-05,
+      "loss": 1.109,
+      "step": 2773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1604130980029653e-05,
+      "loss": 1.1545,
+      "step": 2774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160374966253815e-05,
+      "loss": 1.0697,
+      "step": 2775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160336816519671e-05,
+      "loss": 1.1538,
+      "step": 2776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1602986488012467e-05,
+      "loss": 1.052,
+      "step": 2777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160260463099257e-05,
+      "loss": 1.1275,
+      "step": 2778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160222259414417e-05,
+      "loss": 1.0944,
+      "step": 2779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1601840377474432e-05,
+      "loss": 1.1498,
+      "step": 2780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.16014579809905e-05,
+      "loss": 1.1225,
+      "step": 2781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.160107540469954e-05,
+      "loss": 1.1034,
+      "step": 2782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1600692648608716e-05,
+      "loss": 1.1109,
+      "step": 2783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1600309712725196e-05,
+      "loss": 1.1217,
+      "step": 2784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599926597056154e-05,
+      "loss": 1.1692,
+      "step": 2785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1599543301608764e-05,
+      "loss": 1.1774,
+      "step": 2786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15991598263902e-05,
+      "loss": 1.2316,
+      "step": 2787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159877617140765e-05,
+      "loss": 1.1195,
+      "step": 2788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159839233666829e-05,
+      "loss": 1.1436,
+      "step": 2789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1598008322179316e-05,
+      "loss": 1.1396,
+      "step": 2790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1597624127947915e-05,
+      "loss": 1.1784,
+      "step": 2791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1597239753981283e-05,
+      "loss": 1.1585,
+      "step": 2792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596855200286616e-05,
+      "loss": 1.1115,
+      "step": 2793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159647046687112e-05,
+      "loss": 1.0842,
+      "step": 2794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1596085553742e-05,
+      "loss": 1.0746,
+      "step": 2795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159570046090646e-05,
+      "loss": 1.1371,
+      "step": 2796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1595315188371716e-05,
+      "loss": 1.1385,
+      "step": 2797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159492973614498e-05,
+      "loss": 1.1442,
+      "step": 2798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594544104233467e-05,
+      "loss": 1.0864,
+      "step": 2799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1594158292644404e-05,
+      "loss": 1.1381,
+      "step": 2800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159377230138502e-05,
+      "loss": 1.0862,
+      "step": 2801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1593386130462533e-05,
+      "loss": 1.144,
+      "step": 2802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1592999779884182e-05,
+      "loss": 1.2101,
+      "step": 2803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15926132496572e-05,
+      "loss": 1.2261,
+      "step": 2804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159222653978883e-05,
+      "loss": 1.1558,
+      "step": 2805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1591839650286305e-05,
+      "loss": 1.1217,
+      "step": 2806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1591452581156878e-05,
+      "loss": 1.1424,
+      "step": 2807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1591065332407795e-05,
+      "loss": 1.209,
+      "step": 2808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1590677904046305e-05,
+      "loss": 1.1184,
+      "step": 2809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.159029029607967e-05,
+      "loss": 1.2258,
+      "step": 2810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1589902508515146e-05,
+      "loss": 1.1336,
+      "step": 2811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1589514541359992e-05,
+      "loss": 1.1772,
+      "step": 2812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1589126394621474e-05,
+      "loss": 1.225,
+      "step": 2813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1588738068306867e-05,
+      "loss": 1.1229,
+      "step": 2814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1588349562423435e-05,
+      "loss": 1.0819,
+      "step": 2815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158796087697846e-05,
+      "loss": 1.1821,
+      "step": 2816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1587572011979214e-05,
+      "loss": 1.1008,
+      "step": 2817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158718296743299e-05,
+      "loss": 1.2066,
+      "step": 2818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586793743347064e-05,
+      "loss": 1.177,
+      "step": 2819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586404339728728e-05,
+      "loss": 1.1185,
+      "step": 2820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1586014756585276e-05,
+      "loss": 1.1339,
+      "step": 2821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1585624993924e-05,
+      "loss": 1.1554,
+      "step": 2822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1585235051752202e-05,
+      "loss": 1.147,
+      "step": 2823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158484493007719e-05,
+      "loss": 1.1583,
+      "step": 2824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1584454628906257e-05,
+      "loss": 1.131,
+      "step": 2825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158406414824672e-05,
+      "loss": 1.0929,
+      "step": 2826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158367348810589e-05,
+      "loss": 1.1354,
+      "step": 2827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158328264849109e-05,
+      "loss": 1.1721,
+      "step": 2828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582891629409626e-05,
+      "loss": 1.214,
+      "step": 2829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582500430868827e-05,
+      "loss": 1.1747,
+      "step": 2830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1582109052876024e-05,
+      "loss": 1.093,
+      "step": 2831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158171749543854e-05,
+      "loss": 1.1553,
+      "step": 2832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1581325758563708e-05,
+      "loss": 1.1756,
+      "step": 2833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.158093384225887e-05,
+      "loss": 1.0924,
+      "step": 2834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1580541746531356e-05,
+      "loss": 1.2074,
+      "step": 2835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1580149471388517e-05,
+      "loss": 1.1427,
+      "step": 2836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1579757016837692e-05,
+      "loss": 1.1491,
+      "step": 2837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1579364382886235e-05,
+      "loss": 1.1735,
+      "step": 2838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578971569541503e-05,
+      "loss": 1.1828,
+      "step": 2839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1578578576810843e-05,
+      "loss": 1.1238,
+      "step": 2840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157818540470162e-05,
+      "loss": 1.1207,
+      "step": 2841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15777920532212e-05,
+      "loss": 1.1428,
+      "step": 2842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157739852237694e-05,
+      "loss": 1.1974,
+      "step": 2843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157700481217622e-05,
+      "loss": 1.1705,
+      "step": 2844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1576610922626403e-05,
+      "loss": 1.1333,
+      "step": 2845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1576216853734878e-05,
+      "loss": 1.0901,
+      "step": 2846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157582260550901e-05,
+      "loss": 1.0995,
+      "step": 2847
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157542817795619e-05,
+      "loss": 1.1185,
+      "step": 2848
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157503357108381e-05,
+      "loss": 1.1814,
+      "step": 2849
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1574638784899248e-05,
+      "loss": 1.1432,
+      "step": 2850
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1574243819409904e-05,
+      "loss": 1.1956,
+      "step": 2851
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573848674623173e-05,
+      "loss": 1.1244,
+      "step": 2852
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573453350546456e-05,
+      "loss": 1.1516,
+      "step": 2853
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1573057847187156e-05,
+      "loss": 1.1506,
+      "step": 2854
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157266216455268e-05,
+      "loss": 1.0803,
+      "step": 2855
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1572266302650436e-05,
+      "loss": 1.1348,
+      "step": 2856
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157187026148784e-05,
+      "loss": 1.1194,
+      "step": 2857
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1571474041072306e-05,
+      "loss": 1.1666,
+      "step": 2858
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1571077641411255e-05,
+      "loss": 1.1339,
+      "step": 2859
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.157068106251211e-05,
+      "loss": 1.1095,
+      "step": 2860
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15702843043823e-05,
+      "loss": 1.1457,
+      "step": 2861
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1569887367029252e-05,
+      "loss": 1.1473,
+      "step": 2862
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15694902504604e-05,
+      "loss": 1.1148,
+      "step": 2863
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1569092954683186e-05,
+      "loss": 1.0886,
+      "step": 2864
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156869547970504e-05,
+      "loss": 1.1692,
+      "step": 2865
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1568297825533415e-05,
+      "loss": 1.2088,
+      "step": 2866
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567899992175754e-05,
+      "loss": 1.1188,
+      "step": 2867
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156750197963951e-05,
+      "loss": 1.1297,
+      "step": 2868
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1567103787932132e-05,
+      "loss": 1.1078,
+      "step": 2869
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1566705417061078e-05,
+      "loss": 1.1477,
+      "step": 2870
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156630686703381e-05,
+      "loss": 1.1236,
+      "step": 2871
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1565908137857792e-05,
+      "loss": 1.1103,
+      "step": 2872
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156550922954049e-05,
+      "loss": 1.1255,
+      "step": 2873
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1565110142089376e-05,
+      "loss": 1.1922,
+      "step": 2874
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.156471087551192e-05,
+      "loss": 1.1516,
+      "step": 2875
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15643114298156e-05,
+      "loss": 1.107,
+      "step": 2876
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15639118050079e-05,
+      "loss": 1.1693,
+      "step": 2877
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1563512001096302e-05,
+      "loss": 1.1123,
+      "step": 2878
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1563112018088293e-05,
+      "loss": 1.1356,
+      "step": 2879
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562711855991364e-05,
+      "loss": 1.1586,
+      "step": 2880
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1562311514813004e-05,
+      "loss": 1.1142,
+      "step": 2881
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1561910994560718e-05,
+      "loss": 1.1096,
+      "step": 2882
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1561510295242005e-05,
+      "loss": 1.1241,
+      "step": 2883
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1561109416864362e-05,
+      "loss": 1.1481,
+      "step": 2884
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15607083594353e-05,
+      "loss": 1.1068,
+      "step": 2885
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1560307122962336e-05,
+      "loss": 1.2048,
+      "step": 2886
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155990570745298e-05,
+      "loss": 1.1327,
+      "step": 2887
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155950411291474e-05,
+      "loss": 1.1524,
+      "step": 2888
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155910233935515e-05,
+      "loss": 1.1344,
+      "step": 2889
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155870038678173e-05,
+      "loss": 1.1839,
+      "step": 2890
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1558298255202002e-05,
+      "loss": 1.1692,
+      "step": 2891
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1557895944623505e-05,
+      "loss": 1.1346,
+      "step": 2892
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1557493455053766e-05,
+      "loss": 1.1782,
+      "step": 2893
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1557090786500327e-05,
+      "loss": 1.1746,
+      "step": 2894
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1556687938970724e-05,
+      "loss": 1.0971,
+      "step": 2895
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155628491247251e-05,
+      "loss": 1.1709,
+      "step": 2896
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1555881707013223e-05,
+      "loss": 1.1621,
+      "step": 2897
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1555478322600416e-05,
+      "loss": 1.1809,
+      "step": 2898
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155507475924165e-05,
+      "loss": 1.1712,
+      "step": 2899
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1554671016944473e-05,
+      "loss": 1.1604,
+      "step": 2900
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1554267095716455e-05,
+      "loss": 1.1485,
+      "step": 2901
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155386299556515e-05,
+      "loss": 1.1372,
+      "step": 2902
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1553458716498134e-05,
+      "loss": 1.1771,
+      "step": 2903
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1553054258522976e-05,
+      "loss": 1.0685,
+      "step": 2904
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1552649621647246e-05,
+      "loss": 1.1324,
+      "step": 2905
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155224480587853e-05,
+      "loss": 1.2113,
+      "step": 2906
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15518398112244e-05,
+      "loss": 1.0982,
+      "step": 2907
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1551434637692445e-05,
+      "loss": 1.108,
+      "step": 2908
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1551029285290255e-05,
+      "loss": 1.1216,
+      "step": 2909
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.155062375402542e-05,
+      "loss": 1.1553,
+      "step": 2910
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1550218043905528e-05,
+      "loss": 1.1456,
+      "step": 2911
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1549812154938184e-05,
+      "loss": 1.1422,
+      "step": 2912
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1549406087130986e-05,
+      "loss": 1.2208,
+      "step": 2913
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1548999840491536e-05,
+      "loss": 1.1641,
+      "step": 2914
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1548593415027448e-05,
+      "loss": 1.1394,
+      "step": 2915
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154818681074633e-05,
+      "loss": 1.1078,
+      "step": 2916
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1547780027655794e-05,
+      "loss": 1.1376,
+      "step": 2917
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1547373065763464e-05,
+      "loss": 1.1902,
+      "step": 2918
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546965925076953e-05,
+      "loss": 1.1076,
+      "step": 2919
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546558605603895e-05,
+      "loss": 1.183,
+      "step": 2920
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1546151107351912e-05,
+      "loss": 1.0515,
+      "step": 2921
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545743430328635e-05,
+      "loss": 1.167,
+      "step": 2922
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1545335574541697e-05,
+      "loss": 1.1499,
+      "step": 2923
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544927539998742e-05,
+      "loss": 1.0755,
+      "step": 2924
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544519326707407e-05,
+      "loss": 1.0253,
+      "step": 2925
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1544110934675338e-05,
+      "loss": 1.1945,
+      "step": 2926
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154370236391018e-05,
+      "loss": 1.1873,
+      "step": 2927
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1543293614419592e-05,
+      "loss": 1.091,
+      "step": 2928
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154288468621122e-05,
+      "loss": 1.1861,
+      "step": 2929
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1542475579292725e-05,
+      "loss": 1.1235,
+      "step": 2930
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154206629367177e-05,
+      "loss": 1.1297,
+      "step": 2931
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.154165682935602e-05,
+      "loss": 1.1254,
+      "step": 2932
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1541247186353142e-05,
+      "loss": 1.1557,
+      "step": 2933
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540837364670804e-05,
+      "loss": 1.0754,
+      "step": 2934
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540427364316686e-05,
+      "loss": 1.1447,
+      "step": 2935
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1540017185298466e-05,
+      "loss": 1.1125,
+      "step": 2936
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153960682762382e-05,
+      "loss": 1.1078,
+      "step": 2937
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153919629130044e-05,
+      "loss": 1.1019,
+      "step": 2938
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538785576336007e-05,
+      "loss": 1.1498,
+      "step": 2939
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1538374682738218e-05,
+      "loss": 1.113,
+      "step": 2940
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1537963610514763e-05,
+      "loss": 1.0972,
+      "step": 2941
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1537552359673345e-05,
+      "loss": 1.0857,
+      "step": 2942
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1537140930221663e-05,
+      "loss": 1.1983,
+      "step": 2943
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1536729322167422e-05,
+      "loss": 1.1629,
+      "step": 2944
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153631753551833e-05,
+      "loss": 1.1527,
+      "step": 2945
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.15359055702821e-05,
+      "loss": 1.0591,
+      "step": 2946
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1535493426466447e-05,
+      "loss": 1.0936,
+      "step": 2947
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1535081104079087e-05,
+      "loss": 1.1279,
+      "step": 2948
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534668603127745e-05,
+      "loss": 1.1372,
+      "step": 2949
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1534255923620137e-05,
+      "loss": 1.0992,
+      "step": 2950
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1533843065564004e-05,
+      "loss": 1.1625,
+      "step": 2951
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1533430028967068e-05,
+      "loss": 1.0613,
+      "step": 2952
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.153301681383707e-05,
+      "loss": 1.0804,
+      "step": 2953
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1532603420181745e-05,
+      "loss": 1.1409,
+      "step": 2954
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1532189848008834e-05,
+      "loss": 1.085,
+      "step": 2955
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531776097326085e-05,
+      "loss": 1.1695,
+      "step": 2956
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1531362168141245e-05,
+      "loss": 1.1238,
+      "step": 2957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.1530948060462068e-05,
+      "loss": 1.1169,
+      "step": 2958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15305337742963e-05,
+      "loss": 1.0842,
+      "step": 2959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.153011930965171e-05,
+      "loss": 1.1039,
+      "step": 2960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1529704666536056e-05,
+      "loss": 1.1161,
+      "step": 2961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15292898449571e-05,
+      "loss": 1.173,
+      "step": 2962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1528874844922618e-05,
+      "loss": 1.186,
+      "step": 2963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152845966644037e-05,
+      "loss": 1.1771,
+      "step": 2964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152804430951814e-05,
+      "loss": 1.1378,
+      "step": 2965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152762877416371e-05,
+      "loss": 1.1706,
+      "step": 2966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152721306038485e-05,
+      "loss": 1.1343,
+      "step": 2967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152679716818935e-05,
+      "loss": 1.1894,
+      "step": 2968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1526381097585e-05,
+      "loss": 1.0882,
+      "step": 2969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525964848579594e-05,
+      "loss": 1.1492,
+      "step": 2970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525548421180923e-05,
+      "loss": 1.1412,
+      "step": 2971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1525131815396785e-05,
+      "loss": 1.1233,
+      "step": 2972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152471503123499e-05,
+      "loss": 1.1741,
+      "step": 2973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152429806870333e-05,
+      "loss": 1.1086,
+      "step": 2974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1523880927809622e-05,
+      "loss": 1.1533,
+      "step": 2975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1523463608561677e-05,
+      "loss": 1.1659,
+      "step": 2976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1523046110967306e-05,
+      "loss": 1.1713,
+      "step": 2977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1522628435034332e-05,
+      "loss": 1.1285,
+      "step": 2978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.152221058077058e-05,
+      "loss": 1.2224,
+      "step": 2979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521792548183866e-05,
+      "loss": 1.156,
+      "step": 2980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1521374337282024e-05,
+      "loss": 1.1275,
+      "step": 2981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520955948072888e-05,
+      "loss": 1.168,
+      "step": 2982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520537380564286e-05,
+      "loss": 1.1349,
+      "step": 2983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1520118634764065e-05,
+      "loss": 1.0999,
+      "step": 2984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151969971068006e-05,
+      "loss": 1.1405,
+      "step": 2985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151928060832012e-05,
+      "loss": 1.1234,
+      "step": 2986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1518861327692092e-05,
+      "loss": 1.146,
+      "step": 2987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1518441868803828e-05,
+      "loss": 1.1446,
+      "step": 2988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1518022231663186e-05,
+      "loss": 1.1368,
+      "step": 2989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151760241627802e-05,
+      "loss": 1.2329,
+      "step": 2990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1517182422656195e-05,
+      "loss": 1.1627,
+      "step": 2991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1516762250805575e-05,
+      "loss": 1.1531,
+      "step": 2992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1516341900734028e-05,
+      "loss": 1.1904,
+      "step": 2993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515921372449426e-05,
+      "loss": 1.0966,
+      "step": 2994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515500665959645e-05,
+      "loss": 1.2004,
+      "step": 2995
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1515079781272568e-05,
+      "loss": 1.1801,
+      "step": 2996
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1514658718396067e-05,
+      "loss": 1.1227,
+      "step": 2997
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1514237477338037e-05,
+      "loss": 1.1762,
+      "step": 2998
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1513816058106363e-05,
+      "loss": 1.1388,
+      "step": 2999
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1513394460708935e-05,
+      "loss": 1.13,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_text_loss": 1.0012507438659668,
+      "eval_text_runtime": 1.5562,
+      "eval_text_samples_per_second": 47.552,
+      "eval_text_steps_per_second": 1.285,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "eval_svit_loss": 1.208598256111145,
+      "eval_svit_runtime": 4.5434,
+      "eval_svit_samples_per_second": 149.887,
+      "eval_svit_steps_per_second": 2.421,
+      "step": 3000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1512972685153647e-05,
+      "loss": 1.121,
+      "step": 3001
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15125507314484e-05,
+      "loss": 1.1859,
+      "step": 3002
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.15121285996011e-05,
+      "loss": 1.1032,
+      "step": 3003
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.151170628961965e-05,
+      "loss": 1.1262,
+      "step": 3004
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1511283801511952e-05,
+      "loss": 1.2017,
+      "step": 3005
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1510861135285928e-05,
+      "loss": 1.1553,
+      "step": 3006
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1510438290949486e-05,
+      "loss": 1.1518,
+      "step": 3007
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1510015268510542e-05,
+      "loss": 1.2076,
+      "step": 3008
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150959206797703e-05,
+      "loss": 1.117,
+      "step": 3009
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1509168689356866e-05,
+      "loss": 1.1364,
+      "step": 3010
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1508745132657983e-05,
+      "loss": 1.1762,
+      "step": 3011
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1508321397888305e-05,
+      "loss": 1.1261,
+      "step": 3012
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507897485055782e-05,
+      "loss": 1.1048,
+      "step": 3013
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507473394168337e-05,
+      "loss": 1.1253,
+      "step": 3014
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1507049125233923e-05,
+      "loss": 1.1571,
+      "step": 3015
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1506624678260478e-05,
+      "loss": 1.1654,
+      "step": 3016
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1506200053255954e-05,
+      "loss": 1.2267,
+      "step": 3017
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1505775250228303e-05,
+      "loss": 1.1564,
+      "step": 3018
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150535026918548e-05,
+      "loss": 1.1386,
+      "step": 3019
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1504925110135445e-05,
+      "loss": 1.1637,
+      "step": 3020
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150449977308616e-05,
+      "loss": 1.1184,
+      "step": 3021
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1504074258045584e-05,
+      "loss": 1.1489,
+      "step": 3022
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1503648565021696e-05,
+      "loss": 1.1127,
+      "step": 3023
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150322269402246e-05,
+      "loss": 1.0773,
+      "step": 3024
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1502796645055853e-05,
+      "loss": 1.1012,
+      "step": 3025
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1502370418129856e-05,
+      "loss": 1.1292,
+      "step": 3026
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150194401325245e-05,
+      "loss": 1.0934,
+      "step": 3027
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150151743043162e-05,
+      "loss": 1.1878,
+      "step": 3028
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1501090669675354e-05,
+      "loss": 1.1329,
+      "step": 3029
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.150066373099164e-05,
+      "loss": 1.1847,
+      "step": 3030
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1500236614388484e-05,
+      "loss": 1.1077,
+      "step": 3031
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1499809319873876e-05,
+      "loss": 1.0899,
+      "step": 3032
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149938184745582e-05,
+      "loss": 1.1778,
+      "step": 3033
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1498954197142324e-05,
+      "loss": 1.1988,
+      "step": 3034
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149852636894139e-05,
+      "loss": 1.0735,
+      "step": 3035
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1498098362861035e-05,
+      "loss": 1.1911,
+      "step": 3036
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1497670178909276e-05,
+      "loss": 1.1728,
+      "step": 3037
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1497241817094128e-05,
+      "loss": 1.06,
+      "step": 3038
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496813277423615e-05,
+      "loss": 1.1303,
+      "step": 3039
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1496384559905762e-05,
+      "loss": 1.1015,
+      "step": 3040
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14959556645486e-05,
+      "loss": 1.1484,
+      "step": 3041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149552659136015e-05,
+      "loss": 1.1019,
+      "step": 3042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149509734034846e-05,
+      "loss": 1.1317,
+      "step": 3043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1494667911521563e-05,
+      "loss": 1.1529,
+      "step": 3044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14942383048875e-05,
+      "loss": 1.1501,
+      "step": 3045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149380852045432e-05,
+      "loss": 1.1299,
+      "step": 3046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149337855823007e-05,
+      "loss": 1.1385,
+      "step": 3047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1492948418222802e-05,
+      "loss": 1.0524,
+      "step": 3048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1492518100440568e-05,
+      "loss": 1.0906,
+      "step": 3049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1492087604891434e-05,
+      "loss": 1.1294,
+      "step": 3050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1491656931583453e-05,
+      "loss": 1.1237,
+      "step": 3051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1491226080524694e-05,
+      "loss": 1.1677,
+      "step": 3052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.149079505172323e-05,
+      "loss": 1.0594,
+      "step": 3053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1490363845187126e-05,
+      "loss": 1.171,
+      "step": 3054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1489932460924464e-05,
+      "loss": 1.1593,
+      "step": 3055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148950089894332e-05,
+      "loss": 1.1841,
+      "step": 3056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1489069159251768e-05,
+      "loss": 1.1993,
+      "step": 3057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1488637241857902e-05,
+      "loss": 1.1487,
+      "step": 3058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1488205146769813e-05,
+      "loss": 1.1854,
+      "step": 3059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1487772873995583e-05,
+      "loss": 1.1889,
+      "step": 3060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1487340423543314e-05,
+      "loss": 1.1702,
+      "step": 3061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486907795421104e-05,
+      "loss": 1.1661,
+      "step": 3062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486474989637057e-05,
+      "loss": 1.1271,
+      "step": 3063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1486042006199267e-05,
+      "loss": 1.1122,
+      "step": 3064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1485608845115856e-05,
+      "loss": 1.1385,
+      "step": 3065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148517550639493e-05,
+      "loss": 1.1183,
+      "step": 3066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1484741990044604e-05,
+      "loss": 1.1133,
+      "step": 3067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1484308296072996e-05,
+      "loss": 1.1583,
+      "step": 3068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1483874424488232e-05,
+      "loss": 1.052,
+      "step": 3069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1483440375298433e-05,
+      "loss": 1.0838,
+      "step": 3070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1483006148511727e-05,
+      "loss": 1.1904,
+      "step": 3071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148257174413625e-05,
+      "loss": 1.1259,
+      "step": 3072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.148213716218013e-05,
+      "loss": 1.1028,
+      "step": 3073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1481702402651515e-05,
+      "loss": 1.133,
+      "step": 3074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1481267465558537e-05,
+      "loss": 1.0754,
+      "step": 3075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1480832350909345e-05,
+      "loss": 1.1243,
+      "step": 3076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1480397058712092e-05,
+      "loss": 1.1647,
+      "step": 3077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479961588974925e-05,
+      "loss": 1.1729,
+      "step": 3078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479525941706e-05,
+      "loss": 1.0768,
+      "step": 3079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1479090116913473e-05,
+      "loss": 1.1536,
+      "step": 3080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147865411460551e-05,
+      "loss": 1.1616,
+      "step": 3081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1478217934790274e-05,
+      "loss": 1.064,
+      "step": 3082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1477781577475934e-05,
+      "loss": 1.1556,
+      "step": 3083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147734504267066e-05,
+      "loss": 1.1906,
+      "step": 3084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147690833038263e-05,
+      "loss": 1.2279,
+      "step": 3085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476471440620018e-05,
+      "loss": 1.2238,
+      "step": 3086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1476034373391007e-05,
+      "loss": 1.0943,
+      "step": 3087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1475597128703783e-05,
+      "loss": 1.1603,
+      "step": 3088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1475159706566538e-05,
+      "loss": 1.1401,
+      "step": 3089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1474722106987457e-05,
+      "loss": 1.1837,
+      "step": 3090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147428432997474e-05,
+      "loss": 1.1157,
+      "step": 3091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147384637553658e-05,
+      "loss": 1.1516,
+      "step": 3092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1473408243681186e-05,
+      "loss": 1.0957,
+      "step": 3093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472969934416757e-05,
+      "loss": 1.2267,
+      "step": 3094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472531447751502e-05,
+      "loss": 1.0653,
+      "step": 3095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1472092783693635e-05,
+      "loss": 1.1232,
+      "step": 3096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1471653942251368e-05,
+      "loss": 1.1589,
+      "step": 3097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1471214923432924e-05,
+      "loss": 1.0809,
+      "step": 3098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1470775727246516e-05,
+      "loss": 1.1135,
+      "step": 3099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.147033635370038e-05,
+      "loss": 1.16,
+      "step": 3100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1469896802802735e-05,
+      "loss": 1.1098,
+      "step": 3101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1469457074561814e-05,
+      "loss": 1.1431,
+      "step": 3102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1469017168985855e-05,
+      "loss": 1.1728,
+      "step": 3103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1468577086083098e-05,
+      "loss": 1.1456,
+      "step": 3104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1468136825861776e-05,
+      "loss": 1.2262,
+      "step": 3105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146769638833014e-05,
+      "loss": 1.1375,
+      "step": 3106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1467255773496437e-05,
+      "loss": 1.1024,
+      "step": 3107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1466814981368918e-05,
+      "loss": 1.1328,
+      "step": 3108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146637401195584e-05,
+      "loss": 1.1755,
+      "step": 3109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1465932865265458e-05,
+      "loss": 1.1034,
+      "step": 3110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146549154130604e-05,
+      "loss": 1.1359,
+      "step": 3111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1465050040085836e-05,
+      "loss": 1.1893,
+      "step": 3112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146460836161313e-05,
+      "loss": 1.1593,
+      "step": 3113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1464166505896185e-05,
+      "loss": 1.0058,
+      "step": 3114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1463724472943274e-05,
+      "loss": 1.1714,
+      "step": 3115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1463282262762684e-05,
+      "loss": 1.2227,
+      "step": 3116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1462839875362686e-05,
+      "loss": 1.102,
+      "step": 3117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.146239731075157e-05,
+      "loss": 1.1604,
+      "step": 3118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461954568937624e-05,
+      "loss": 1.1311,
+      "step": 3119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461511649929137e-05,
+      "loss": 1.1131,
+      "step": 3120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1461068553734405e-05,
+      "loss": 1.1484,
+      "step": 3121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1460625280361726e-05,
+      "loss": 1.1771,
+      "step": 3122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1460181829819404e-05,
+      "loss": 1.093,
+      "step": 3123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145973820211574e-05,
+      "loss": 1.159,
+      "step": 3124
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145929439725904e-05,
+      "loss": 1.1618,
+      "step": 3125
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145885041525762e-05,
+      "loss": 1.0806,
+      "step": 3126
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1458406256119786e-05,
+      "loss": 1.1822,
+      "step": 3127
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1457961919853867e-05,
+      "loss": 1.1692,
+      "step": 3128
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145751740646818e-05,
+      "loss": 1.1009,
+      "step": 3129
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1457072715971046e-05,
+      "loss": 1.1365,
+      "step": 3130
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1456627848370793e-05,
+      "loss": 1.1592,
+      "step": 3131
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1456182803675757e-05,
+      "loss": 1.1635,
+      "step": 3132
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145573758189427e-05,
+      "loss": 1.1533,
+      "step": 3133
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1455292183034668e-05,
+      "loss": 1.1042,
+      "step": 3134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145484660710529e-05,
+      "loss": 1.1581,
+      "step": 3135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145440085411449e-05,
+      "loss": 1.1018,
+      "step": 3136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1453954924070606e-05,
+      "loss": 1.1145,
+      "step": 3137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145350881698199e-05,
+      "loss": 1.0927,
+      "step": 3138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1453062532857e-05,
+      "loss": 1.0894,
+      "step": 3139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1452616071703994e-05,
+      "loss": 1.178,
+      "step": 3140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145216943353133e-05,
+      "loss": 1.0536,
+      "step": 3141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145172261834737e-05,
+      "loss": 1.1566,
+      "step": 3142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.145127562616049e-05,
+      "loss": 1.1091,
+      "step": 3143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1450828456979054e-05,
+      "loss": 1.1156,
+      "step": 3144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1450381110811434e-05,
+      "loss": 1.1317,
+      "step": 3145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1449933587666014e-05,
+      "loss": 1.1863,
+      "step": 3146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144948588755117e-05,
+      "loss": 1.1162,
+      "step": 3147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1449038010475292e-05,
+      "loss": 1.1181,
+      "step": 3148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144858995644676e-05,
+      "loss": 1.2145,
+      "step": 3149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144814172547397e-05,
+      "loss": 1.1095,
+      "step": 3150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1447693317565312e-05,
+      "loss": 1.1101,
+      "step": 3151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144724473272919e-05,
+      "loss": 1.1214,
+      "step": 3152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1446795970973995e-05,
+      "loss": 1.15,
+      "step": 3153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1446347032308137e-05,
+      "loss": 1.0888,
+      "step": 3154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445897916740024e-05,
+      "loss": 1.0937,
+      "step": 3155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1445448624278062e-05,
+      "loss": 1.1549,
+      "step": 3156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1444999154930674e-05,
+      "loss": 1.1379,
+      "step": 3157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144454950870627e-05,
+      "loss": 1.1306,
+      "step": 3158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144409968561327e-05,
+      "loss": 1.1386,
+      "step": 3159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14436496856601e-05,
+      "loss": 1.1249,
+      "step": 3160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1443199508855188e-05,
+      "loss": 1.1071,
+      "step": 3161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144274915520696e-05,
+      "loss": 1.1485,
+      "step": 3162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1442298624723855e-05,
+      "loss": 1.1525,
+      "step": 3163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1441847917414308e-05,
+      "loss": 1.0911,
+      "step": 3164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1441397033286765e-05,
+      "loss": 1.1579,
+      "step": 3165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1440945972349656e-05,
+      "loss": 1.1595,
+      "step": 3166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.144049473461144e-05,
+      "loss": 1.1062,
+      "step": 3167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1440043320080567e-05,
+      "loss": 1.0914,
+      "step": 3168
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1439591728765483e-05,
+      "loss": 1.0825,
+      "step": 3169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143913996067465e-05,
+      "loss": 1.093,
+      "step": 3170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143868801581653e-05,
+      "loss": 1.0876,
+      "step": 3171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143823589419958e-05,
+      "loss": 1.0359,
+      "step": 3172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1437783595832276e-05,
+      "loss": 1.0391,
+      "step": 3173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1437331120723083e-05,
+      "loss": 1.1692,
+      "step": 3174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1436878468880473e-05,
+      "loss": 1.1399,
+      "step": 3175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1436425640312925e-05,
+      "loss": 1.234,
+      "step": 3176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143597263502892e-05,
+      "loss": 1.1075,
+      "step": 3177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143551945303694e-05,
+      "loss": 1.1387,
+      "step": 3178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1435066094345472e-05,
+      "loss": 1.1348,
+      "step": 3179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143461255896301e-05,
+      "loss": 1.1566,
+      "step": 3180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143415884689804e-05,
+      "loss": 1.1828,
+      "step": 3181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.143370495815906e-05,
+      "loss": 1.1212,
+      "step": 3182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1433250892754576e-05,
+      "loss": 1.1669,
+      "step": 3183
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1432796650693087e-05,
+      "loss": 1.1573,
+      "step": 3184
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.14323422319831e-05,
+      "loss": 1.203,
+      "step": 3185
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1431887636633126e-05,
+      "loss": 1.1469,
+      "step": 3186
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1431432864651673e-05,
+      "loss": 1.1476,
+      "step": 3187
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1430977916047266e-05,
+      "loss": 1.1611,
+      "step": 3188
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1430522790828423e-05,
+      "loss": 1.0862,
+      "step": 3189
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1430067489003664e-05,
+      "loss": 1.1421,
+      "step": 3190
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1429612010581513e-05,
+      "loss": 1.1305,
+      "step": 3191
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1429156355570507e-05,
+      "loss": 1.1008,
+      "step": 3192
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.142870052397917e-05,
+      "loss": 1.1767,
+      "step": 3193
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1428244515816048e-05,
+      "loss": 1.0723,
+      "step": 3194
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.1427788331089674e-05,
+      "loss": 1.1326,
+      "step": 3195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1427331969808596e-05,
+      "loss": 1.0961,
+      "step": 3196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1426875431981356e-05,
+      "loss": 1.1504,
+      "step": 3197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142641871761651e-05,
+      "loss": 1.1076,
+      "step": 3198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14259618267226e-05,
+      "loss": 1.1691,
+      "step": 3199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142550475930819e-05,
+      "loss": 1.1693,
+      "step": 3200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1425047515381836e-05,
+      "loss": 1.0535,
+      "step": 3201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1424590094952106e-05,
+      "loss": 1.1169,
+      "step": 3202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142413249802756e-05,
+      "loss": 1.1025,
+      "step": 3203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1423674724616773e-05,
+      "loss": 1.0468,
+      "step": 3204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1423216774728313e-05,
+      "loss": 1.1569,
+      "step": 3205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142275864837076e-05,
+      "loss": 1.1051,
+      "step": 3206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142230034555269e-05,
+      "loss": 1.0715,
+      "step": 3207
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1421841866282687e-05,
+      "loss": 1.1445,
+      "step": 3208
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142138321056934e-05,
+      "loss": 1.1622,
+      "step": 3209
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142092437842123e-05,
+      "loss": 1.1779,
+      "step": 3210
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1420465369846962e-05,
+      "loss": 1.1243,
+      "step": 3211
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.142000618485512e-05,
+      "loss": 1.1456,
+      "step": 3212
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141954682345431e-05,
+      "loss": 1.1106,
+      "step": 3213
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141908728565313e-05,
+      "loss": 1.164,
+      "step": 3214
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141862757146019e-05,
+      "loss": 1.1866,
+      "step": 3215
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1418167680884098e-05,
+      "loss": 1.1073,
+      "step": 3216
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1417707613933464e-05,
+      "loss": 1.1948,
+      "step": 3217
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141724737061691e-05,
+      "loss": 1.082,
+      "step": 3218
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1416786950943046e-05,
+      "loss": 1.1911,
+      "step": 3219
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14163263549205e-05,
+      "loss": 1.1706,
+      "step": 3220
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1415865582557894e-05,
+      "loss": 1.12,
+      "step": 3221
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141540463386386e-05,
+      "loss": 1.0399,
+      "step": 3222
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1414943508847035e-05,
+      "loss": 1.0941,
+      "step": 3223
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1414482207516046e-05,
+      "loss": 1.0835,
+      "step": 3224
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1414020729879532e-05,
+      "loss": 1.1182,
+      "step": 3225
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1413559075946142e-05,
+      "loss": 1.1051,
+      "step": 3226
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1413097245724514e-05,
+      "loss": 1.123,
+      "step": 3227
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14126352392233e-05,
+      "loss": 1.1201,
+      "step": 3228
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1412173056451155e-05,
+      "loss": 1.1715,
+      "step": 3229
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1411710697416728e-05,
+      "loss": 1.1134,
+      "step": 3230
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1411248162128686e-05,
+      "loss": 1.1152,
+      "step": 3231
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.141078545059568e-05,
+      "loss": 1.1302,
+      "step": 3232
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1410322562826387e-05,
+      "loss": 1.1172,
+      "step": 3233
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1409859498829467e-05,
+      "loss": 1.1593,
+      "step": 3234
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1409396258613593e-05,
+      "loss": 1.1209,
+      "step": 3235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1408932842187443e-05,
+      "loss": 1.1289,
+      "step": 3236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1408469249559696e-05,
+      "loss": 1.1509,
+      "step": 3237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1408005480739027e-05,
+      "loss": 1.1671,
+      "step": 3238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1407541535734133e-05,
+      "loss": 1.1999,
+      "step": 3239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140707741455369e-05,
+      "loss": 1.155,
+      "step": 3240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1406613117206397e-05,
+      "loss": 1.1216,
+      "step": 3241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1406148643700946e-05,
+      "loss": 1.1896,
+      "step": 3242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1405683994046036e-05,
+      "loss": 1.1331,
+      "step": 3243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1405219168250367e-05,
+      "loss": 1.1602,
+      "step": 3244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404754166322647e-05,
+      "loss": 1.0818,
+      "step": 3245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1404288988271583e-05,
+      "loss": 1.158,
+      "step": 3246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1403823634105882e-05,
+      "loss": 1.0546,
+      "step": 3247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1403358103834267e-05,
+      "loss": 1.0741,
+      "step": 3248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.140289239746545e-05,
+      "loss": 1.1504,
+      "step": 3249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1402426515008153e-05,
+      "loss": 1.2043,
+      "step": 3250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.14019604564711e-05,
+      "loss": 1.2573,
+      "step": 3251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401494221863027e-05,
+      "loss": 1.1291,
+      "step": 3252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1401027811192653e-05,
+      "loss": 1.1591,
+      "step": 3253
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1400561224468718e-05,
+      "loss": 1.1157,
+      "step": 3254
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1400094461699962e-05,
+      "loss": 1.1645,
+      "step": 3255
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1399627522895124e-05,
+      "loss": 1.1649,
+      "step": 3256
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1399160408062948e-05,
+      "loss": 1.0877,
+      "step": 3257
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139869311721218e-05,
+      "loss": 1.1288,
+      "step": 3258
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1398225650351575e-05,
+      "loss": 1.15,
+      "step": 3259
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1397758007489884e-05,
+      "loss": 1.1221,
+      "step": 3260
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139729018863587e-05,
+      "loss": 1.1663,
+      "step": 3261
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1396822193798284e-05,
+      "loss": 1.1538,
+      "step": 3262
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1396354022985897e-05,
+      "loss": 1.1978,
+      "step": 3263
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1395885676207478e-05,
+      "loss": 1.1581,
+      "step": 3264
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1395417153471794e-05,
+      "loss": 1.0863,
+      "step": 3265
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1394948454787618e-05,
+      "loss": 1.1261,
+      "step": 3266
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139447958016373e-05,
+      "loss": 1.171,
+      "step": 3267
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139401052960891e-05,
+      "loss": 1.1057,
+      "step": 3268
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1393541303131946e-05,
+      "loss": 1.1996,
+      "step": 3269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139307190074162e-05,
+      "loss": 1.1775,
+      "step": 3270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.139260232244672e-05,
+      "loss": 1.1122,
+      "step": 3271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1392132568256046e-05,
+      "loss": 1.1527,
+      "step": 3272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1391662638178397e-05,
+      "loss": 1.094,
+      "step": 3273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1391192532222562e-05,
+      "loss": 1.1169,
+      "step": 3274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390722250397354e-05,
+      "loss": 1.1439,
+      "step": 3275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1390251792711583e-05,
+      "loss": 1.1713,
+      "step": 3276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1389781159174046e-05,
+      "loss": 1.1583,
+      "step": 3277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138931034979357e-05,
+      "loss": 1.1222,
+      "step": 3278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1388839364578967e-05,
+      "loss": 1.1393,
+      "step": 3279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1388368203539052e-05,
+      "loss": 1.0378,
+      "step": 3280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138789686668266e-05,
+      "loss": 1.1411,
+      "step": 3281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1387425354018606e-05,
+      "loss": 1.164,
+      "step": 3282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1386953665555723e-05,
+      "loss": 1.1349,
+      "step": 3283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1386481801302852e-05,
+      "loss": 1.135,
+      "step": 3284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1386009761268825e-05,
+      "loss": 1.1153,
+      "step": 3285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385537545462474e-05,
+      "loss": 1.1547,
+      "step": 3286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1385065153892653e-05,
+      "loss": 1.2408,
+      "step": 3287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1384592586568205e-05,
+      "loss": 1.1031,
+      "step": 3288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1384119843497977e-05,
+      "loss": 1.1636,
+      "step": 3289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1383646924690827e-05,
+      "loss": 1.228,
+      "step": 3290
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138317383015561e-05,
+      "loss": 1.1883,
+      "step": 3291
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382700559901183e-05,
+      "loss": 1.0811,
+      "step": 3292
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1382227113936408e-05,
+      "loss": 1.1431,
+      "step": 3293
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.138175349227016e-05,
+      "loss": 1.1116,
+      "step": 3294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1381279694911296e-05,
+      "loss": 1.1543,
+      "step": 3295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1380805721868695e-05,
+      "loss": 1.131,
+      "step": 3296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1380331573151238e-05,
+      "loss": 1.2003,
+      "step": 3297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1379857248767797e-05,
+      "loss": 1.0903,
+      "step": 3298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1379382748727257e-05,
+      "loss": 1.1703,
+      "step": 3299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1378908073038507e-05,
+      "loss": 1.1098,
+      "step": 3300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137843322171043e-05,
+      "loss": 1.1378,
+      "step": 3301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1377958194751922e-05,
+      "loss": 1.1369,
+      "step": 3302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137748299217188e-05,
+      "loss": 1.1822,
+      "step": 3303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1377007613979206e-05,
+      "loss": 1.0815,
+      "step": 3304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1376532060182794e-05,
+      "loss": 1.1212,
+      "step": 3305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1376056330791556e-05,
+      "loss": 1.0538,
+      "step": 3306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1375580425814397e-05,
+      "loss": 1.0832,
+      "step": 3307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1375104345260234e-05,
+      "loss": 1.1045,
+      "step": 3308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137462808913798e-05,
+      "loss": 1.1514,
+      "step": 3309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1374151657456557e-05,
+      "loss": 1.1349,
+      "step": 3310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1373675050224883e-05,
+      "loss": 1.0551,
+      "step": 3311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137319826745188e-05,
+      "loss": 1.1341,
+      "step": 3312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1372721309146484e-05,
+      "loss": 1.1082,
+      "step": 3313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1372244175317625e-05,
+      "loss": 1.1089,
+      "step": 3314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371766865974238e-05,
+      "loss": 1.1617,
+      "step": 3315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1371289381125263e-05,
+      "loss": 1.0433,
+      "step": 3316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1370811720779636e-05,
+      "loss": 1.1499,
+      "step": 3317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.137033388494631e-05,
+      "loss": 1.1172,
+      "step": 3318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1369855873634227e-05,
+      "loss": 1.1324,
+      "step": 3319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1369377686852345e-05,
+      "loss": 1.1645,
+      "step": 3320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1368899324609614e-05,
+      "loss": 1.086,
+      "step": 3321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1368420786914992e-05,
+      "loss": 1.1748,
+      "step": 3322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367942073777448e-05,
+      "loss": 1.1522,
+      "step": 3323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1367463185205936e-05,
+      "loss": 1.1751,
+      "step": 3324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366984121209432e-05,
+      "loss": 1.171,
+      "step": 3325
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366504881796904e-05,
+      "loss": 1.144,
+      "step": 3326
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1366025466977328e-05,
+      "loss": 1.1608,
+      "step": 3327
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1365545876759684e-05,
+      "loss": 1.126,
+      "step": 3328
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1365066111152952e-05,
+      "loss": 1.139,
+      "step": 3329
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136458617016611e-05,
+      "loss": 1.1476,
+      "step": 3330
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136410605380816e-05,
+      "loss": 1.2558,
+      "step": 3331
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1363625762088077e-05,
+      "loss": 1.1246,
+      "step": 3332
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136314529501487e-05,
+      "loss": 1.1125,
+      "step": 3333
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1362664652597523e-05,
+      "loss": 1.1042,
+      "step": 3334
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1362183834845045e-05,
+      "loss": 1.1943,
+      "step": 3335
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1361702841766443e-05,
+      "loss": 1.03,
+      "step": 3336
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136122167337072e-05,
+      "loss": 1.1038,
+      "step": 3337
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136074032966689e-05,
+      "loss": 1.1281,
+      "step": 3338
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.136025881066396e-05,
+      "loss": 1.0727,
+      "step": 3339
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1359777116370955e-05,
+      "loss": 1.167,
+      "step": 3340
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135929524679689e-05,
+      "loss": 1.1279,
+      "step": 3341
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1358813201950796e-05,
+      "loss": 1.1426,
+      "step": 3342
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1358330981841697e-05,
+      "loss": 1.2104,
+      "step": 3343
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1357848586478618e-05,
+      "loss": 1.081,
+      "step": 3344
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1357366015870602e-05,
+      "loss": 1.0791,
+      "step": 3345
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1356883270026683e-05,
+      "loss": 1.1159,
+      "step": 3346
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1356400348955895e-05,
+      "loss": 1.1964,
+      "step": 3347
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1355917252667292e-05,
+      "loss": 1.1685,
+      "step": 3348
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135543398116991e-05,
+      "loss": 1.1568,
+      "step": 3349
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1354950534472814e-05,
+      "loss": 1.2085,
+      "step": 3350
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135446691258504e-05,
+      "loss": 1.1519,
+      "step": 3351
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135398311551566e-05,
+      "loss": 1.1203,
+      "step": 3352
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1353499143273724e-05,
+      "loss": 1.1568,
+      "step": 3353
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13530149958683e-05,
+      "loss": 1.0879,
+      "step": 3354
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352530673308456e-05,
+      "loss": 1.1631,
+      "step": 3355
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1352046175603256e-05,
+      "loss": 1.1161,
+      "step": 3356
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.135156150276178e-05,
+      "loss": 1.1573,
+      "step": 3357
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1351076654793098e-05,
+      "loss": 1.1492,
+      "step": 3358
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1350591631706297e-05,
+      "loss": 1.2624,
+      "step": 3359
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1350106433510453e-05,
+      "loss": 1.1118,
+      "step": 3360
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1349621060214655e-05,
+      "loss": 1.1554,
+      "step": 3361
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1349135511827994e-05,
+      "loss": 1.1917,
+      "step": 3362
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1348649788359563e-05,
+      "loss": 1.188,
+      "step": 3363
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1348163889818457e-05,
+      "loss": 1.1251,
+      "step": 3364
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1347677816213775e-05,
+      "loss": 1.1686,
+      "step": 3365
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134719156755462e-05,
+      "loss": 1.1032,
+      "step": 3366
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1346705143850098e-05,
+      "loss": 1.0985,
+      "step": 3367
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134621854510932e-05,
+      "loss": 1.1384,
+      "step": 3368
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1345731771341395e-05,
+      "loss": 1.1496,
+      "step": 3369
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1345244822555444e-05,
+      "loss": 1.0852,
+      "step": 3370
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1344757698760584e-05,
+      "loss": 1.0851,
+      "step": 3371
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134427039996593e-05,
+      "loss": 1.0718,
+      "step": 3372
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1343782926180622e-05,
+      "loss": 1.0759,
+      "step": 3373
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1343295277413776e-05,
+      "loss": 1.152,
+      "step": 3374
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134280745367453e-05,
+      "loss": 1.0982,
+      "step": 3375
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.134231945497202e-05,
+      "loss": 1.1172,
+      "step": 3376
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341831281315384e-05,
+      "loss": 1.1825,
+      "step": 3377
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1341342932713762e-05,
+      "loss": 1.0664,
+      "step": 3378
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1340854409176302e-05,
+      "loss": 1.1369,
+      "step": 3379
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1340365710712153e-05,
+      "loss": 1.1531,
+      "step": 3380
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1339876837330466e-05,
+      "loss": 1.1101,
+      "step": 3381
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133938778904039e-05,
+      "loss": 1.1507,
+      "step": 3382
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1338898565851095e-05,
+      "loss": 1.1046,
+      "step": 3383
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1338409167771735e-05,
+      "loss": 1.2696,
+      "step": 3384
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133791959481148e-05,
+      "loss": 1.0909,
+      "step": 3385
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1337429846979492e-05,
+      "loss": 1.2161,
+      "step": 3386
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1336939924284943e-05,
+      "loss": 1.0878,
+      "step": 3387
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1336449826737017e-05,
+      "loss": 1.1482,
+      "step": 3388
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1335959554344878e-05,
+      "loss": 1.046,
+      "step": 3389
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133546910711772e-05,
+      "loss": 1.0948,
+      "step": 3390
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133497848506472e-05,
+      "loss": 1.0746,
+      "step": 3391
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1334487688195067e-05,
+      "loss": 1.1219,
+      "step": 3392
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133399671651796e-05,
+      "loss": 1.1762,
+      "step": 3393
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133350557004258e-05,
+      "loss": 1.1105,
+      "step": 3394
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1333014248778136e-05,
+      "loss": 1.163,
+      "step": 3395
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1332522752733823e-05,
+      "loss": 1.1655,
+      "step": 3396
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.133203108191885e-05,
+      "loss": 1.1457,
+      "step": 3397
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1331539236342417e-05,
+      "loss": 1.1931,
+      "step": 3398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1331047216013742e-05,
+      "loss": 1.1795,
+      "step": 3399
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1330555020942035e-05,
+      "loss": 1.1016,
+      "step": 3400
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1330062651136514e-05,
+      "loss": 1.0671,
+      "step": 3401
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1329570106606402e-05,
+      "loss": 1.1284,
+      "step": 3402
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1329077387360922e-05,
+      "loss": 1.1668,
+      "step": 3403
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.13285844934093e-05,
+      "loss": 1.138,
+      "step": 3404
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1328091424760767e-05,
+      "loss": 1.1213,
+      "step": 3405
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1327598181424557e-05,
+      "loss": 1.0529,
+      "step": 3406
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132710476340991e-05,
+      "loss": 1.1294,
+      "step": 3407
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1326611170726057e-05,
+      "loss": 1.172,
+      "step": 3408
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1326117403382252e-05,
+      "loss": 1.0819,
+      "step": 3409
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1325623461387734e-05,
+      "loss": 1.1772,
+      "step": 3410
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1325129344751757e-05,
+      "loss": 1.0837,
+      "step": 3411
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1324635053483576e-05,
+      "loss": 1.1363,
+      "step": 3412
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1324140587592445e-05,
+      "loss": 1.1066,
+      "step": 3413
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1323645947087623e-05,
+      "loss": 1.0656,
+      "step": 3414
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1323151131978377e-05,
+      "loss": 1.1438,
+      "step": 3415
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1322656142273967e-05,
+      "loss": 1.0992,
+      "step": 3416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132216097798367e-05,
+      "loss": 1.1519,
+      "step": 3417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132166563911675e-05,
+      "loss": 1.1288,
+      "step": 3418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1321170125682498e-05,
+      "loss": 1.1808,
+      "step": 3419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1320674437690177e-05,
+      "loss": 1.1982,
+      "step": 3420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.132017857514908e-05,
+      "loss": 1.1111,
+      "step": 3421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1319682538068487e-05,
+      "loss": 1.1366,
+      "step": 3422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1319186326457695e-05,
+      "loss": 1.1229,
+      "step": 3423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131868994032599e-05,
+      "loss": 1.112,
+      "step": 3424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131819337968267e-05,
+      "loss": 1.1284,
+      "step": 3425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.131769664453703e-05,
+      "loss": 1.0627,
+      "step": 3426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1317199734898383e-05,
+      "loss": 1.1582,
+      "step": 3427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1316702650776024e-05,
+      "loss": 1.0564,
+      "step": 3428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1316205392179268e-05,
+      "loss": 1.1308,
+      "step": 3429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1315707959117425e-05,
+      "loss": 1.1592,
+      "step": 3430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.1315210351599808e-05,
+      "loss": 1.1703,
+      "step": 3431
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131471256963574e-05,
+      "loss": 1.1359,
+      "step": 3432
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1314214613234542e-05,
+      "loss": 1.1388,
+      "step": 3433
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1313716482405537e-05,
+      "loss": 1.07,
+      "step": 3434
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1313218177158056e-05,
+      "loss": 1.2073,
+      "step": 3435
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131271969750143e-05,
+      "loss": 1.0965,
+      "step": 3436
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1312221043444994e-05,
+      "loss": 1.1765,
+      "step": 3437
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1311722214998085e-05,
+      "loss": 1.131,
+      "step": 3438
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1311223212170047e-05,
+      "loss": 1.1506,
+      "step": 3439
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131072403497022e-05,
+      "loss": 1.0844,
+      "step": 3440
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.131022468340796e-05,
+      "loss": 1.1492,
+      "step": 3441
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1309725157492616e-05,
+      "loss": 1.0708,
+      "step": 3442
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130922545723354e-05,
+      "loss": 1.1413,
+      "step": 3443
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1308725582640086e-05,
+      "loss": 1.1407,
+      "step": 3444
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1308225533721625e-05,
+      "loss": 1.1233,
+      "step": 3445
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1307725310487514e-05,
+      "loss": 1.0696,
+      "step": 3446
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1307224912947124e-05,
+      "loss": 1.1333,
+      "step": 3447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1306724341109823e-05,
+      "loss": 1.1508,
+      "step": 3448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1306223594984988e-05,
+      "loss": 1.1553,
+      "step": 3449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1305722674581998e-05,
+      "loss": 1.2383,
+      "step": 3450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130522157991023e-05,
+      "loss": 1.1196,
+      "step": 3451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130472031097907e-05,
+      "loss": 1.1615,
+      "step": 3452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1304218867797907e-05,
+      "loss": 1.0781,
+      "step": 3453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1303717250376126e-05,
+      "loss": 1.2505,
+      "step": 3454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1303215458723124e-05,
+      "loss": 1.1517,
+      "step": 3455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.13027134928483e-05,
+      "loss": 1.1093,
+      "step": 3456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130221135276105e-05,
+      "loss": 1.1267,
+      "step": 3457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1301709038470784e-05,
+      "loss": 1.1178,
+      "step": 3458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1301206549986904e-05,
+      "loss": 1.1832,
+      "step": 3459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1300703887318823e-05,
+      "loss": 1.15,
+      "step": 3460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.130020105047595e-05,
+      "loss": 1.1128,
+      "step": 3461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1299698039467707e-05,
+      "loss": 1.1238,
+      "step": 3462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1299194854303507e-05,
+      "loss": 1.0835,
+      "step": 3463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1298691494992778e-05,
+      "loss": 1.1512,
+      "step": 3464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129818796154495e-05,
+      "loss": 1.1649,
+      "step": 3465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1297684253969447e-05,
+      "loss": 1.1595,
+      "step": 3466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12971803722757e-05,
+      "loss": 1.1435,
+      "step": 3467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1296676316473152e-05,
+      "loss": 1.1432,
+      "step": 3468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1296172086571237e-05,
+      "loss": 1.1214,
+      "step": 3469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1295667682579403e-05,
+      "loss": 1.0793,
+      "step": 3470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129516310450709e-05,
+      "loss": 1.183,
+      "step": 3471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129465835236375e-05,
+      "loss": 1.1317,
+      "step": 3472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1294153426158842e-05,
+      "loss": 1.1639,
+      "step": 3473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129364832590181e-05,
+      "loss": 1.1828,
+      "step": 3474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1293143051602117e-05,
+      "loss": 1.1391,
+      "step": 3475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129263760326923e-05,
+      "loss": 1.1693,
+      "step": 3476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129213198091261e-05,
+      "loss": 1.1048,
+      "step": 3477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129162618454173e-05,
+      "loss": 1.133,
+      "step": 3478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1291120214166055e-05,
+      "loss": 1.1965,
+      "step": 3479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.129061406979507e-05,
+      "loss": 1.1344,
+      "step": 3480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1290107751438243e-05,
+      "loss": 1.0899,
+      "step": 3481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289601259105064e-05,
+      "loss": 1.1234,
+      "step": 3482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1289094592805013e-05,
+      "loss": 1.1313,
+      "step": 3483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128858775254758e-05,
+      "loss": 1.1762,
+      "step": 3484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128808073834226e-05,
+      "loss": 1.1228,
+      "step": 3485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1287573550198544e-05,
+      "loss": 1.0484,
+      "step": 3486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128706618812593e-05,
+      "loss": 1.1325,
+      "step": 3487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1286558652133924e-05,
+      "loss": 1.0847,
+      "step": 3488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1286050942232023e-05,
+      "loss": 1.0486,
+      "step": 3489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128554305842974e-05,
+      "loss": 1.11,
+      "step": 3490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128503500073659e-05,
+      "loss": 1.1638,
+      "step": 3491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1284526769162078e-05,
+      "loss": 1.1353,
+      "step": 3492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1284018363715728e-05,
+      "loss": 1.1616,
+      "step": 3493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128350978440706e-05,
+      "loss": 1.1111,
+      "step": 3494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1283001031245594e-05,
+      "loss": 1.1492,
+      "step": 3495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1282492104240865e-05,
+      "loss": 1.093,
+      "step": 3496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1281983003402397e-05,
+      "loss": 1.1313,
+      "step": 3497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.128147372873973e-05,
+      "loss": 1.1128,
+      "step": 3498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1280964280262396e-05,
+      "loss": 1.0947,
+      "step": 3499
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1280454657979935e-05,
+      "loss": 1.1038,
+      "step": 3500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1279944861901895e-05,
+      "loss": 1.0939,
+      "step": 3501
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127943489203782e-05,
+      "loss": 1.159,
+      "step": 3502
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1278924748397264e-05,
+      "loss": 1.1141,
+      "step": 3503
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127841443098978e-05,
+      "loss": 1.1266,
+      "step": 3504
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1277903939824915e-05,
+      "loss": 1.0796,
+      "step": 3505
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1277393274912242e-05,
+      "loss": 1.1738,
+      "step": 3506
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1276882436261318e-05,
+      "loss": 1.197,
+      "step": 3507
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127637142388171e-05,
+      "loss": 1.1629,
+      "step": 3508
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1275860237782987e-05,
+      "loss": 1.1436,
+      "step": 3509
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1275348877974723e-05,
+      "loss": 1.148,
+      "step": 3510
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1274837344466497e-05,
+      "loss": 1.155,
+      "step": 3511
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1274325637267883e-05,
+      "loss": 1.0802,
+      "step": 3512
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1273813756388467e-05,
+      "loss": 1.1424,
+      "step": 3513
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1273301701837837e-05,
+      "loss": 1.1281,
+      "step": 3514
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1272789473625575e-05,
+      "loss": 1.097,
+      "step": 3515
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1272277071761285e-05,
+      "loss": 1.0996,
+      "step": 3516
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.127176449625455e-05,
+      "loss": 1.0865,
+      "step": 3517
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1271251747114976e-05,
+      "loss": 1.1472,
+      "step": 3518
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1270738824352166e-05,
+      "loss": 1.1041,
+      "step": 3519
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1270225727975728e-05,
+      "loss": 1.1341,
+      "step": 3520
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126971245799526e-05,
+      "loss": 1.14,
+      "step": 3521
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1269199014420384e-05,
+      "loss": 1.1324,
+      "step": 3522
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1268685397260716e-05,
+      "loss": 1.1157,
+      "step": 3523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1268171606525865e-05,
+      "loss": 1.1371,
+      "step": 3524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126765764222546e-05,
+      "loss": 1.0512,
+      "step": 3525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1267143504369122e-05,
+      "loss": 1.1628,
+      "step": 3526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126662919296649e-05,
+      "loss": 1.0966,
+      "step": 3527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1266114708027178e-05,
+      "loss": 1.2056,
+      "step": 3528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1265600049560835e-05,
+      "loss": 1.1393,
+      "step": 3529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1265085217577092e-05,
+      "loss": 1.1637,
+      "step": 3530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1264570212085592e-05,
+      "loss": 1.1465,
+      "step": 3531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126405503309598e-05,
+      "loss": 1.1214,
+      "step": 3532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1263539680617906e-05,
+      "loss": 1.1242,
+      "step": 3533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1263024154661018e-05,
+      "loss": 1.1449,
+      "step": 3534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1262508455234975e-05,
+      "loss": 1.165,
+      "step": 3535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1261992582349424e-05,
+      "loss": 1.1141,
+      "step": 3536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1261476536014037e-05,
+      "loss": 1.1482,
+      "step": 3537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.126096031623847e-05,
+      "loss": 1.1551,
+      "step": 3538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1260443923032397e-05,
+      "loss": 1.0689,
+      "step": 3539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125992735640548e-05,
+      "loss": 1.1526,
+      "step": 3540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1259410616367403e-05,
+      "loss": 1.0935,
+      "step": 3541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1258893702927837e-05,
+      "loss": 1.107,
+      "step": 3542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125837661609646e-05,
+      "loss": 1.1329,
+      "step": 3543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1257859355882963e-05,
+      "loss": 1.1385,
+      "step": 3544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1257341922297028e-05,
+      "loss": 1.0835,
+      "step": 3545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1256824315348345e-05,
+      "loss": 1.1238,
+      "step": 3546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1256306535046608e-05,
+      "loss": 1.1901,
+      "step": 3547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125578858140151e-05,
+      "loss": 1.1634,
+      "step": 3548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125527045442276e-05,
+      "loss": 1.102,
+      "step": 3549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1254752154120048e-05,
+      "loss": 1.176,
+      "step": 3550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125423368050309e-05,
+      "loss": 1.1753,
+      "step": 3551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1253715033581598e-05,
+      "loss": 1.1377,
+      "step": 3552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1253196213365273e-05,
+      "loss": 1.1841,
+      "step": 3553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125267721986384e-05,
+      "loss": 1.131,
+      "step": 3554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1252158053087015e-05,
+      "loss": 1.1186,
+      "step": 3555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.125163871304452e-05,
+      "loss": 1.1245,
+      "step": 3556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1251119199746082e-05,
+      "loss": 1.1898,
+      "step": 3557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1250599513201433e-05,
+      "loss": 1.1203,
+      "step": 3558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1250079653420302e-05,
+      "loss": 1.1366,
+      "step": 3559
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124955962041242e-05,
+      "loss": 1.1626,
+      "step": 3560
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1249039414187533e-05,
+      "loss": 1.1046,
+      "step": 3561
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124851903475538e-05,
+      "loss": 1.1188,
+      "step": 3562
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1247998482125707e-05,
+      "loss": 1.1598,
+      "step": 3563
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124747775630826e-05,
+      "loss": 1.1307,
+      "step": 3564
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1246956857312795e-05,
+      "loss": 1.1218,
+      "step": 3565
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1246435785149062e-05,
+      "loss": 1.223,
+      "step": 3566
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1245914539826817e-05,
+      "loss": 1.2048,
+      "step": 3567
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1245393121355832e-05,
+      "loss": 1.1545,
+      "step": 3568
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1244871529745865e-05,
+      "loss": 1.1624,
+      "step": 3569
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1244349765006683e-05,
+      "loss": 1.0819,
+      "step": 3570
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1243827827148058e-05,
+      "loss": 1.0775,
+      "step": 3571
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1243305716179768e-05,
+      "loss": 1.1177,
+      "step": 3572
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124278343211158e-05,
+      "loss": 1.1772,
+      "step": 3573
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1242260974953288e-05,
+      "loss": 1.0628,
+      "step": 3574
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124173834471467e-05,
+      "loss": 1.1426,
+      "step": 3575
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1241215541405513e-05,
+      "loss": 1.1532,
+      "step": 3576
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1240692565035607e-05,
+      "loss": 1.1608,
+      "step": 3577
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.124016941561475e-05,
+      "loss": 1.0966,
+      "step": 3578
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123964609315273e-05,
+      "loss": 1.1195,
+      "step": 3579
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1239122597659358e-05,
+      "loss": 1.2003,
+      "step": 3580
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1238598929144434e-05,
+      "loss": 1.0753,
+      "step": 3581
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123807508761776e-05,
+      "loss": 1.1249,
+      "step": 3582
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123755107308915e-05,
+      "loss": 1.0978,
+      "step": 3583
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1237026885568418e-05,
+      "loss": 1.0747,
+      "step": 3584
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123650252506538e-05,
+      "loss": 1.176,
+      "step": 3585
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1235977991589855e-05,
+      "loss": 1.1125,
+      "step": 3586
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1235453285151665e-05,
+      "loss": 1.1509,
+      "step": 3587
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1234928405760636e-05,
+      "loss": 1.1243,
+      "step": 3588
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12344033534266e-05,
+      "loss": 1.1179,
+      "step": 3589
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123387812815939e-05,
+      "loss": 1.136,
+      "step": 3590
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1233352729968838e-05,
+      "loss": 1.1693,
+      "step": 3591
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123282715886479e-05,
+      "loss": 1.0497,
+      "step": 3592
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1232301414857076e-05,
+      "loss": 1.1382,
+      "step": 3593
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1231775497955556e-05,
+      "loss": 1.2083,
+      "step": 3594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1231249408170068e-05,
+      "loss": 1.1454,
+      "step": 3595
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123072314551047e-05,
+      "loss": 1.1856,
+      "step": 3596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.123019670998662e-05,
+      "loss": 1.1475,
+      "step": 3597
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1229670101608367e-05,
+      "loss": 1.1565,
+      "step": 3598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122914332038558e-05,
+      "loss": 1.1997,
+      "step": 3599
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1228616366328122e-05,
+      "loss": 1.0941,
+      "step": 3600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1228089239445864e-05,
+      "loss": 1.1155,
+      "step": 3601
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1227561939748672e-05,
+      "loss": 1.0755,
+      "step": 3602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1227034467246428e-05,
+      "loss": 1.1848,
+      "step": 3603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1226506821949002e-05,
+      "loss": 1.0979,
+      "step": 3604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1225979003866277e-05,
+      "loss": 1.1406,
+      "step": 3605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1225451013008147e-05,
+      "loss": 1.1012,
+      "step": 3606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1224922849384488e-05,
+      "loss": 1.111,
+      "step": 3607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1224394513005196e-05,
+      "loss": 1.0573,
+      "step": 3608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1223866003880162e-05,
+      "loss": 1.1097,
+      "step": 3609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122333732201929e-05,
+      "loss": 1.1569,
+      "step": 3610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122280846743247e-05,
+      "loss": 1.14,
+      "step": 3611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1222279440129616e-05,
+      "loss": 1.1112,
+      "step": 3612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1221750240120634e-05,
+      "loss": 1.1904,
+      "step": 3613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1221220867415434e-05,
+      "loss": 1.186,
+      "step": 3614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122069132202392e-05,
+      "loss": 1.1145,
+      "step": 3615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.122016160395602e-05,
+      "loss": 1.0565,
+      "step": 3616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121963171322165e-05,
+      "loss": 1.138,
+      "step": 3617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1219101649830735e-05,
+      "loss": 1.1036,
+      "step": 3618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1218571413793204e-05,
+      "loss": 1.085,
+      "step": 3619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1218041005118977e-05,
+      "loss": 1.1241,
+      "step": 3620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1217510423817992e-05,
+      "loss": 1.095,
+      "step": 3621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121697966990019e-05,
+      "loss": 1.1706,
+      "step": 3622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1216448743375505e-05,
+      "loss": 1.0525,
+      "step": 3623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121591764425388e-05,
+      "loss": 1.1707,
+      "step": 3624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1215386372545265e-05,
+      "loss": 1.1519,
+      "step": 3625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1214854928259605e-05,
+      "loss": 1.1452,
+      "step": 3626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1214323311406853e-05,
+      "loss": 1.1729,
+      "step": 3627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1213791521996968e-05,
+      "loss": 1.1359,
+      "step": 3628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1213259560039903e-05,
+      "loss": 1.0992,
+      "step": 3629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1212727425545627e-05,
+      "loss": 1.1135,
+      "step": 3630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.12121951185241e-05,
+      "loss": 1.1105,
+      "step": 3631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.121166263898529e-05,
+      "loss": 1.063,
+      "step": 3632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1211129986939175e-05,
+      "loss": 1.0785,
+      "step": 3633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1210597162395723e-05,
+      "loss": 1.1168,
+      "step": 3634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1210064165364914e-05,
+      "loss": 1.0855,
+      "step": 3635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1209530995856737e-05,
+      "loss": 1.1159,
+      "step": 3636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1208997653881166e-05,
+      "loss": 1.1326,
+      "step": 3637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1208464139448193e-05,
+      "loss": 1.1346,
+      "step": 3638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120793045256781e-05,
+      "loss": 1.1809,
+      "step": 3639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120739659325001e-05,
+      "loss": 1.1364,
+      "step": 3640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1206862561504793e-05,
+      "loss": 1.0621,
+      "step": 3641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1206328357342157e-05,
+      "loss": 1.1513,
+      "step": 3642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1205793980772104e-05,
+      "loss": 1.14,
+      "step": 3643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120525943180465e-05,
+      "loss": 1.1059,
+      "step": 3644
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1204724710449802e-05,
+      "loss": 1.0847,
+      "step": 3645
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1204189816717566e-05,
+      "loss": 1.1309,
+      "step": 3646
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120365475061797e-05,
+      "loss": 1.1306,
+      "step": 3647
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1203119512161025e-05,
+      "loss": 1.1918,
+      "step": 3648
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120258410135676e-05,
+      "loss": 1.0927,
+      "step": 3649
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1202048518215204e-05,
+      "loss": 1.1591,
+      "step": 3650
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120151276274638e-05,
+      "loss": 1.0795,
+      "step": 3651
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1200976834960322e-05,
+      "loss": 1.0981,
+      "step": 3652
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.120044073486707e-05,
+      "loss": 1.0556,
+      "step": 3653
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1199904462476663e-05,
+      "loss": 1.1627,
+      "step": 3654
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1199368017799142e-05,
+      "loss": 1.1571,
+      "step": 3655
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1198831400844555e-05,
+      "loss": 1.1614,
+      "step": 3656
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1198294611622952e-05,
+      "loss": 1.1422,
+      "step": 3657
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1197757650144382e-05,
+      "loss": 1.108,
+      "step": 3658
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1197220516418904e-05,
+      "loss": 1.1258,
+      "step": 3659
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119668321045657e-05,
+      "loss": 1.1353,
+      "step": 3660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1196145732267452e-05,
+      "loss": 1.1475,
+      "step": 3661
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119560808186161e-05,
+      "loss": 1.1235,
+      "step": 3662
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1195070259249115e-05,
+      "loss": 1.1426,
+      "step": 3663
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1194532264440036e-05,
+      "loss": 1.1502,
+      "step": 3664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119399409744445e-05,
+      "loss": 1.0979,
+      "step": 3665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1193455758272432e-05,
+      "loss": 1.1863,
+      "step": 3666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.119291724693407e-05,
+      "loss": 1.154,
+      "step": 3667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.1192378563439442e-05,
+      "loss": 1.1312,
+      "step": 3668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1191839707798643e-05,
+      "loss": 1.1297,
+      "step": 3669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.119130068002176e-05,
+      "loss": 1.1396,
+      "step": 3670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.119076148011888e-05,
+      "loss": 1.135,
+      "step": 3671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1190222108100116e-05,
+      "loss": 1.1248,
+      "step": 3672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1189682563975555e-05,
+      "loss": 1.0707,
+      "step": 3673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118914284775531e-05,
+      "loss": 1.1296,
+      "step": 3674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1188602959449487e-05,
+      "loss": 1.1295,
+      "step": 3675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1188062899068194e-05,
+      "loss": 1.1994,
+      "step": 3676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1187522666621543e-05,
+      "loss": 1.0681,
+      "step": 3677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1186982262119658e-05,
+      "loss": 1.1159,
+      "step": 3678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118644168557265e-05,
+      "loss": 1.1337,
+      "step": 3679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118590093699065e-05,
+      "loss": 1.1316,
+      "step": 3680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118536001638378e-05,
+      "loss": 1.0892,
+      "step": 3681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1184818923762176e-05,
+      "loss": 1.15,
+      "step": 3682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1184277659135964e-05,
+      "loss": 1.0648,
+      "step": 3683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118373622251528e-05,
+      "loss": 1.0365,
+      "step": 3684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118319461391027e-05,
+      "loss": 1.1093,
+      "step": 3685
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1182652833331073e-05,
+      "loss": 1.1356,
+      "step": 3686
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1182110880787833e-05,
+      "loss": 1.0847,
+      "step": 3687
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11815687562907e-05,
+      "loss": 1.198,
+      "step": 3688
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1181026459849834e-05,
+      "loss": 1.1363,
+      "step": 3689
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.118048399147538e-05,
+      "loss": 1.1242,
+      "step": 3690
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1179941351177504e-05,
+      "loss": 1.1232,
+      "step": 3691
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117939853896636e-05,
+      "loss": 1.1169,
+      "step": 3692
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1178855554852125e-05,
+      "loss": 1.1503,
+      "step": 3693
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1178312398844958e-05,
+      "loss": 1.0888,
+      "step": 3694
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1177769070955033e-05,
+      "loss": 1.1041,
+      "step": 3695
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1177225571192524e-05,
+      "loss": 1.1152,
+      "step": 3696
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1176681899567613e-05,
+      "loss": 1.1502,
+      "step": 3697
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117613805609048e-05,
+      "loss": 1.1494,
+      "step": 3698
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117559404077131e-05,
+      "loss": 1.0677,
+      "step": 3699
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117504985362029e-05,
+      "loss": 1.1639,
+      "step": 3700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1174505494647606e-05,
+      "loss": 1.0933,
+      "step": 3701
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117396096386346e-05,
+      "loss": 1.1078,
+      "step": 3702
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1173416261278046e-05,
+      "loss": 1.115,
+      "step": 3703
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117287138690156e-05,
+      "loss": 1.0683,
+      "step": 3704
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1172326340744216e-05,
+      "loss": 1.0937,
+      "step": 3705
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1171781122816217e-05,
+      "loss": 1.1895,
+      "step": 3706
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.117123573312777e-05,
+      "loss": 1.1118,
+      "step": 3707
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1170690171689092e-05,
+      "loss": 1.1681,
+      "step": 3708
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1170144438510396e-05,
+      "loss": 1.1042,
+      "step": 3709
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116959853360191e-05,
+      "loss": 1.0992,
+      "step": 3710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1169052456973847e-05,
+      "loss": 1.1454,
+      "step": 3711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116850620863644e-05,
+      "loss": 1.1088,
+      "step": 3712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1167959788599914e-05,
+      "loss": 1.1486,
+      "step": 3713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.116741319687451e-05,
+      "loss": 1.0977,
+      "step": 3714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1166866433470455e-05,
+      "loss": 1.105,
+      "step": 3715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1166319498397994e-05,
+      "loss": 1.1165,
+      "step": 3716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1165772391667367e-05,
+      "loss": 1.0794,
+      "step": 3717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1165225113288817e-05,
+      "loss": 1.2291,
+      "step": 3718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11646776632726e-05,
+      "loss": 1.1208,
+      "step": 3719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1164130041628964e-05,
+      "loss": 1.1593,
+      "step": 3720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1163582248368162e-05,
+      "loss": 1.0589,
+      "step": 3721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1163034283500457e-05,
+      "loss": 1.1612,
+      "step": 3722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1162486147036105e-05,
+      "loss": 1.191,
+      "step": 3723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1161937838985378e-05,
+      "loss": 1.1523,
+      "step": 3724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1161389359358543e-05,
+      "loss": 1.1066,
+      "step": 3725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1160840708165868e-05,
+      "loss": 1.0728,
+      "step": 3726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1160291885417628e-05,
+      "loss": 1.1279,
+      "step": 3727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11597428911241e-05,
+      "loss": 1.1904,
+      "step": 3728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115919372529557e-05,
+      "loss": 1.1512,
+      "step": 3729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1158644387942318e-05,
+      "loss": 1.0645,
+      "step": 3730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1158094879074635e-05,
+      "loss": 1.1254,
+      "step": 3731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1157545198702805e-05,
+      "loss": 1.1693,
+      "step": 3732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115699534683713e-05,
+      "loss": 1.0484,
+      "step": 3733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1156445323487903e-05,
+      "loss": 1.1277,
+      "step": 3734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115589512866542e-05,
+      "loss": 1.1348,
+      "step": 3735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1155344762379998e-05,
+      "loss": 1.041,
+      "step": 3736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1154794224641926e-05,
+      "loss": 1.1797,
+      "step": 3737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1154243515461526e-05,
+      "loss": 1.1581,
+      "step": 3738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1153692634849112e-05,
+      "loss": 1.1602,
+      "step": 3739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1153141582814993e-05,
+      "loss": 1.1631,
+      "step": 3740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115259035936949e-05,
+      "loss": 1.1354,
+      "step": 3741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115203896452293e-05,
+      "loss": 1.1049,
+      "step": 3742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1151487398285636e-05,
+      "loss": 1.1019,
+      "step": 3743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115093566066794e-05,
+      "loss": 1.2152,
+      "step": 3744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.115038375168017e-05,
+      "loss": 1.1349,
+      "step": 3745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1149831671332666e-05,
+      "loss": 1.1468,
+      "step": 3746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1149279419635762e-05,
+      "loss": 1.1536,
+      "step": 3747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1148726996599807e-05,
+      "loss": 1.118,
+      "step": 3748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1148174402235138e-05,
+      "loss": 1.1039,
+      "step": 3749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1147621636552112e-05,
+      "loss": 1.1019,
+      "step": 3750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114706869956107e-05,
+      "loss": 1.1818,
+      "step": 3751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114651559127238e-05,
+      "loss": 1.2032,
+      "step": 3752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114596231169639e-05,
+      "loss": 1.0984,
+      "step": 3753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1145408860843468e-05,
+      "loss": 1.0957,
+      "step": 3754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114485523872397e-05,
+      "loss": 1.1848,
+      "step": 3755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114430144534827e-05,
+      "loss": 1.0659,
+      "step": 3756
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114374748072674e-05,
+      "loss": 1.1427,
+      "step": 3757
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114319334486975e-05,
+      "loss": 1.135,
+      "step": 3758
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114263903778768e-05,
+      "loss": 1.0915,
+      "step": 3759
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.114208455949091e-05,
+      "loss": 1.1455,
+      "step": 3760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1141529909989823e-05,
+      "loss": 1.1776,
+      "step": 3761
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1140975089294804e-05,
+      "loss": 1.1907,
+      "step": 3762
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1140420097416253e-05,
+      "loss": 1.1577,
+      "step": 3763
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113986493436455e-05,
+      "loss": 1.1321,
+      "step": 3764
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1139309600150098e-05,
+      "loss": 1.1972,
+      "step": 3765
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1138754094783298e-05,
+      "loss": 1.1242,
+      "step": 3766
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1138198418274547e-05,
+      "loss": 1.0797,
+      "step": 3767
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113764257063426e-05,
+      "loss": 1.087,
+      "step": 3768
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113708655187284e-05,
+      "loss": 1.1188,
+      "step": 3769
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11365303620007e-05,
+      "loss": 1.096,
+      "step": 3770
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1135974001028262e-05,
+      "loss": 1.1267,
+      "step": 3771
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1135417468965938e-05,
+      "loss": 1.1139,
+      "step": 3772
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113486076582415e-05,
+      "loss": 1.1478,
+      "step": 3773
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1134303891613328e-05,
+      "loss": 1.1347,
+      "step": 3774
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1133746846343896e-05,
+      "loss": 1.1886,
+      "step": 3775
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113318963002629e-05,
+      "loss": 1.1426,
+      "step": 3776
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1132632242670947e-05,
+      "loss": 1.0892,
+      "step": 3777
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1132074684288295e-05,
+      "loss": 1.1538,
+      "step": 3778
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1131516954888786e-05,
+      "loss": 1.1118,
+      "step": 3779
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.113095905448286e-05,
+      "loss": 1.1911,
+      "step": 3780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1130400983080964e-05,
+      "loss": 1.1,
+      "step": 3781
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112984274069355e-05,
+      "loss": 1.1579,
+      "step": 3782
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112928432733108e-05,
+      "loss": 1.1121,
+      "step": 3783
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1128725743003994e-05,
+      "loss": 1.1526,
+      "step": 3784
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1128166987722767e-05,
+      "loss": 1.1264,
+      "step": 3785
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127608061497857e-05,
+      "loss": 1.1456,
+      "step": 3786
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1127048964339737e-05,
+      "loss": 1.149,
+      "step": 3787
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112648969625887e-05,
+      "loss": 1.182,
+      "step": 3788
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1125930257265735e-05,
+      "loss": 1.1228,
+      "step": 3789
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11253706473708e-05,
+      "loss": 1.1282,
+      "step": 3790
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1124810866584554e-05,
+      "loss": 1.1136,
+      "step": 3791
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1124250914917478e-05,
+      "loss": 1.1412,
+      "step": 3792
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112369079238006e-05,
+      "loss": 1.1767,
+      "step": 3793
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112313049898278e-05,
+      "loss": 1.117,
+      "step": 3794
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.112257003473614e-05,
+      "loss": 1.1675,
+      "step": 3795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1122009399650633e-05,
+      "loss": 1.1763,
+      "step": 3796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1121448593736756e-05,
+      "loss": 1.2065,
+      "step": 3797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1120887617005017e-05,
+      "loss": 1.1305,
+      "step": 3798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1120326469465917e-05,
+      "loss": 1.1797,
+      "step": 3799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111976515112996e-05,
+      "loss": 1.1287,
+      "step": 3800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111920366200767e-05,
+      "loss": 1.1621,
+      "step": 3801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1118642002109555e-05,
+      "loss": 1.1921,
+      "step": 3802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1118080171446132e-05,
+      "loss": 1.0631,
+      "step": 3803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111751817002792e-05,
+      "loss": 1.1834,
+      "step": 3804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1116955997865454e-05,
+      "loss": 1.0745,
+      "step": 3805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111639365496925e-05,
+      "loss": 1.1288,
+      "step": 3806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.111583114134985e-05,
+      "loss": 1.1451,
+      "step": 3807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1115268457017782e-05,
+      "loss": 1.1357,
+      "step": 3808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1114705601983578e-05,
+      "loss": 1.2021,
+      "step": 3809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1114142576257788e-05,
+      "loss": 1.1805,
+      "step": 3810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1113579379850953e-05,
+      "loss": 1.0604,
+      "step": 3811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1113016012773622e-05,
+      "loss": 1.1409,
+      "step": 3812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1112452475036338e-05,
+      "loss": 1.1761,
+      "step": 3813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1111888766649664e-05,
+      "loss": 1.1274,
+      "step": 3814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1111324887624146e-05,
+      "loss": 1.1133,
+      "step": 3815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1110760837970354e-05,
+      "loss": 1.0976,
+      "step": 3816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1110196617698847e-05,
+      "loss": 1.094,
+      "step": 3817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1109632226820186e-05,
+      "loss": 1.1668,
+      "step": 3818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110906766534495e-05,
+      "loss": 1.0753,
+      "step": 3819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1108502933283702e-05,
+      "loss": 1.1919,
+      "step": 3820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1107938030647027e-05,
+      "loss": 1.1686,
+      "step": 3821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.11073729574455e-05,
+      "loss": 1.2131,
+      "step": 3822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1106807713689695e-05,
+      "loss": 1.1645,
+      "step": 3823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1106242299390212e-05,
+      "loss": 1.096,
+      "step": 3824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105676714557633e-05,
+      "loss": 1.115,
+      "step": 3825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1105110959202542e-05,
+      "loss": 1.1219,
+      "step": 3826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1104545033335548e-05,
+      "loss": 1.1252,
+      "step": 3827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110397893696724e-05,
+      "loss": 1.0963,
+      "step": 3828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.110341267010822e-05,
+      "loss": 1.1209,
+      "step": 3829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1102846232769096e-05,
+      "loss": 1.1806,
+      "step": 3830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1102279624960474e-05,
+      "loss": 1.1648,
+      "step": 3831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1101712846692963e-05,
+      "loss": 1.0816,
+      "step": 3832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1101145897977178e-05,
+      "loss": 1.1463,
+      "step": 3833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1100578778823738e-05,
+      "loss": 1.1517,
+      "step": 3834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1100011489243264e-05,
+      "loss": 1.1665,
+      "step": 3835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1099444029246378e-05,
+      "loss": 1.1176,
+      "step": 3836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1098876398843705e-05,
+      "loss": 1.0916,
+      "step": 3837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109830859804588e-05,
+      "loss": 1.2076,
+      "step": 3838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109774062686353e-05,
+      "loss": 1.063,
+      "step": 3839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10971724853073e-05,
+      "loss": 1.1697,
+      "step": 3840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109660417338782e-05,
+      "loss": 1.1053,
+      "step": 3841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1096035691115735e-05,
+      "loss": 1.167,
+      "step": 3842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1095467038501698e-05,
+      "loss": 1.0637,
+      "step": 3843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094898215556348e-05,
+      "loss": 1.162,
+      "step": 3844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1094329222290348e-05,
+      "loss": 1.2005,
+      "step": 3845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1093760058714347e-05,
+      "loss": 1.1421,
+      "step": 3846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1093190724839e-05,
+      "loss": 1.1523,
+      "step": 3847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1092621220674983e-05,
+      "loss": 1.0569,
+      "step": 3848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1092051546232944e-05,
+      "loss": 1.1537,
+      "step": 3849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1091481701523564e-05,
+      "loss": 1.1068,
+      "step": 3850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.109091168655751e-05,
+      "loss": 1.1596,
+      "step": 3851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1090341501345455e-05,
+      "loss": 1.1292,
+      "step": 3852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1089771145898078e-05,
+      "loss": 1.0814,
+      "step": 3853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108920062022606e-05,
+      "loss": 1.0969,
+      "step": 3854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108862992434009e-05,
+      "loss": 1.0723,
+      "step": 3855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1088059058250846e-05,
+      "loss": 1.1038,
+      "step": 3856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1087488021969027e-05,
+      "loss": 1.0854,
+      "step": 3857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1086916815505324e-05,
+      "loss": 1.068,
+      "step": 3858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1086345438870435e-05,
+      "loss": 1.0948,
+      "step": 3859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108577389207506e-05,
+      "loss": 1.1776,
+      "step": 3860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1085202175129894e-05,
+      "loss": 1.1089,
+      "step": 3861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1084630288045656e-05,
+      "loss": 1.0601,
+      "step": 3862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108405823083305e-05,
+      "loss": 1.0851,
+      "step": 3863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1083486003502794e-05,
+      "loss": 1.1047,
+      "step": 3864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1082913606065592e-05,
+      "loss": 1.1275,
+      "step": 3865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1082341038532178e-05,
+      "loss": 1.1291,
+      "step": 3866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1081768300913263e-05,
+      "loss": 1.0507,
+      "step": 3867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1081195393219583e-05,
+      "loss": 1.1445,
+      "step": 3868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1080622315461852e-05,
+      "loss": 1.0982,
+      "step": 3869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.108004906765082e-05,
+      "loss": 1.1915,
+      "step": 3870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107947564979721e-05,
+      "loss": 1.1087,
+      "step": 3871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1078902061911764e-05,
+      "loss": 1.1734,
+      "step": 3872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1078328304005225e-05,
+      "loss": 1.1348,
+      "step": 3873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1077754376088335e-05,
+      "loss": 1.1243,
+      "step": 3874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107718027817184e-05,
+      "loss": 1.1611,
+      "step": 3875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.10766060102665e-05,
+      "loss": 1.1278,
+      "step": 3876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107603157238306e-05,
+      "loss": 1.0908,
+      "step": 3877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1075456964532285e-05,
+      "loss": 1.2104,
+      "step": 3878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1074882186724928e-05,
+      "loss": 1.1341,
+      "step": 3879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107430723897176e-05,
+      "loss": 1.0844,
+      "step": 3880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1073732121283546e-05,
+      "loss": 1.1049,
+      "step": 3881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107315683367105e-05,
+      "loss": 1.1073,
+      "step": 3882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1072581376145055e-05,
+      "loss": 1.1552,
+      "step": 3883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1072005748716333e-05,
+      "loss": 1.1293,
+      "step": 3884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1071429951395665e-05,
+      "loss": 1.1616,
+      "step": 3885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107085398419383e-05,
+      "loss": 1.1068,
+      "step": 3886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.107027784712162e-05,
+      "loss": 1.0953,
+      "step": 3887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1069701540189823e-05,
+      "loss": 1.201,
+      "step": 3888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1069125063409228e-05,
+      "loss": 1.0571,
+      "step": 3889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1068548416790634e-05,
+      "loss": 1.1015,
+      "step": 3890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1067971600344836e-05,
+      "loss": 1.179,
+      "step": 3891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1067394614082642e-05,
+      "loss": 1.152,
+      "step": 3892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106681745801485e-05,
+      "loss": 1.0891,
+      "step": 3893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1066240132152276e-05,
+      "loss": 1.1286,
+      "step": 3894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106566263650573e-05,
+      "loss": 1.1293,
+      "step": 3895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106508497108602e-05,
+      "loss": 1.0346,
+      "step": 3896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1064507135903974e-05,
+      "loss": 1.15,
+      "step": 3897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1063929130970405e-05,
+      "loss": 1.1066,
+      "step": 3898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106335095629614e-05,
+      "loss": 1.1161,
+      "step": 3899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106277261189201e-05,
+      "loss": 1.1909,
+      "step": 3900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1062194097768842e-05,
+      "loss": 1.0956,
+      "step": 3901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106161541393747e-05,
+      "loss": 1.1305,
+      "step": 3902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.106103656040873e-05,
+      "loss": 1.1618,
+      "step": 3903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1060457537193464e-05,
+      "loss": 1.2311,
+      "step": 3904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.1059878344302516e-05,
+      "loss": 1.121,
+      "step": 3905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1059298981746732e-05,
+      "loss": 1.073,
+      "step": 3906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1058719449536964e-05,
+      "loss": 1.1821,
+      "step": 3907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1058139747684057e-05,
+      "loss": 1.2003,
+      "step": 3908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1057559876198874e-05,
+      "loss": 1.1373,
+      "step": 3909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1056979835092277e-05,
+      "loss": 1.1582,
+      "step": 3910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1056399624375122e-05,
+      "loss": 1.1254,
+      "step": 3911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1055819244058274e-05,
+      "loss": 1.1259,
+      "step": 3912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1055238694152606e-05,
+      "loss": 1.2399,
+      "step": 3913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105465797466899e-05,
+      "loss": 1.1472,
+      "step": 3914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10540770856183e-05,
+      "loss": 1.1118,
+      "step": 3915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1053496027011413e-05,
+      "loss": 1.1608,
+      "step": 3916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1052914798859213e-05,
+      "loss": 1.0767,
+      "step": 3917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1052333401172583e-05,
+      "loss": 1.1531,
+      "step": 3918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1051751833962412e-05,
+      "loss": 1.167,
+      "step": 3919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105117009723959e-05,
+      "loss": 1.115,
+      "step": 3920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.105058819101501e-05,
+      "loss": 1.1284,
+      "step": 3921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1050006115299575e-05,
+      "loss": 1.1564,
+      "step": 3922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1049423870104178e-05,
+      "loss": 1.1036,
+      "step": 3923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104884145543973e-05,
+      "loss": 1.1414,
+      "step": 3924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1048258871317126e-05,
+      "loss": 1.1064,
+      "step": 3925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104767611774729e-05,
+      "loss": 1.1406,
+      "step": 3926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1047093194741126e-05,
+      "loss": 1.1336,
+      "step": 3927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104651010230956e-05,
+      "loss": 1.1582,
+      "step": 3928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10459268404635e-05,
+      "loss": 1.1023,
+      "step": 3929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104534340921388e-05,
+      "loss": 1.091,
+      "step": 3930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1044759808571617e-05,
+      "loss": 1.1836,
+      "step": 3931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104417603854764e-05,
+      "loss": 1.1717,
+      "step": 3932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1043592099152892e-05,
+      "loss": 1.0977,
+      "step": 3933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1043007990398296e-05,
+      "loss": 1.0988,
+      "step": 3934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10424237122948e-05,
+      "loss": 1.2039,
+      "step": 3935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1041839264853337e-05,
+      "loss": 1.1064,
+      "step": 3936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104125464808486e-05,
+      "loss": 1.1354,
+      "step": 3937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104066986200031e-05,
+      "loss": 1.0864,
+      "step": 3938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.104008490661064e-05,
+      "loss": 1.1046,
+      "step": 3939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103949978192681e-05,
+      "loss": 1.0987,
+      "step": 3940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1038914487959778e-05,
+      "loss": 1.1618,
+      "step": 3941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1038329024720498e-05,
+      "loss": 1.1084,
+      "step": 3942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103774339221994e-05,
+      "loss": 1.1493,
+      "step": 3943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103715759046906e-05,
+      "loss": 1.1215,
+      "step": 3944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1036571619478842e-05,
+      "loss": 1.1406,
+      "step": 3945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1035985479260253e-05,
+      "loss": 1.098,
+      "step": 3946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1035399169824274e-05,
+      "loss": 1.116,
+      "step": 3947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1034812691181877e-05,
+      "loss": 1.0425,
+      "step": 3948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1034226043344052e-05,
+      "loss": 1.1515,
+      "step": 3949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103363922632178e-05,
+      "loss": 1.1812,
+      "step": 3950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1033052240126057e-05,
+      "loss": 1.1813,
+      "step": 3951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103246508476787e-05,
+      "loss": 1.103,
+      "step": 3952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1031877760258215e-05,
+      "loss": 1.0547,
+      "step": 3953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1031290266608098e-05,
+      "loss": 1.1472,
+      "step": 3954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1030702603828508e-05,
+      "loss": 1.2045,
+      "step": 3955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.103011477193046e-05,
+      "loss": 1.1588,
+      "step": 3956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102952677092496e-05,
+      "loss": 1.148,
+      "step": 3957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1028938600823023e-05,
+      "loss": 1.1331,
+      "step": 3958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1028350261635653e-05,
+      "loss": 1.132,
+      "step": 3959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102776175337388e-05,
+      "loss": 1.0525,
+      "step": 3960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102717307604872e-05,
+      "loss": 1.117,
+      "step": 3961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1026584229671194e-05,
+      "loss": 1.1346,
+      "step": 3962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1025995214252333e-05,
+      "loss": 1.1441,
+      "step": 3963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1025406029803166e-05,
+      "loss": 1.0953,
+      "step": 3964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102481667633473e-05,
+      "loss": 1.0964,
+      "step": 3965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102422715385806e-05,
+      "loss": 1.1013,
+      "step": 3966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1023637462384194e-05,
+      "loss": 1.1525,
+      "step": 3967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1023047601924178e-05,
+      "loss": 1.1062,
+      "step": 3968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1022457572489057e-05,
+      "loss": 1.1513,
+      "step": 3969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1021867374089878e-05,
+      "loss": 1.1522,
+      "step": 3970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1021277006737702e-05,
+      "loss": 1.0973,
+      "step": 3971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1020686470443575e-05,
+      "loss": 1.173,
+      "step": 3972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.102009576521856e-05,
+      "loss": 1.1785,
+      "step": 3973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101950489107372e-05,
+      "loss": 1.141,
+      "step": 3974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101891384802012e-05,
+      "loss": 1.1129,
+      "step": 3975
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1018322636068826e-05,
+      "loss": 1.0606,
+      "step": 3976
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1017731255230914e-05,
+      "loss": 1.1048,
+      "step": 3977
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1017139705517454e-05,
+      "loss": 1.0966,
+      "step": 3978
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1016547986939527e-05,
+      "loss": 1.1724,
+      "step": 3979
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1015956099508217e-05,
+      "loss": 1.1058,
+      "step": 3980
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.10153640432346e-05,
+      "loss": 1.0877,
+      "step": 3981
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1014771818129772e-05,
+      "loss": 1.1504,
+      "step": 3982
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1014179424204822e-05,
+      "loss": 1.1273,
+      "step": 3983
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101358686147084e-05,
+      "loss": 1.0833,
+      "step": 3984
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1012994129938924e-05,
+      "loss": 1.0792,
+      "step": 3985
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1012401229620175e-05,
+      "loss": 1.1328,
+      "step": 3986
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1011808160525696e-05,
+      "loss": 1.1118,
+      "step": 3987
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1011214922666594e-05,
+      "loss": 1.1305,
+      "step": 3988
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.101062151605398e-05,
+      "loss": 1.1786,
+      "step": 3989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1010027940698963e-05,
+      "loss": 1.1489,
+      "step": 3990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1009434196612663e-05,
+      "loss": 1.0757,
+      "step": 3991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1008840283806196e-05,
+      "loss": 1.0941,
+      "step": 3992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1008246202290684e-05,
+      "loss": 1.1705,
+      "step": 3993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1007651952077255e-05,
+      "loss": 1.0092,
+      "step": 3994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1007057533177035e-05,
+      "loss": 1.0466,
+      "step": 3995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1006462945601157e-05,
+      "loss": 1.1329,
+      "step": 3996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100586818936076e-05,
+      "loss": 1.1115,
+      "step": 3997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1005273264466973e-05,
+      "loss": 1.0882,
+      "step": 3998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1004678170930943e-05,
+      "loss": 1.0902,
+      "step": 3999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1004082908763816e-05,
+      "loss": 1.096,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "eval_text_loss": 1.002455472946167,
+      "eval_text_runtime": 1.5472,
+      "eval_text_samples_per_second": 47.829,
+      "eval_text_steps_per_second": 1.293,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "eval_svit_loss": 1.1932272911071777,
+      "eval_svit_runtime": 4.1755,
+      "eval_svit_samples_per_second": 163.093,
+      "eval_svit_steps_per_second": 2.634,
+      "step": 4000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1003487477976736e-05,
+      "loss": 1.0783,
+      "step": 4001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1002891878580853e-05,
+      "loss": 1.0969,
+      "step": 4002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1002296110587323e-05,
+      "loss": 1.094,
+      "step": 4003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1001700174007302e-05,
+      "loss": 1.1365,
+      "step": 4004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.100110406885195e-05,
+      "loss": 1.1088,
+      "step": 4005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.1000507795132432e-05,
+      "loss": 1.1153,
+      "step": 4006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0999911352859915e-05,
+      "loss": 1.0799,
+      "step": 4007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0999314742045564e-05,
+      "loss": 1.0808,
+      "step": 4008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0998717962700557e-05,
+      "loss": 1.1245,
+      "step": 4009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0998121014836066e-05,
+      "loss": 1.1054,
+      "step": 4010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099752389846327e-05,
+      "loss": 1.0873,
+      "step": 4011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0996926613593354e-05,
+      "loss": 1.0275,
+      "step": 4012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0996329160237502e-05,
+      "loss": 1.1075,
+      "step": 4013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0995731538406906e-05,
+      "loss": 1.101,
+      "step": 4014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099513374811275e-05,
+      "loss": 1.073,
+      "step": 4015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0994535789366235e-05,
+      "loss": 1.1169,
+      "step": 4016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099393766217856e-05,
+      "loss": 1.1486,
+      "step": 4017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099333936656092e-05,
+      "loss": 1.1298,
+      "step": 4018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0992740902524524e-05,
+      "loss": 1.0531,
+      "step": 4019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.099214227008058e-05,
+      "loss": 1.1148,
+      "step": 4020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0991543469240293e-05,
+      "loss": 1.0146,
+      "step": 4021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0990944500014885e-05,
+      "loss": 1.0928,
+      "step": 4022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0990345362415567e-05,
+      "loss": 1.1742,
+      "step": 4023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098974605645356e-05,
+      "loss": 1.1126,
+      "step": 4024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0989146582140088e-05,
+      "loss": 1.0991,
+      "step": 4025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0988546939486376e-05,
+      "loss": 1.0885,
+      "step": 4026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0987947128503658e-05,
+      "loss": 1.1284,
+      "step": 4027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0987347149203163e-05,
+      "loss": 1.0823,
+      "step": 4028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0986747001596124e-05,
+      "loss": 1.0708,
+      "step": 4029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0986146685693786e-05,
+      "loss": 1.152,
+      "step": 4030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098554620150739e-05,
+      "loss": 1.0963,
+      "step": 4031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0984945549048175e-05,
+      "loss": 1.1393,
+      "step": 4032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0984344728327397e-05,
+      "loss": 1.1239,
+      "step": 4033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0983743739356308e-05,
+      "loss": 1.1462,
+      "step": 4034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0983142582146155e-05,
+      "loss": 1.0299,
+      "step": 4035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0982541256708205e-05,
+      "loss": 1.1322,
+      "step": 4036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098193976305371e-05,
+      "loss": 1.162,
+      "step": 4037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098133810119394e-05,
+      "loss": 1.094,
+      "step": 4038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0980736271140165e-05,
+      "loss": 1.0896,
+      "step": 4039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.098013427290365e-05,
+      "loss": 1.2045,
+      "step": 4040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0979532106495666e-05,
+      "loss": 1.1384,
+      "step": 4041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0978929771927497e-05,
+      "loss": 1.1988,
+      "step": 4042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0978327269210424e-05,
+      "loss": 1.1533,
+      "step": 4043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097772459835572e-05,
+      "loss": 1.1637,
+      "step": 4044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097712175937468e-05,
+      "loss": 1.1337,
+      "step": 4045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097651875227859e-05,
+      "loss": 1.1349,
+      "step": 4046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0975915577078747e-05,
+      "loss": 1.1182,
+      "step": 4047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0975312233786438e-05,
+      "loss": 1.1095,
+      "step": 4048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097470872241297e-05,
+      "loss": 1.1124,
+      "step": 4049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097410504296964e-05,
+      "loss": 1.1264,
+      "step": 4050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0973501195467758e-05,
+      "loss": 1.1635,
+      "step": 4051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0972897179918626e-05,
+      "loss": 1.0804,
+      "step": 4052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.097229299633356e-05,
+      "loss": 1.0651,
+      "step": 4053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0971688644723876e-05,
+      "loss": 1.0775,
+      "step": 4054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0971084125100884e-05,
+      "loss": 1.179,
+      "step": 4055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0970479437475913e-05,
+      "loss": 1.1266,
+      "step": 4056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0969874581860284e-05,
+      "loss": 1.0922,
+      "step": 4057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0969269558265325e-05,
+      "loss": 1.0919,
+      "step": 4058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0968664366702365e-05,
+      "loss": 1.1951,
+      "step": 4059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0968059007182736e-05,
+      "loss": 1.1491,
+      "step": 4060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0967453479717777e-05,
+      "loss": 1.1591,
+      "step": 4061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0966847784318827e-05,
+      "loss": 1.0675,
+      "step": 4062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0966241920997228e-05,
+      "loss": 1.1485,
+      "step": 4063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0965635889764333e-05,
+      "loss": 1.1299,
+      "step": 4064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096502969063148e-05,
+      "loss": 1.2198,
+      "step": 4065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0964423323610027e-05,
+      "loss": 1.1448,
+      "step": 4066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0963816788711332e-05,
+      "loss": 1.1286,
+      "step": 4067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0963210085946746e-05,
+      "loss": 1.0397,
+      "step": 4068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096260321532764e-05,
+      "loss": 1.0648,
+      "step": 4069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.096199617686537e-05,
+      "loss": 1.074,
+      "step": 4070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0961388970571313e-05,
+      "loss": 1.2179,
+      "step": 4071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0960781596456833e-05,
+      "loss": 1.095,
+      "step": 4072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0960174054533308e-05,
+      "loss": 1.0766,
+      "step": 4073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0959566344812112e-05,
+      "loss": 1.1364,
+      "step": 4074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0958958467304633e-05,
+      "loss": 1.121,
+      "step": 4075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0958350422022246e-05,
+      "loss": 1.1616,
+      "step": 4076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0957742208976346e-05,
+      "loss": 1.136,
+      "step": 4077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0957133828178312e-05,
+      "loss": 1.0884,
+      "step": 4078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095652527963955e-05,
+      "loss": 1.1626,
+      "step": 4079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0955916563371446e-05,
+      "loss": 1.1829,
+      "step": 4080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0955307679385406e-05,
+      "loss": 1.1124,
+      "step": 4081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095469862769283e-05,
+      "loss": 1.0239,
+      "step": 4082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095408940830512e-05,
+      "loss": 1.1646,
+      "step": 4083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0953480021233697e-05,
+      "loss": 1.1342,
+      "step": 4084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.095287046648996e-05,
+      "loss": 1.0848,
+      "step": 4085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0952260744085332e-05,
+      "loss": 1.1014,
+      "step": 4086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0951650854031223e-05,
+      "loss": 1.1331,
+      "step": 4087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0951040796339065e-05,
+      "loss": 1.1171,
+      "step": 4088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0950430571020272e-05,
+      "loss": 1.1457,
+      "step": 4089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0949820178086284e-05,
+      "loss": 1.1965,
+      "step": 4090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094920961754852e-05,
+      "loss": 1.1349,
+      "step": 4091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094859888941842e-05,
+      "loss": 1.0873,
+      "step": 4092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094798799370742e-05,
+      "loss": 1.0704,
+      "step": 4093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094737693042696e-05,
+      "loss": 1.0713,
+      "step": 4094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0946765699588487e-05,
+      "loss": 1.0887,
+      "step": 4095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0946154301203446e-05,
+      "loss": 1.1232,
+      "step": 4096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094554273528328e-05,
+      "loss": 1.143,
+      "step": 4097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.094493100183945e-05,
+      "loss": 1.1777,
+      "step": 4098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0944319100883403e-05,
+      "loss": 1.0781,
+      "step": 4099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0943707032426613e-05,
+      "loss": 1.1704,
+      "step": 4100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0943094796480523e-05,
+      "loss": 1.1318,
+      "step": 4101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0942482393056615e-05,
+      "loss": 1.115,
+      "step": 4102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0941869822166352e-05,
+      "loss": 1.1751,
+      "step": 4103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0941257083821203e-05,
+      "loss": 1.1583,
+      "step": 4104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0940644178032645e-05,
+      "loss": 1.0432,
+      "step": 4105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0940031104812156e-05,
+      "loss": 1.1013,
+      "step": 4106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0939417864171217e-05,
+      "loss": 1.0841,
+      "step": 4107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093880445612131e-05,
+      "loss": 1.0485,
+      "step": 4108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0938190880673926e-05,
+      "loss": 1.1636,
+      "step": 4109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0937577137840553e-05,
+      "loss": 1.1216,
+      "step": 4110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0936963227632683e-05,
+      "loss": 1.0825,
+      "step": 4111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0936349150061816e-05,
+      "loss": 1.1881,
+      "step": 4112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093573490513945e-05,
+      "loss": 1.1221,
+      "step": 4113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0935120492877088e-05,
+      "loss": 1.1268,
+      "step": 4114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0934505913286243e-05,
+      "loss": 1.1251,
+      "step": 4115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093389116637841e-05,
+      "loss": 1.1488,
+      "step": 4116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0933276252165116e-05,
+      "loss": 1.1497,
+      "step": 4117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932661170657864e-05,
+      "loss": 1.1534,
+      "step": 4118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0932045921868185e-05,
+      "loss": 1.1421,
+      "step": 4119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0931430505807588e-05,
+      "loss": 1.1006,
+      "step": 4120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.093081492248761e-05,
+      "loss": 1.0754,
+      "step": 4121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0930199171919772e-05,
+      "loss": 1.2207,
+      "step": 4122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0929583254115604e-05,
+      "loss": 1.1143,
+      "step": 4123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0928967169086647e-05,
+      "loss": 1.1402,
+      "step": 4124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092835091684443e-05,
+      "loss": 1.1143,
+      "step": 4125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09277344974005e-05,
+      "loss": 1.1095,
+      "step": 4126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09271179107664e-05,
+      "loss": 1.1549,
+      "step": 4127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0926501156953673e-05,
+      "loss": 1.0912,
+      "step": 4128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0925884235973877e-05,
+      "loss": 1.1765,
+      "step": 4129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0925267147838553e-05,
+      "loss": 1.1293,
+      "step": 4130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092464989255927e-05,
+      "loss": 1.1089,
+      "step": 4131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0924032470147578e-05,
+      "loss": 1.0954,
+      "step": 4132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0923414880615044e-05,
+      "loss": 1.1227,
+      "step": 4133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092279712397323e-05,
+      "loss": 1.1706,
+      "step": 4134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092217920023371e-05,
+      "loss": 1.2328,
+      "step": 4135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092156110940805e-05,
+      "loss": 1.0825,
+      "step": 4136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0920942851507836e-05,
+      "loss": 1.1352,
+      "step": 4137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.092032442654463e-05,
+      "loss": 1.14,
+      "step": 4138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0919705834530022e-05,
+      "loss": 1.1291,
+      "step": 4139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.09190870754756e-05,
+      "loss": 1.1591,
+      "step": 4140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.0918468149392948e-05,
+      "loss": 1.1223,
+      "step": 4141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0917849056293653e-05,
+      "loss": 1.1649,
+      "step": 4142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0917229796189313e-05,
+      "loss": 1.1005,
+      "step": 4143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0916610369091523e-05,
+      "loss": 1.1207,
+      "step": 4144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0915990775011884e-05,
+      "loss": 1.0847,
+      "step": 4145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0915371013962e-05,
+      "loss": 1.184,
+      "step": 4146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0914751085953476e-05,
+      "loss": 1.1456,
+      "step": 4147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0914130990997922e-05,
+      "loss": 1.1386,
+      "step": 4148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0913510729106953e-05,
+      "loss": 1.0934,
+      "step": 4149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0912890300292177e-05,
+      "loss": 1.0988,
+      "step": 4150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0912269704565222e-05,
+      "loss": 1.1183,
+      "step": 4151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0911648941937706e-05,
+      "loss": 1.1623,
+      "step": 4152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091102801242125e-05,
+      "loss": 1.1463,
+      "step": 4153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.091040691602749e-05,
+      "loss": 1.1159,
+      "step": 4154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090978565276805e-05,
+      "loss": 1.1542,
+      "step": 4155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0909164222654568e-05,
+      "loss": 1.1044,
+      "step": 4156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0908542625698684e-05,
+      "loss": 1.0766,
+      "step": 4157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0907920861912035e-05,
+      "loss": 1.1557,
+      "step": 4158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0907298931306264e-05,
+      "loss": 1.1947,
+      "step": 4159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0906676833893025e-05,
+      "loss": 1.0803,
+      "step": 4160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0906054569683956e-05,
+      "loss": 1.1133,
+      "step": 4161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090543213869072e-05,
+      "loss": 1.1687,
+      "step": 4162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090480954092497e-05,
+      "loss": 1.143,
+      "step": 4163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0904186776398366e-05,
+      "loss": 1.1387,
+      "step": 4164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0903563845122573e-05,
+      "loss": 1.0811,
+      "step": 4165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090294074710925e-05,
+      "loss": 1.0918,
+      "step": 4166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.090231748237007e-05,
+      "loss": 1.1059,
+      "step": 4167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0901694050916706e-05,
+      "loss": 1.1736,
+      "step": 4168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0901070452760832e-05,
+      "loss": 1.1083,
+      "step": 4169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0900446687914125e-05,
+      "loss": 1.1358,
+      "step": 4170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0899822756388265e-05,
+      "loss": 1.2287,
+      "step": 4171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089919865819494e-05,
+      "loss": 1.1254,
+      "step": 4172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0898574393345837e-05,
+      "loss": 1.1141,
+      "step": 4173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0897949961852647e-05,
+      "loss": 1.1276,
+      "step": 4174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089732536372706e-05,
+      "loss": 1.078,
+      "step": 4175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0896700598980778e-05,
+      "loss": 1.1979,
+      "step": 4176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0896075667625494e-05,
+      "loss": 1.1644,
+      "step": 4177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0895450569672917e-05,
+      "loss": 1.1267,
+      "step": 4178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0894825305134753e-05,
+      "loss": 1.0595,
+      "step": 4179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0894199874022708e-05,
+      "loss": 1.1496,
+      "step": 4180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08935742763485e-05,
+      "loss": 1.0764,
+      "step": 4181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0892948512123834e-05,
+      "loss": 1.0667,
+      "step": 4182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0892322581360443e-05,
+      "loss": 1.066,
+      "step": 4183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089169648407004e-05,
+      "loss": 1.1736,
+      "step": 4184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.089107022026435e-05,
+      "loss": 1.1881,
+      "step": 4185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0890443789955104e-05,
+      "loss": 1.1335,
+      "step": 4186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0889817193154033e-05,
+      "loss": 1.1254,
+      "step": 4187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0889190429872865e-05,
+      "loss": 1.0906,
+      "step": 4188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088856350012335e-05,
+      "loss": 1.1133,
+      "step": 4189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0887936403917218e-05,
+      "loss": 1.117,
+      "step": 4190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0887309141266214e-05,
+      "loss": 1.1155,
+      "step": 4191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088668171218209e-05,
+      "loss": 1.0666,
+      "step": 4192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0886054116676587e-05,
+      "loss": 1.0943,
+      "step": 4193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0885426354761467e-05,
+      "loss": 1.1135,
+      "step": 4194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0884798426448483e-05,
+      "loss": 1.131,
+      "step": 4195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0884170331749395e-05,
+      "loss": 1.1472,
+      "step": 4196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0883542070675964e-05,
+      "loss": 1.1098,
+      "step": 4197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0882913643239953e-05,
+      "loss": 1.0307,
+      "step": 4198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0882285049453135e-05,
+      "loss": 1.128,
+      "step": 4199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.088165628932728e-05,
+      "loss": 1.1055,
+      "step": 4200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0881027362874166e-05,
+      "loss": 1.1434,
+      "step": 4201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0880398270105565e-05,
+      "loss": 1.1165,
+      "step": 4202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087976901103326e-05,
+      "loss": 1.0071,
+      "step": 4203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0879139585669037e-05,
+      "loss": 1.0952,
+      "step": 4204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087850999402468e-05,
+      "loss": 1.1144,
+      "step": 4205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0877880236111987e-05,
+      "loss": 1.1598,
+      "step": 4206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0877250311942743e-05,
+      "loss": 1.1218,
+      "step": 4207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0876620221528746e-05,
+      "loss": 1.1002,
+      "step": 4208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0875989964881798e-05,
+      "loss": 1.1135,
+      "step": 4209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0875359542013708e-05,
+      "loss": 1.1198,
+      "step": 4210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087472895293627e-05,
+      "loss": 1.1283,
+      "step": 4211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08740981976613e-05,
+      "loss": 1.1051,
+      "step": 4212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0873467276200604e-05,
+      "loss": 1.1338,
+      "step": 4213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0872836188566007e-05,
+      "loss": 1.1392,
+      "step": 4214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0872204934769322e-05,
+      "loss": 1.1935,
+      "step": 4215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087157351482237e-05,
+      "loss": 1.1894,
+      "step": 4216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0870941928736975e-05,
+      "loss": 1.1212,
+      "step": 4217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.087031017652497e-05,
+      "loss": 1.1325,
+      "step": 4218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086967825819818e-05,
+      "loss": 1.1668,
+      "step": 4219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0869046173768442e-05,
+      "loss": 1.1713,
+      "step": 4220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086841392324759e-05,
+      "loss": 1.0496,
+      "step": 4221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0867781506647467e-05,
+      "loss": 1.1258,
+      "step": 4222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0867148923979917e-05,
+      "loss": 1.1748,
+      "step": 4223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0866516175256786e-05,
+      "loss": 1.1335,
+      "step": 4224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0865883260489922e-05,
+      "loss": 1.1024,
+      "step": 4225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0865250179691177e-05,
+      "loss": 1.0905,
+      "step": 4226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086461693287241e-05,
+      "loss": 1.0914,
+      "step": 4227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086398352004548e-05,
+      "loss": 1.1232,
+      "step": 4228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086334994122224e-05,
+      "loss": 1.1774,
+      "step": 4229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0862716196414563e-05,
+      "loss": 1.065,
+      "step": 4230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.086208228563432e-05,
+      "loss": 1.0754,
+      "step": 4231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0861448208893376e-05,
+      "loss": 1.145,
+      "step": 4232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0860813966203613e-05,
+      "loss": 1.0585,
+      "step": 4233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0860179557576896e-05,
+      "loss": 1.1491,
+      "step": 4234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0859544983025113e-05,
+      "loss": 1.1313,
+      "step": 4235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085891024256015e-05,
+      "loss": 1.1426,
+      "step": 4236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0858275336193894e-05,
+      "loss": 1.1656,
+      "step": 4237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085764026393823e-05,
+      "loss": 1.1285,
+      "step": 4238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0857005025805053e-05,
+      "loss": 1.0768,
+      "step": 4239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085636962180626e-05,
+      "loss": 1.1527,
+      "step": 4240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085573405195375e-05,
+      "loss": 1.0994,
+      "step": 4241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0855098316259423e-05,
+      "loss": 1.0683,
+      "step": 4242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0854462414735186e-05,
+      "loss": 1.0861,
+      "step": 4243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085382634739295e-05,
+      "loss": 1.1475,
+      "step": 4244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0853190114244624e-05,
+      "loss": 1.1245,
+      "step": 4245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0852553715302125e-05,
+      "loss": 1.0959,
+      "step": 4246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.085191715057737e-05,
+      "loss": 1.0606,
+      "step": 4247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0851280420082277e-05,
+      "loss": 1.1364,
+      "step": 4248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0850643523828774e-05,
+      "loss": 1.1101,
+      "step": 4249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0850006461828785e-05,
+      "loss": 1.1893,
+      "step": 4250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0849369234094244e-05,
+      "loss": 1.1561,
+      "step": 4251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084873184063708e-05,
+      "loss": 1.1204,
+      "step": 4252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0848094281469233e-05,
+      "loss": 1.0985,
+      "step": 4253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0847456556602642e-05,
+      "loss": 1.0602,
+      "step": 4254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084681866604925e-05,
+      "loss": 1.1208,
+      "step": 4255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0846180609821e-05,
+      "loss": 1.1303,
+      "step": 4256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0845542387929845e-05,
+      "loss": 1.1417,
+      "step": 4257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084490400038773e-05,
+      "loss": 1.0888,
+      "step": 4258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084426544720662e-05,
+      "loss": 1.1198,
+      "step": 4259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0843626728398467e-05,
+      "loss": 1.161,
+      "step": 4260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0842987843975234e-05,
+      "loss": 1.1329,
+      "step": 4261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0842348793948884e-05,
+      "loss": 1.0318,
+      "step": 4262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0841709578331384e-05,
+      "loss": 1.161,
+      "step": 4263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084107019713471e-05,
+      "loss": 1.1886,
+      "step": 4264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.084043065037083e-05,
+      "loss": 1.1078,
+      "step": 4265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0839790938051722e-05,
+      "loss": 1.1404,
+      "step": 4266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0839151060189366e-05,
+      "loss": 1.1478,
+      "step": 4267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083851101679575e-05,
+      "loss": 1.0771,
+      "step": 4268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083787080788285e-05,
+      "loss": 1.0816,
+      "step": 4269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0837230433462663e-05,
+      "loss": 1.0949,
+      "step": 4270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0836589893547182e-05,
+      "loss": 1.0702,
+      "step": 4271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0835949188148395e-05,
+      "loss": 1.1381,
+      "step": 4272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083530831727831e-05,
+      "loss": 1.0765,
+      "step": 4273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083466728094892e-05,
+      "loss": 1.1073,
+      "step": 4274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0834026079172236e-05,
+      "loss": 1.0915,
+      "step": 4275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0833384711960265e-05,
+      "loss": 1.1048,
+      "step": 4276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0832743179325012e-05,
+      "loss": 1.1482,
+      "step": 4277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08321014812785e-05,
+      "loss": 1.2053,
+      "step": 4278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0831459617832737e-05,
+      "loss": 1.0996,
+      "step": 4279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083081758899975e-05,
+      "loss": 1.0985,
+      "step": 4280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.083017539479156e-05,
+      "loss": 1.1039,
+      "step": 4281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0829533035220194e-05,
+      "loss": 1.1717,
+      "step": 4282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082889051029768e-05,
+      "loss": 1.1287,
+      "step": 4283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0828247820036052e-05,
+      "loss": 1.1903,
+      "step": 4284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0827604964447344e-05,
+      "loss": 1.1477,
+      "step": 4285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0826961943543596e-05,
+      "loss": 1.0826,
+      "step": 4286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0826318757336854e-05,
+      "loss": 1.1342,
+      "step": 4287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0825675405839155e-05,
+      "loss": 1.1539,
+      "step": 4288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0825031889062552e-05,
+      "loss": 1.1302,
+      "step": 4289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0824388207019095e-05,
+      "loss": 1.1334,
+      "step": 4290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0823744359720834e-05,
+      "loss": 1.1338,
+      "step": 4291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.082310034717984e-05,
+      "loss": 1.1329,
+      "step": 4292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0822456169408157e-05,
+      "loss": 1.1287,
+      "step": 4293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0821811826417857e-05,
+      "loss": 1.1056,
+      "step": 4294
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0821167318221008e-05,
+      "loss": 1.1482,
+      "step": 4295
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0820522644829678e-05,
+      "loss": 1.1464,
+      "step": 4296
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0819877806255935e-05,
+      "loss": 1.154,
+      "step": 4297
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0819232802511862e-05,
+      "loss": 1.1466,
+      "step": 4298
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081858763360953e-05,
+      "loss": 1.0978,
+      "step": 4299
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081794229956103e-05,
+      "loss": 1.1503,
+      "step": 4300
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0817296800378444e-05,
+      "loss": 1.0999,
+      "step": 4301
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081665113607386e-05,
+      "loss": 1.1392,
+      "step": 4302
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0816005306659367e-05,
+      "loss": 1.1401,
+      "step": 4303
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081535931214706e-05,
+      "loss": 1.1234,
+      "step": 4304
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081471315254904e-05,
+      "loss": 1.049,
+      "step": 4305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0814066827877404e-05,
+      "loss": 1.0697,
+      "step": 4306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0813420338144258e-05,
+      "loss": 1.1793,
+      "step": 4307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.081277368336171e-05,
+      "loss": 1.1071,
+      "step": 4308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0812126863541868e-05,
+      "loss": 1.1421,
+      "step": 4309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0811479878696846e-05,
+      "loss": 1.16,
+      "step": 4310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0810832728838757e-05,
+      "loss": 1.0883,
+      "step": 4311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0810185413979725e-05,
+      "loss": 1.1226,
+      "step": 4312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0809537934131867e-05,
+      "loss": 1.1195,
+      "step": 4313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080889028930731e-05,
+      "loss": 1.1619,
+      "step": 4314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080824247951819e-05,
+      "loss": 1.0919,
+      "step": 4315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0807594504776628e-05,
+      "loss": 1.0759,
+      "step": 4316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0806946365094763e-05,
+      "loss": 1.1883,
+      "step": 4317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0806298060484734e-05,
+      "loss": 1.1075,
+      "step": 4318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080564959095868e-05,
+      "loss": 1.158,
+      "step": 4319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0805000956528745e-05,
+      "loss": 1.1285,
+      "step": 4320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0804352157207078e-05,
+      "loss": 1.1262,
+      "step": 4321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080370319300583e-05,
+      "loss": 1.0884,
+      "step": 4322
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0803054063937145e-05,
+      "loss": 1.1153,
+      "step": 4323
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0802404770013187e-05,
+      "loss": 1.1366,
+      "step": 4324
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.080175531124612e-05,
+      "loss": 1.222,
+      "step": 4325
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.08011056876481e-05,
+      "loss": 1.1757,
+      "step": 4326
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0800455899231287e-05,
+      "loss": 1.1342,
+      "step": 4327
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0799805946007864e-05,
+      "loss": 1.1518,
+      "step": 4328
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0799155827989992e-05,
+      "loss": 1.1654,
+      "step": 4329
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0798505545189847e-05,
+      "loss": 1.0833,
+      "step": 4330
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079785509761961e-05,
+      "loss": 1.1248,
+      "step": 4331
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0797204485291456e-05,
+      "loss": 1.1328,
+      "step": 4332
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0796553708217578e-05,
+      "loss": 1.1375,
+      "step": 4333
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0795902766410157e-05,
+      "loss": 1.0959,
+      "step": 4334
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0795251659881387e-05,
+      "loss": 1.2204,
+      "step": 4335
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0794600388643455e-05,
+      "loss": 1.1426,
+      "step": 4336
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0793948952708563e-05,
+      "loss": 1.0789,
+      "step": 4337
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079329735208891e-05,
+      "loss": 1.1051,
+      "step": 4338
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.07926455867967e-05,
+      "loss": 1.1411,
+      "step": 4339
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079199365684413e-05,
+      "loss": 1.1045,
+      "step": 4340
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.079134156224342e-05,
+      "loss": 1.1851,
+      "step": 4341
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0790689303006776e-05,
+      "loss": 1.1536,
+      "step": 4342
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0790036879146414e-05,
+      "loss": 1.1414,
+      "step": 4343
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078938429067455e-05,
+      "loss": 1.103,
+      "step": 4344
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0788731537603407e-05,
+      "loss": 1.1336,
+      "step": 4345
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0788078619945208e-05,
+      "loss": 1.0436,
+      "step": 4346
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078742553771218e-05,
+      "loss": 1.1409,
+      "step": 4347
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0786772290916558e-05,
+      "loss": 1.1086,
+      "step": 4348
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078611887957057e-05,
+      "loss": 1.1422,
+      "step": 4349
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0785465303686452e-05,
+      "loss": 1.1485,
+      "step": 4350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078481156327645e-05,
+      "loss": 1.1212,
+      "step": 4351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.07841576583528e-05,
+      "loss": 1.1004,
+      "step": 4352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078350358892775e-05,
+      "loss": 1.0938,
+      "step": 4353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078284935501355e-05,
+      "loss": 1.1113,
+      "step": 4354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078219495662245e-05,
+      "loss": 1.187,
+      "step": 4355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0781540393766705e-05,
+      "loss": 1.1355,
+      "step": 4356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0780885666458576e-05,
+      "loss": 1.1646,
+      "step": 4357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.078023077471032e-05,
+      "loss": 1.0867,
+      "step": 4358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0779575718534202e-05,
+      "loss": 1.0556,
+      "step": 4359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077892049794249e-05,
+      "loss": 1.1287,
+      "step": 4360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0778265112947457e-05,
+      "loss": 1.0377,
+      "step": 4361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0777609563561374e-05,
+      "loss": 1.1282,
+      "step": 4362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0776953849796517e-05,
+      "loss": 1.1132,
+      "step": 4363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077629797166517e-05,
+      "loss": 1.0723,
+      "step": 4364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0775641929179605e-05,
+      "loss": 1.0407,
+      "step": 4365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077498572235212e-05,
+      "loss": 1.1397,
+      "step": 4366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0774329351194993e-05,
+      "loss": 1.1238,
+      "step": 4367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0773672815720525e-05,
+      "loss": 1.0891,
+      "step": 4368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0773016115941005e-05,
+      "loss": 1.136,
+      "step": 4369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077235925186874e-05,
+      "loss": 1.1343,
+      "step": 4370
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0771702223516018e-05,
+      "loss": 1.0933,
+      "step": 4371
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.077104503089515e-05,
+      "loss": 1.1168,
+      "step": 4372
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0770387674018443e-05,
+      "loss": 1.0345,
+      "step": 4373
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.076973015289821e-05,
+      "loss": 1.1979,
+      "step": 4374
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.076907246754676e-05,
+      "loss": 1.1465,
+      "step": 4375
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.076841461797641e-05,
+      "loss": 1.1138,
+      "step": 4376
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0767756604199485e-05,
+      "loss": 1.11,
+      "step": 4377
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0767098426228298e-05,
+      "loss": 1.0776,
+      "step": 4378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0766440084075184e-05,
+      "loss": 1.1296,
+      "step": 4379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0765781577752468e-05,
+      "loss": 1.0834,
+      "step": 4380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0765122907272483e-05,
+      "loss": 1.0604,
+      "step": 4381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0764464072647562e-05,
+      "loss": 1.0774,
+      "step": 4382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076380507389004e-05,
+      "loss": 1.1501,
+      "step": 4383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0763145911012267e-05,
+      "loss": 1.0751,
+      "step": 4384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076248658402658e-05,
+      "loss": 1.1295,
+      "step": 4385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076182709294533e-05,
+      "loss": 1.1813,
+      "step": 4386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0761167437780864e-05,
+      "loss": 1.0408,
+      "step": 4387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0760507618545537e-05,
+      "loss": 1.1375,
+      "step": 4388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0759847635251708e-05,
+      "loss": 1.1416,
+      "step": 4389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0759187487911734e-05,
+      "loss": 1.1078,
+      "step": 4390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0758527176537975e-05,
+      "loss": 1.1919,
+      "step": 4391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.07578667011428e-05,
+      "loss": 1.1444,
+      "step": 4392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.075720606173858e-05,
+      "loss": 1.1309,
+      "step": 4393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0756545258337683e-05,
+      "loss": 1.1531,
+      "step": 4394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0755884290952482e-05,
+      "loss": 1.1036,
+      "step": 4395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0755223159595363e-05,
+      "loss": 1.1136,
+      "step": 4396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0754561864278697e-05,
+      "loss": 1.2148,
+      "step": 4397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0753900405014877e-05,
+      "loss": 1.0791,
+      "step": 4398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0753238781816283e-05,
+      "loss": 1.0656,
+      "step": 4399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0752576994695313e-05,
+      "loss": 1.1213,
+      "step": 4400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0751915043664353e-05,
+      "loss": 1.0718,
+      "step": 4401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.07512529287358e-05,
+      "loss": 1.18,
+      "step": 4402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0750590649922062e-05,
+      "loss": 1.1568,
+      "step": 4403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074992820723553e-05,
+      "loss": 1.1452,
+      "step": 4404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0749265600688616e-05,
+      "loss": 1.092,
+      "step": 4405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074860283029373e-05,
+      "loss": 1.1305,
+      "step": 4406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074793989606328e-05,
+      "loss": 1.0916,
+      "step": 4407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074727679800968e-05,
+      "loss": 1.1529,
+      "step": 4408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074661353614535e-05,
+      "loss": 1.0235,
+      "step": 4409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0745950110482716e-05,
+      "loss": 1.163,
+      "step": 4410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0745286521034192e-05,
+      "loss": 1.1012,
+      "step": 4411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074462276781221e-05,
+      "loss": 1.114,
+      "step": 4412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0743958850829205e-05,
+      "loss": 1.0956,
+      "step": 4413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0743294770097597e-05,
+      "loss": 1.0599,
+      "step": 4414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074263052562984e-05,
+      "loss": 1.1212,
+      "step": 4415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074196611743836e-05,
+      "loss": 1.1186,
+      "step": 4416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0741301545535602e-05,
+      "loss": 1.1027,
+      "step": 4417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.074063680993401e-05,
+      "loss": 1.1011,
+      "step": 4418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0739971910646042e-05,
+      "loss": 1.0561,
+      "step": 4419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073930684768414e-05,
+      "loss": 1.0991,
+      "step": 4420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0738641621060763e-05,
+      "loss": 1.1595,
+      "step": 4421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0737976230788365e-05,
+      "loss": 1.0969,
+      "step": 4422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073731067687941e-05,
+      "loss": 1.1367,
+      "step": 4423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073664495934636e-05,
+      "loss": 1.1406,
+      "step": 4424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0735979078201685e-05,
+      "loss": 1.0177,
+      "step": 4425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073531303345785e-05,
+      "loss": 1.1488,
+      "step": 4426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0734646825127336e-05,
+      "loss": 1.0991,
+      "step": 4427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073398045322261e-05,
+      "loss": 1.0601,
+      "step": 4428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0733313917756155e-05,
+      "loss": 1.0971,
+      "step": 4429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073264721874046e-05,
+      "loss": 1.1588,
+      "step": 4430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0731980356187996e-05,
+      "loss": 1.076,
+      "step": 4431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0731313330111265e-05,
+      "loss": 1.0553,
+      "step": 4432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.073064614052275e-05,
+      "loss": 1.125,
+      "step": 4433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072997878743495e-05,
+      "loss": 1.1334,
+      "step": 4434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0729311270860355e-05,
+      "loss": 1.155,
+      "step": 4435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072864359081148e-05,
+      "loss": 1.1203,
+      "step": 4436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0727975747300815e-05,
+      "loss": 1.1515,
+      "step": 4437
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072730774034087e-05,
+      "loss": 1.0037,
+      "step": 4438
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0726639569944163e-05,
+      "loss": 1.1017,
+      "step": 4439
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0725971236123192e-05,
+      "loss": 1.1161,
+      "step": 4440
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0725302738890487e-05,
+      "loss": 1.0535,
+      "step": 4441
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072463407825856e-05,
+      "loss": 1.1132,
+      "step": 4442
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0723965254239935e-05,
+      "loss": 1.1581,
+      "step": 4443
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.072329626684714e-05,
+      "loss": 1.1185,
+      "step": 4444
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0722627116092693e-05,
+      "loss": 1.1416,
+      "step": 4445
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0721957801989133e-05,
+      "loss": 1.0589,
+      "step": 4446
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0721288324548996e-05,
+      "loss": 1.1235,
+      "step": 4447
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0720618683784818e-05,
+      "loss": 1.142,
+      "step": 4448
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0719948879709133e-05,
+      "loss": 1.1345,
+      "step": 4449
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0719278912334495e-05,
+      "loss": 1.0739,
+      "step": 4450
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071860878167344e-05,
+      "loss": 1.1496,
+      "step": 4451
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0717938487738523e-05,
+      "loss": 1.0874,
+      "step": 4452
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.07172680305423e-05,
+      "loss": 1.1265,
+      "step": 4453
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071659741009732e-05,
+      "loss": 1.1022,
+      "step": 4454
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0715926626416146e-05,
+      "loss": 1.0975,
+      "step": 4455
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0715255679511336e-05,
+      "loss": 1.1807,
+      "step": 4456
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0714584569395455e-05,
+      "loss": 1.108,
+      "step": 4457
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071391329608108e-05,
+      "loss": 1.1299,
+      "step": 4458
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071324185958077e-05,
+      "loss": 1.1058,
+      "step": 4459
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0712570259907107e-05,
+      "loss": 1.1218,
+      "step": 4460
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0711898497072664e-05,
+      "loss": 1.21,
+      "step": 4461
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071122657109002e-05,
+      "loss": 1.0608,
+      "step": 4462
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071055448197176e-05,
+      "loss": 1.1258,
+      "step": 4463
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0709882229730477e-05,
+      "loss": 1.159,
+      "step": 4464
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0709209814378753e-05,
+      "loss": 1.0735,
+      "step": 4465
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0708537235929177e-05,
+      "loss": 1.1,
+      "step": 4466
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070786449439435e-05,
+      "loss": 1.0771,
+      "step": 4467
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0707191589786872e-05,
+      "loss": 1.1317,
+      "step": 4468
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0706518522119337e-05,
+      "loss": 1.1727,
+      "step": 4469
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.070584529140436e-05,
+      "loss": 1.0738,
+      "step": 4470
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0705171897654537e-05,
+      "loss": 1.1034,
+      "step": 4471
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0704498340882493e-05,
+      "loss": 1.1502,
+      "step": 4472
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0703824621100826e-05,
+      "loss": 1.118,
+      "step": 4473
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0703150738322162e-05,
+      "loss": 1.1129,
+      "step": 4474
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0702476692559123e-05,
+      "loss": 1.1409,
+      "step": 4475
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0701802483824323e-05,
+      "loss": 1.1182,
+      "step": 4476
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0701128112130395e-05,
+      "loss": 1.101,
+      "step": 4477
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0700453577489968e-05,
+      "loss": 1.1171,
+      "step": 4478
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069977887991567e-05,
+      "loss": 1.0519,
+      "step": 4479
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0699104019420135e-05,
+      "loss": 1.1803,
+      "step": 4480
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069842899601601e-05,
+      "loss": 1.0612,
+      "step": 4481
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0697753809715927e-05,
+      "loss": 1.1498,
+      "step": 4482
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0697078460532534e-05,
+      "loss": 1.0848,
+      "step": 4483
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0696402948478478e-05,
+      "loss": 1.0959,
+      "step": 4484
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0695727273566414e-05,
+      "loss": 1.1114,
+      "step": 4485
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0695051435808982e-05,
+      "loss": 1.1197,
+      "step": 4486
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069437543521885e-05,
+      "loss": 1.1317,
+      "step": 4487
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0693699271808677e-05,
+      "loss": 1.1291,
+      "step": 4488
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0693022945591123e-05,
+      "loss": 1.1015,
+      "step": 4489
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0692346456578854e-05,
+      "loss": 1.1579,
+      "step": 4490
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069166980478454e-05,
+      "loss": 1.1202,
+      "step": 4491
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0690992990220845e-05,
+      "loss": 1.1339,
+      "step": 4492
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0690316012900455e-05,
+      "loss": 1.1116,
+      "step": 4493
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068963887283604e-05,
+      "loss": 1.1177,
+      "step": 4494
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0688961570040286e-05,
+      "loss": 1.1014,
+      "step": 4495
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0688284104525874e-05,
+      "loss": 1.0588,
+      "step": 4496
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068760647630549e-05,
+      "loss": 1.2038,
+      "step": 4497
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0686928685391827e-05,
+      "loss": 1.1315,
+      "step": 4498
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0686250731797576e-05,
+      "loss": 1.1487,
+      "step": 4499
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0685572615535435e-05,
+      "loss": 1.1083,
+      "step": 4500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0684894336618096e-05,
+      "loss": 1.0538,
+      "step": 4501
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068421589505827e-05,
+      "loss": 1.0803,
+      "step": 4502
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0683537290868662e-05,
+      "loss": 1.0836,
+      "step": 4503
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068285852406197e-05,
+      "loss": 1.0658,
+      "step": 4504
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068217959465092e-05,
+      "loss": 1.1646,
+      "step": 4505
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0681500502648218e-05,
+      "loss": 1.1858,
+      "step": 4506
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.068082124806658e-05,
+      "loss": 1.1057,
+      "step": 4507
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0680141830918726e-05,
+      "loss": 1.1295,
+      "step": 4508
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0679462251217385e-05,
+      "loss": 1.135,
+      "step": 4509
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067878250897528e-05,
+      "loss": 1.078,
+      "step": 4510
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0678102604205142e-05,
+      "loss": 1.0394,
+      "step": 4511
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0677422536919704e-05,
+      "loss": 1.1575,
+      "step": 4512
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0676742307131697e-05,
+      "loss": 1.0723,
+      "step": 4513
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067606191485387e-05,
+      "loss": 1.1597,
+      "step": 4514
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.067538136009895e-05,
+      "loss": 1.071,
+      "step": 4515
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0674700642879695e-05,
+      "loss": 1.0488,
+      "step": 4516
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0674019763208847e-05,
+      "loss": 1.1417,
+      "step": 4517
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0673338721099153e-05,
+      "loss": 1.091,
+      "step": 4518
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0672657516563377e-05,
+      "loss": 1.1475,
+      "step": 4519
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0671976149614268e-05,
+      "loss": 1.0949,
+      "step": 4520
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0671294620264585e-05,
+      "loss": 1.1084,
+      "step": 4521
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06706129285271e-05,
+      "loss": 1.194,
+      "step": 4522
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066993107441457e-05,
+      "loss": 1.1436,
+      "step": 4523
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066924905793977e-05,
+      "loss": 1.1169,
+      "step": 4524
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0668566879115466e-05,
+      "loss": 1.0383,
+      "step": 4525
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0667884537954442e-05,
+      "loss": 1.1236,
+      "step": 4526
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0667202034469467e-05,
+      "loss": 1.1127,
+      "step": 4527
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0666519368673328e-05,
+      "loss": 1.1261,
+      "step": 4528
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066583654057881e-05,
+      "loss": 1.0924,
+      "step": 4529
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0665153550198692e-05,
+      "loss": 1.0939,
+      "step": 4530
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0664470397545776e-05,
+      "loss": 1.1357,
+      "step": 4531
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066378708263285e-05,
+      "loss": 1.0512,
+      "step": 4532
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0663103605472704e-05,
+      "loss": 1.1031,
+      "step": 4533
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0662419966078152e-05,
+      "loss": 1.1387,
+      "step": 4534
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066173616446198e-05,
+      "loss": 1.2164,
+      "step": 4535
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066105220063701e-05,
+      "loss": 1.0984,
+      "step": 4536
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.066036807461604e-05,
+      "loss": 1.1401,
+      "step": 4537
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0659683786411885e-05,
+      "loss": 1.1749,
+      "step": 4538
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0658999336037358e-05,
+      "loss": 1.0651,
+      "step": 4539
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0658314723505277e-05,
+      "loss": 1.0922,
+      "step": 4540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0657629948828465e-05,
+      "loss": 1.1587,
+      "step": 4541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0656945012019745e-05,
+      "loss": 1.1358,
+      "step": 4542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065625991309194e-05,
+      "loss": 1.1018,
+      "step": 4543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0655574652057887e-05,
+      "loss": 1.0856,
+      "step": 4544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0654889228930415e-05,
+      "loss": 1.1196,
+      "step": 4545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0654203643722357e-05,
+      "loss": 1.1111,
+      "step": 4546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0653517896446557e-05,
+      "loss": 1.0716,
+      "step": 4547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0652831987115854e-05,
+      "loss": 1.1845,
+      "step": 4548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0652145915743096e-05,
+      "loss": 1.135,
+      "step": 4549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0651459682341128e-05,
+      "loss": 1.0877,
+      "step": 4550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06507732869228e-05,
+      "loss": 1.1082,
+      "step": 4551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0650086729500972e-05,
+      "loss": 1.1096,
+      "step": 4552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0649400010088497e-05,
+      "loss": 1.127,
+      "step": 4553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0648713128698233e-05,
+      "loss": 1.0686,
+      "step": 4554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064802608534305e-05,
+      "loss": 1.1066,
+      "step": 4555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064733888003581e-05,
+      "loss": 1.1051,
+      "step": 4556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064665151278938e-05,
+      "loss": 1.1566,
+      "step": 4557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0645963983616635e-05,
+      "loss": 1.1133,
+      "step": 4558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0645276292530456e-05,
+      "loss": 1.0844,
+      "step": 4559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0644588439543712e-05,
+      "loss": 1.0675,
+      "step": 4560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0643900424669286e-05,
+      "loss": 1.0877,
+      "step": 4561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.064321224792007e-05,
+      "loss": 1.1124,
+      "step": 4562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0642523909308945e-05,
+      "loss": 1.1107,
+      "step": 4563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0641835408848802e-05,
+      "loss": 1.1558,
+      "step": 4564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0641146746552536e-05,
+      "loss": 1.1192,
+      "step": 4565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0640457922433044e-05,
+      "loss": 1.1651,
+      "step": 4566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0639768936503226e-05,
+      "loss": 1.1573,
+      "step": 4567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0639079788775982e-05,
+      "loss": 1.1558,
+      "step": 4568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063839047926422e-05,
+      "loss": 1.11,
+      "step": 4569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0637701007980846e-05,
+      "loss": 1.1598,
+      "step": 4570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0637011374938774e-05,
+      "loss": 1.0663,
+      "step": 4571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063632158015092e-05,
+      "loss": 1.0936,
+      "step": 4572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0635631623630197e-05,
+      "loss": 1.1841,
+      "step": 4573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0634941505389533e-05,
+      "loss": 1.1299,
+      "step": 4574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0634251225441847e-05,
+      "loss": 1.0777,
+      "step": 4575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0633560783800066e-05,
+      "loss": 1.1148,
+      "step": 4576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0632870180477122e-05,
+      "loss": 1.1652,
+      "step": 4577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063217941548595e-05,
+      "loss": 1.0915,
+      "step": 4578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0631488488839476e-05,
+      "loss": 1.1201,
+      "step": 4579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0630797400550653e-05,
+      "loss": 1.129,
+      "step": 4580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.063010615063241e-05,
+      "loss": 1.1123,
+      "step": 4581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06294147390977e-05,
+      "loss": 1.1531,
+      "step": 4582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0628723165959474e-05,
+      "loss": 1.1315,
+      "step": 4583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0628031431230675e-05,
+      "loss": 1.1474,
+      "step": 4584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062733953492426e-05,
+      "loss": 1.1469,
+      "step": 4585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0626647477053186e-05,
+      "loss": 1.1242,
+      "step": 4586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0625955257630413e-05,
+      "loss": 1.1366,
+      "step": 4587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0625262876668905e-05,
+      "loss": 1.1416,
+      "step": 4588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0624570334181634e-05,
+      "loss": 1.1028,
+      "step": 4589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0623877630181557e-05,
+      "loss": 1.0585,
+      "step": 4590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0623184764681655e-05,
+      "loss": 1.1395,
+      "step": 4591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0622491737694902e-05,
+      "loss": 1.1759,
+      "step": 4592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0621798549234277e-05,
+      "loss": 1.1223,
+      "step": 4593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0621105199312756e-05,
+      "loss": 1.1303,
+      "step": 4594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.062041168794333e-05,
+      "loss": 1.1237,
+      "step": 4595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0619718015138984e-05,
+      "loss": 1.1774,
+      "step": 4596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0619024180912708e-05,
+      "loss": 1.0568,
+      "step": 4597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0618330185277494e-05,
+      "loss": 1.1385,
+      "step": 4598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0617636028246342e-05,
+      "loss": 1.122,
+      "step": 4599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0616941709832253e-05,
+      "loss": 1.1163,
+      "step": 4600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061624723004822e-05,
+      "loss": 1.1119,
+      "step": 4601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0615552588907258e-05,
+      "loss": 1.0956,
+      "step": 4602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0614857786422374e-05,
+      "loss": 1.1998,
+      "step": 4603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0614162822606576e-05,
+      "loss": 1.1647,
+      "step": 4604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0613467697472877e-05,
+      "loss": 1.071,
+      "step": 4605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06127724110343e-05,
+      "loss": 1.0072,
+      "step": 4606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061207696330387e-05,
+      "loss": 1.108,
+      "step": 4607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06113813542946e-05,
+      "loss": 1.1289,
+      "step": 4608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.061068558401952e-05,
+      "loss": 1.1214,
+      "step": 4609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0609989652491662e-05,
+      "loss": 1.0985,
+      "step": 4610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.060929355972406e-05,
+      "loss": 1.0607,
+      "step": 4611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.060859730572975e-05,
+      "loss": 1.1275,
+      "step": 4612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0607900890521763e-05,
+      "loss": 1.1362,
+      "step": 4613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0607204314113145e-05,
+      "loss": 1.1927,
+      "step": 4614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0606507576516945e-05,
+      "loss": 1.0957,
+      "step": 4615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0605810677746212e-05,
+      "loss": 1.1825,
+      "step": 4616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0605113617813986e-05,
+      "loss": 1.1836,
+      "step": 4617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0604416396733332e-05,
+      "loss": 1.1197,
+      "step": 4618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.06037190145173e-05,
+      "loss": 1.1248,
+      "step": 4619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0603021471178953e-05,
+      "loss": 1.0452,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.060232376673135e-05,
+      "loss": 1.1022,
+      "step": 4621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0601625901187566e-05,
+      "loss": 1.0972,
+      "step": 4622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0600927874560663e-05,
+      "loss": 1.1047,
+      "step": 4623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0600229686863712e-05,
+      "loss": 1.1179,
+      "step": 4624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0599531338109792e-05,
+      "loss": 1.0468,
+      "step": 4625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059883282831198e-05,
+      "loss": 1.1148,
+      "step": 4626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0598134157483355e-05,
+      "loss": 1.0737,
+      "step": 4627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0597435325637e-05,
+      "loss": 1.1568,
+      "step": 4628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0596736332786008e-05,
+      "loss": 1.1144,
+      "step": 4629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0596037178943464e-05,
+      "loss": 1.1401,
+      "step": 4630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0595337864122464e-05,
+      "loss": 1.1021,
+      "step": 4631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0594638388336104e-05,
+      "loss": 1.0597,
+      "step": 4632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0593938751597478e-05,
+      "loss": 1.0977,
+      "step": 4633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.059323895391969e-05,
+      "loss": 1.1099,
+      "step": 4634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0592538995315853e-05,
+      "loss": 1.1086,
+      "step": 4635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0591838875799066e-05,
+      "loss": 1.0988,
+      "step": 4636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0591138595382444e-05,
+      "loss": 1.0632,
+      "step": 4637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05904381540791e-05,
+      "loss": 1.0237,
+      "step": 4638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0589737551902154e-05,
+      "loss": 1.1808,
+      "step": 4639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0589036788864722e-05,
+      "loss": 1.1715,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058833586497993e-05,
+      "loss": 1.1126,
+      "step": 4641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0587634780260902e-05,
+      "loss": 1.1988,
+      "step": 4642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058693353472077e-05,
+      "loss": 1.1274,
+      "step": 4643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0586232128372665e-05,
+      "loss": 1.1033,
+      "step": 4644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0585530561229723e-05,
+      "loss": 1.1118,
+      "step": 4645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058482883330508e-05,
+      "loss": 1.1848,
+      "step": 4646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0584126944611877e-05,
+      "loss": 1.0976,
+      "step": 4647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.058342489516327e-05,
+      "loss": 1.1247,
+      "step": 4648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0582722684972385e-05,
+      "loss": 1.0505,
+      "step": 4649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0582020314052392e-05,
+      "loss": 1.1094,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0581317782416428e-05,
+      "loss": 1.1601,
+      "step": 4651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0580615090077663e-05,
+      "loss": 1.0411,
+      "step": 4652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057991223704925e-05,
+      "loss": 1.1205,
+      "step": 4653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0579209223344352e-05,
+      "loss": 1.021,
+      "step": 4654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0578506048976136e-05,
+      "loss": 1.1088,
+      "step": 4655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0577802713957766e-05,
+      "loss": 1.1838,
+      "step": 4656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0577099218302417e-05,
+      "loss": 1.0705,
+      "step": 4657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0576395562023265e-05,
+      "loss": 1.0326,
+      "step": 4658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0575691745133483e-05,
+      "loss": 1.1331,
+      "step": 4659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0574987767646258e-05,
+      "loss": 1.0789,
+      "step": 4660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0574283629574766e-05,
+      "loss": 1.1121,
+      "step": 4661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0573579330932193e-05,
+      "loss": 1.0335,
+      "step": 4662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0572874871731737e-05,
+      "loss": 1.0184,
+      "step": 4663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0572170251986583e-05,
+      "loss": 1.1008,
+      "step": 4664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.057146547170993e-05,
+      "loss": 1.1024,
+      "step": 4665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0570760530914973e-05,
+      "loss": 1.2168,
+      "step": 4666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0570055429614916e-05,
+      "loss": 1.0958,
+      "step": 4667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0569350167822966e-05,
+      "loss": 1.0544,
+      "step": 4668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0568644745552327e-05,
+      "loss": 1.0579,
+      "step": 4669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056793916281621e-05,
+      "loss": 1.0785,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0567233419627825e-05,
+      "loss": 1.0782,
+      "step": 4671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0566527516000397e-05,
+      "loss": 1.0617,
+      "step": 4672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0565821451947138e-05,
+      "loss": 1.0618,
+      "step": 4673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0565115227481274e-05,
+      "loss": 1.0616,
+      "step": 4674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056440884261603e-05,
+      "loss": 1.1227,
+      "step": 4675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0563702297364634e-05,
+      "loss": 1.133,
+      "step": 4676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0562995591740316e-05,
+      "loss": 1.0726,
+      "step": 4677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0562288725756313e-05,
+      "loss": 1.0538,
+      "step": 4678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0561581699425863e-05,
+      "loss": 1.1124,
+      "step": 4679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0560874512762203e-05,
+      "loss": 1.1138,
+      "step": 4680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.056016716577858e-05,
+      "loss": 1.082,
+      "step": 4681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055945965848824e-05,
+      "loss": 1.085,
+      "step": 4682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0558751990904434e-05,
+      "loss": 1.0617,
+      "step": 4683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0558044163040407e-05,
+      "loss": 1.1292,
+      "step": 4684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0557336174909423e-05,
+      "loss": 1.1473,
+      "step": 4685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0556628026524737e-05,
+      "loss": 1.2016,
+      "step": 4686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0555919717899613e-05,
+      "loss": 1.1722,
+      "step": 4687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0555211249047307e-05,
+      "loss": 1.0623,
+      "step": 4688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05545026199811e-05,
+      "loss": 1.1252,
+      "step": 4689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055379383071425e-05,
+      "loss": 1.1354,
+      "step": 4690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.055308488126004e-05,
+      "loss": 1.1056,
+      "step": 4691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0552375771631743e-05,
+      "loss": 1.0993,
+      "step": 4692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0551666501842636e-05,
+      "loss": 1.086,
+      "step": 4693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0550957071906005e-05,
+      "loss": 1.1649,
+      "step": 4694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0550247481835133e-05,
+      "loss": 1.0966,
+      "step": 4695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0549537731643307e-05,
+      "loss": 1.1082,
+      "step": 4696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0548827821343826e-05,
+      "loss": 1.0887,
+      "step": 4697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054811775094998e-05,
+      "loss": 1.1085,
+      "step": 4698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0547407520475063e-05,
+      "loss": 1.1096,
+      "step": 4699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054669712993238e-05,
+      "loss": 1.0815,
+      "step": 4700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0545986579335233e-05,
+      "loss": 1.1516,
+      "step": 4701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054527586869693e-05,
+      "loss": 1.1061,
+      "step": 4702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0544564998030778e-05,
+      "loss": 1.1513,
+      "step": 4703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0543853967350094e-05,
+      "loss": 1.0277,
+      "step": 4704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0543142776668187e-05,
+      "loss": 1.0385,
+      "step": 4705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0542431425998378e-05,
+      "loss": 1.0854,
+      "step": 4706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0541719915353993e-05,
+      "loss": 1.0919,
+      "step": 4707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.054100824474835e-05,
+      "loss": 1.0559,
+      "step": 4708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0540296414194784e-05,
+      "loss": 1.1254,
+      "step": 4709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0539584423706618e-05,
+      "loss": 1.1541,
+      "step": 4710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0538872273297187e-05,
+      "loss": 1.0644,
+      "step": 4711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0538159962979832e-05,
+      "loss": 1.1636,
+      "step": 4712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0537447492767884e-05,
+      "loss": 1.1215,
+      "step": 4713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0536734862674694e-05,
+      "loss": 1.101,
+      "step": 4714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053602207271361e-05,
+      "loss": 1.0786,
+      "step": 4715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0535309122897966e-05,
+      "loss": 1.1127,
+      "step": 4716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0534596013241125e-05,
+      "loss": 1.1239,
+      "step": 4717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.053388274375644e-05,
+      "loss": 1.1257,
+      "step": 4718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0533169314457265e-05,
+      "loss": 1.0667,
+      "step": 4719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0532455725356963e-05,
+      "loss": 1.0941,
+      "step": 4720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05317419764689e-05,
+      "loss": 1.1702,
+      "step": 4721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0531028067806435e-05,
+      "loss": 1.0892,
+      "step": 4722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0530313999382944e-05,
+      "loss": 1.2166,
+      "step": 4723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0529599771211797e-05,
+      "loss": 1.1213,
+      "step": 4724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0528885383306368e-05,
+      "loss": 1.13,
+      "step": 4725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0528170835680038e-05,
+      "loss": 1.1464,
+      "step": 4726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052745612834619e-05,
+      "loss": 1.1501,
+      "step": 4727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0526741261318202e-05,
+      "loss": 1.0456,
+      "step": 4728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0526026234609465e-05,
+      "loss": 1.1354,
+      "step": 4729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0525311048233375e-05,
+      "loss": 1.114,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0524595702203313e-05,
+      "loss": 1.1121,
+      "step": 4731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0523880196532684e-05,
+      "loss": 1.1669,
+      "step": 4732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0523164531234885e-05,
+      "loss": 1.0367,
+      "step": 4733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052244870632332e-05,
+      "loss": 1.1239,
+      "step": 4734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0521732721811392e-05,
+      "loss": 1.0811,
+      "step": 4735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.052101657771251e-05,
+      "loss": 1.1065,
+      "step": 4736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0520300274040086e-05,
+      "loss": 1.1068,
+      "step": 4737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0519583810807533e-05,
+      "loss": 1.1395,
+      "step": 4738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0518867188028267e-05,
+      "loss": 1.1007,
+      "step": 4739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0518150405715713e-05,
+      "loss": 1.1042,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0517433463883294e-05,
+      "loss": 1.0901,
+      "step": 4741
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0516716362544426e-05,
+      "loss": 1.0918,
+      "step": 4742
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0515999101712552e-05,
+      "loss": 1.1046,
+      "step": 4743
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0515281681401097e-05,
+      "loss": 1.1179,
+      "step": 4744
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.05145641016235e-05,
+      "loss": 1.0745,
+      "step": 4745
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0513846362393193e-05,
+      "loss": 1.086,
+      "step": 4746
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.051312846372362e-05,
+      "loss": 1.133,
+      "step": 4747
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0512410405628226e-05,
+      "loss": 1.0532,
+      "step": 4748
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0511692188120458e-05,
+      "loss": 1.0258,
+      "step": 4749
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0510973811213764e-05,
+      "loss": 1.1553,
+      "step": 4750
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0510255274921598e-05,
+      "loss": 1.1427,
+      "step": 4751
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0509536579257422e-05,
+      "loss": 1.0604,
+      "step": 4752
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0508817724234687e-05,
+      "loss": 1.1371,
+      "step": 4753
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050809870986686e-05,
+      "loss": 1.0719,
+      "step": 4754
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0507379536167402e-05,
+      "loss": 1.0917,
+      "step": 4755
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0506660203149782e-05,
+      "loss": 1.0679,
+      "step": 4756
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0505940710827477e-05,
+      "loss": 1.1251,
+      "step": 4757
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050522105921395e-05,
+      "loss": 1.1415,
+      "step": 4758
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0504501248322687e-05,
+      "loss": 1.1639,
+      "step": 4759
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0503781278167168e-05,
+      "loss": 1.0232,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050306114876087e-05,
+      "loss": 1.0428,
+      "step": 4761
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050234086011728e-05,
+      "loss": 1.0645,
+      "step": 4762
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050162041224989e-05,
+      "loss": 1.0901,
+      "step": 4763
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.050089980517219e-05,
+      "loss": 1.1296,
+      "step": 4764
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0500179038897674e-05,
+      "loss": 1.1037,
+      "step": 4765
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0499458113439847e-05,
+      "loss": 1.1229,
+      "step": 4766
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0498737028812197e-05,
+      "loss": 1.1175,
+      "step": 4767
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049801578502824e-05,
+      "loss": 1.102,
+      "step": 4768
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0497294382101477e-05,
+      "loss": 1.1725,
+      "step": 4769
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0496572820045413e-05,
+      "loss": 1.1286,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049585109887357e-05,
+      "loss": 1.1021,
+      "step": 4771
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0495129218599463e-05,
+      "loss": 1.1194,
+      "step": 4772
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0494407179236604e-05,
+      "loss": 1.1383,
+      "step": 4773
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0493684980798514e-05,
+      "loss": 1.047,
+      "step": 4774
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.049296262329873e-05,
+      "loss": 1.1258,
+      "step": 4775
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0492240106750766e-05,
+      "loss": 1.1369,
+      "step": 4776
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0491517431168158e-05,
+      "loss": 1.2458,
+      "step": 4777
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0490794596564445e-05,
+      "loss": 1.0683,
+      "step": 4778
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0490071602953152e-05,
+      "loss": 1.1158,
+      "step": 4779
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0489348450347826e-05,
+      "loss": 1.0783,
+      "step": 4780
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048862513876201e-05,
+      "loss": 1.13,
+      "step": 4781
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0487901668209246e-05,
+      "loss": 1.0228,
+      "step": 4782
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0487178038703086e-05,
+      "loss": 1.1398,
+      "step": 4783
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0486454250257075e-05,
+      "loss": 1.1331,
+      "step": 4784
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0485730302884777e-05,
+      "loss": 1.0525,
+      "step": 4785
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0485006196599746e-05,
+      "loss": 1.1848,
+      "step": 4786
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0484281931415536e-05,
+      "loss": 1.0219,
+      "step": 4787
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.048355750734572e-05,
+      "loss": 1.1079,
+      "step": 4788
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0482832924403855e-05,
+      "loss": 1.0778,
+      "step": 4789
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0482108182603515e-05,
+      "loss": 1.0672,
+      "step": 4790
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0481383281958275e-05,
+      "loss": 1.1148,
+      "step": 4791
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0480658222481705e-05,
+      "loss": 1.1727,
+      "step": 4792
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047993300418739e-05,
+      "loss": 1.086,
+      "step": 4793
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0479207627088897e-05,
+      "loss": 1.0777,
+      "step": 4794
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047848209119983e-05,
+      "loss": 1.1475,
+      "step": 4795
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0477756396533763e-05,
+      "loss": 1.1103,
+      "step": 4796
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0477030543104288e-05,
+      "loss": 1.0795,
+      "step": 4797
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0476304530924998e-05,
+      "loss": 1.0968,
+      "step": 4798
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047557836000949e-05,
+      "loss": 1.0734,
+      "step": 4799
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047485203037137e-05,
+      "loss": 1.1006,
+      "step": 4800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047412554202423e-05,
+      "loss": 1.1144,
+      "step": 4801
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0473398894981677e-05,
+      "loss": 1.1185,
+      "step": 4802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.047267208925732e-05,
+      "loss": 1.062,
+      "step": 4803
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0471945124864772e-05,
+      "loss": 1.0464,
+      "step": 4804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0471218001817645e-05,
+      "loss": 1.1676,
+      "step": 4805
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0470490720129556e-05,
+      "loss": 1.2082,
+      "step": 4806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0469763279814126e-05,
+      "loss": 1.0978,
+      "step": 4807
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0469035680884974e-05,
+      "loss": 1.1329,
+      "step": 4808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0468307923355733e-05,
+      "loss": 1.1751,
+      "step": 4809
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046758000724002e-05,
+      "loss": 1.2149,
+      "step": 4810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0466851932551482e-05,
+      "loss": 1.0903,
+      "step": 4811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046612369930374e-05,
+      "loss": 1.1014,
+      "step": 4812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046539530751044e-05,
+      "loss": 1.1443,
+      "step": 4813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046466675718522e-05,
+      "loss": 1.0911,
+      "step": 4814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0463938048341722e-05,
+      "loss": 1.1232,
+      "step": 4815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0463209180993596e-05,
+      "loss": 1.0829,
+      "step": 4816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046248015515449e-05,
+      "loss": 1.1628,
+      "step": 4817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0461750970838055e-05,
+      "loss": 1.0316,
+      "step": 4818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.046102162805795e-05,
+      "loss": 1.131,
+      "step": 4819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0460292126827828e-05,
+      "loss": 1.1862,
+      "step": 4820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0459562467161356e-05,
+      "loss": 1.0577,
+      "step": 4821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045883264907219e-05,
+      "loss": 1.1055,
+      "step": 4822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045810267257401e-05,
+      "loss": 1.0302,
+      "step": 4823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045737253768048e-05,
+      "loss": 1.0958,
+      "step": 4824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045664224440527e-05,
+      "loss": 1.1615,
+      "step": 4825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0455911792762057e-05,
+      "loss": 1.1588,
+      "step": 4826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0455181182764528e-05,
+      "loss": 1.194,
+      "step": 4827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0454450414426355e-05,
+      "loss": 1.1014,
+      "step": 4828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0453719487761228e-05,
+      "loss": 1.089,
+      "step": 4829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0452988402782838e-05,
+      "loss": 1.1749,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045225715950487e-05,
+      "loss": 1.1199,
+      "step": 4831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045152575794102e-05,
+      "loss": 1.1133,
+      "step": 4832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.045079419810499e-05,
+      "loss": 1.0491,
+      "step": 4833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0450062480010473e-05,
+      "loss": 1.1102,
+      "step": 4834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0449330603671176e-05,
+      "loss": 1.1597,
+      "step": 4835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.04485985691008e-05,
+      "loss": 1.1259,
+      "step": 4836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.044786637631306e-05,
+      "loss": 1.212,
+      "step": 4837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0447134025321666e-05,
+      "loss": 1.087,
+      "step": 4838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0446401516140333e-05,
+      "loss": 1.1315,
+      "step": 4839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0445668848782774e-05,
+      "loss": 1.1208,
+      "step": 4840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0444936023262716e-05,
+      "loss": 1.1301,
+      "step": 4841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.044420303959388e-05,
+      "loss": 1.1645,
+      "step": 4842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0443469897789994e-05,
+      "loss": 1.0907,
+      "step": 4843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0442736597864784e-05,
+      "loss": 1.0931,
+      "step": 4844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0442003139831983e-05,
+      "loss": 1.1101,
+      "step": 4845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0441269523705332e-05,
+      "loss": 1.099,
+      "step": 4846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0440535749498565e-05,
+      "loss": 1.0988,
+      "step": 4847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0439801817225427e-05,
+      "loss": 1.0687,
+      "step": 4848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0439067726899655e-05,
+      "loss": 1.1902,
+      "step": 4849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0438333478535005e-05,
+      "loss": 1.0616,
+      "step": 4850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.0437599072145218e-05,
+      "loss": 1.1245,
+      "step": 4851
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.043686450774406e-05,
+      "loss": 1.1515,
+      "step": 4852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0436129785345274e-05,
+      "loss": 1.1108,
+      "step": 4853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.043539490496263e-05,
+      "loss": 1.1131,
+      "step": 4854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0434659866609878e-05,
+      "loss": 1.0713,
+      "step": 4855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0433924670300797e-05,
+      "loss": 1.1186,
+      "step": 4856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0433189316049147e-05,
+      "loss": 1.1083,
+      "step": 4857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.04324538038687e-05,
+      "loss": 1.1245,
+      "step": 4858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0431718133773227e-05,
+      "loss": 1.0784,
+      "step": 4859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0430982305776514e-05,
+      "loss": 1.1606,
+      "step": 4860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.043024631989233e-05,
+      "loss": 1.1032,
+      "step": 4861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0429510176134466e-05,
+      "loss": 1.1483,
+      "step": 4862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0428773874516706e-05,
+      "loss": 1.1313,
+      "step": 4863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0428037415052835e-05,
+      "loss": 1.1356,
+      "step": 4864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0427300797756646e-05,
+      "loss": 1.1232,
+      "step": 4865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042656402264194e-05,
+      "loss": 1.1058,
+      "step": 4866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0425827089722508e-05,
+      "loss": 1.0906,
+      "step": 4867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0425089999012147e-05,
+      "loss": 1.1482,
+      "step": 4868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0424352750524672e-05,
+      "loss": 1.0526,
+      "step": 4869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0423615344273878e-05,
+      "loss": 1.1604,
+      "step": 4870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0422877780273582e-05,
+      "loss": 1.0752,
+      "step": 4871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0422140058537598e-05,
+      "loss": 1.1651,
+      "step": 4872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0421402179079732e-05,
+      "loss": 1.0715,
+      "step": 4873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.042066414191381e-05,
+      "loss": 1.1326,
+      "step": 4874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041992594705365e-05,
+      "loss": 1.1192,
+      "step": 4875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041918759451308e-05,
+      "loss": 1.0388,
+      "step": 4876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0418449084305923e-05,
+      "loss": 1.0614,
+      "step": 4877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041771041644601e-05,
+      "loss": 1.1324,
+      "step": 4878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0416971590947174e-05,
+      "loss": 1.1447,
+      "step": 4879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041623260782325e-05,
+      "loss": 1.169,
+      "step": 4880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.041549346708808e-05,
+      "loss": 1.1523,
+      "step": 4881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0414754168755506e-05,
+      "loss": 1.1356,
+      "step": 4882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0414014712839374e-05,
+      "loss": 1.1563,
+      "step": 4883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0413275099353527e-05,
+      "loss": 1.1111,
+      "step": 4884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0412535328311814e-05,
+      "loss": 1.0352,
+      "step": 4885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.04117953997281e-05,
+      "loss": 1.0833,
+      "step": 4886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0411055313616228e-05,
+      "loss": 1.07,
+      "step": 4887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0410315069990066e-05,
+      "loss": 1.0958,
+      "step": 4888
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0409574668863474e-05,
+      "loss": 1.1268,
+      "step": 4889
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040883411025032e-05,
+      "loss": 1.0755,
+      "step": 4890
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040809339416447e-05,
+      "loss": 1.0445,
+      "step": 4891
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0407352520619795e-05,
+      "loss": 1.1539,
+      "step": 4892
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040661148963017e-05,
+      "loss": 1.1265,
+      "step": 4893
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040587030120947e-05,
+      "loss": 1.0969,
+      "step": 4894
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0405128955371584e-05,
+      "loss": 1.134,
+      "step": 4895
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.040438745213039e-05,
+      "loss": 1.1488,
+      "step": 4896
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0403645791499772e-05,
+      "loss": 1.1494,
+      "step": 4897
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0402903973493617e-05,
+      "loss": 1.1583,
+      "step": 4898
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0402161998125823e-05,
+      "loss": 1.0803,
+      "step": 4899
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0401419865410282e-05,
+      "loss": 1.1442,
+      "step": 4900
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0400677575360895e-05,
+      "loss": 1.115,
+      "step": 4901
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0399935127991557e-05,
+      "loss": 1.1018,
+      "step": 4902
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0399192523316177e-05,
+      "loss": 1.1898,
+      "step": 4903
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039844976134866e-05,
+      "loss": 1.0275,
+      "step": 4904
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0397706842102917e-05,
+      "loss": 1.1486,
+      "step": 4905
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0396963765592857e-05,
+      "loss": 1.1464,
+      "step": 4906
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0396220531832402e-05,
+      "loss": 0.9625,
+      "step": 4907
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0395477140835465e-05,
+      "loss": 1.1434,
+      "step": 4908
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039473359261597e-05,
+      "loss": 1.1015,
+      "step": 4909
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039398988718784e-05,
+      "loss": 1.1234,
+      "step": 4910
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0393246024565005e-05,
+      "loss": 1.0579,
+      "step": 4911
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039250200476139e-05,
+      "loss": 1.1002,
+      "step": 4912
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0391757827790933e-05,
+      "loss": 1.1984,
+      "step": 4913
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.039101349366757e-05,
+      "loss": 1.0553,
+      "step": 4914
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0390269002405244e-05,
+      "loss": 1.1441,
+      "step": 4915
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0389524354017885e-05,
+      "loss": 1.0865,
+      "step": 4916
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0388779548519447e-05,
+      "loss": 1.1494,
+      "step": 4917
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038803458592388e-05,
+      "loss": 1.0205,
+      "step": 4918
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038728946624513e-05,
+      "loss": 1.1456,
+      "step": 4919
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038654418949715e-05,
+      "loss": 1.0636,
+      "step": 4920
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.03857987556939e-05,
+      "loss": 1.0917,
+      "step": 4921
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0385053164849343e-05,
+      "loss": 1.1278,
+      "step": 4922
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0384307416977433e-05,
+      "loss": 1.0442,
+      "step": 4923
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038356151209214e-05,
+      "loss": 1.1197,
+      "step": 4924
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0382815450207436e-05,
+      "loss": 1.1549,
+      "step": 4925
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038206923133729e-05,
+      "loss": 1.1789,
+      "step": 4926
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.038132285549567e-05,
+      "loss": 1.1542,
+      "step": 4927
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0380576322696565e-05,
+      "loss": 1.1,
+      "step": 4928
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0379829632953952e-05,
+      "loss": 1.0745,
+      "step": 4929
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0379082786281806e-05,
+      "loss": 1.1582,
+      "step": 4930
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0378335782694124e-05,
+      "loss": 1.1145,
+      "step": 4931
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0377588622204887e-05,
+      "loss": 1.0684,
+      "step": 4932
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0376841304828093e-05,
+      "loss": 1.1428,
+      "step": 4933
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0376093830577734e-05,
+      "loss": 1.084,
+      "step": 4934
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037534619946781e-05,
+      "loss": 1.1344,
+      "step": 4935
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0374598411512317e-05,
+      "loss": 1.0988,
+      "step": 4936
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0373850466725262e-05,
+      "loss": 1.1668,
+      "step": 4937
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037310236512066e-05,
+      "loss": 1.1405,
+      "step": 4938
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0372354106712506e-05,
+      "loss": 1.0765,
+      "step": 4939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0371605691514822e-05,
+      "loss": 1.1674,
+      "step": 4940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.037085711954162e-05,
+      "loss": 1.0553,
+      "step": 4941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0370108390806923e-05,
+      "loss": 1.1058,
+      "step": 4942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036935950532474e-05,
+      "loss": 1.1382,
+      "step": 4943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0368610463109115e-05,
+      "loss": 1.1785,
+      "step": 4944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036786126417406e-05,
+      "loss": 1.0798,
+      "step": 4945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0367111908533614e-05,
+      "loss": 1.0942,
+      "step": 4946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0366362396201807e-05,
+      "loss": 1.0768,
+      "step": 4947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0365612727192672e-05,
+      "loss": 1.1773,
+      "step": 4948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036486290152025e-05,
+      "loss": 1.0545,
+      "step": 4949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0364112919198586e-05,
+      "loss": 1.017,
+      "step": 4950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036336278024172e-05,
+      "loss": 1.0932,
+      "step": 4951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0362612484663708e-05,
+      "loss": 1.1572,
+      "step": 4952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0361862032478592e-05,
+      "loss": 1.1259,
+      "step": 4953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.036111142370043e-05,
+      "loss": 1.1466,
+      "step": 4954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0360360658343276e-05,
+      "loss": 1.0988,
+      "step": 4955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0359609736421194e-05,
+      "loss": 1.0459,
+      "step": 4956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0358858657948242e-05,
+      "loss": 1.0877,
+      "step": 4957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035810742293849e-05,
+      "loss": 1.1095,
+      "step": 4958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0357356031406005e-05,
+      "loss": 1.0383,
+      "step": 4959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0356604483364856e-05,
+      "loss": 1.0851,
+      "step": 4960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0355852778829117e-05,
+      "loss": 1.0997,
+      "step": 4961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0355100917812868e-05,
+      "loss": 1.0345,
+      "step": 4962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035434890033019e-05,
+      "loss": 1.0928,
+      "step": 4963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0353596726395164e-05,
+      "loss": 1.1276,
+      "step": 4964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0352844396021872e-05,
+      "loss": 1.0552,
+      "step": 4965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035209190922441e-05,
+      "loss": 1.1458,
+      "step": 4966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0351339266016866e-05,
+      "loss": 1.1358,
+      "step": 4967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.035058646641334e-05,
+      "loss": 1.1536,
+      "step": 4968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034983351042792e-05,
+      "loss": 1.1436,
+      "step": 4969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0349080398074714e-05,
+      "loss": 1.1405,
+      "step": 4970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0348327129367823e-05,
+      "loss": 1.0554,
+      "step": 4971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0347573704321358e-05,
+      "loss": 1.0808,
+      "step": 4972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034682012294942e-05,
+      "loss": 1.0963,
+      "step": 4973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0346066385266128e-05,
+      "loss": 1.1542,
+      "step": 4974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0345312491285594e-05,
+      "loss": 1.0701,
+      "step": 4975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0344558441021938e-05,
+      "loss": 1.1081,
+      "step": 4976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.034380423448928e-05,
+      "loss": 1.1279,
+      "step": 4977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0343049871701747e-05,
+      "loss": 1.0727,
+      "step": 4978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0342295352673463e-05,
+      "loss": 1.0937,
+      "step": 4979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0341540677418558e-05,
+      "loss": 1.0649,
+      "step": 4980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0340785845951163e-05,
+      "loss": 1.14,
+      "step": 4981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0340030858285418e-05,
+      "loss": 1.0543,
+      "step": 4982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0339275714435464e-05,
+      "loss": 1.1429,
+      "step": 4983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0338520414415433e-05,
+      "loss": 1.0411,
+      "step": 4984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0337764958239478e-05,
+      "loss": 1.1573,
+      "step": 4985
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0337009345921744e-05,
+      "loss": 1.0775,
+      "step": 4986
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033625357747638e-05,
+      "loss": 1.1034,
+      "step": 4987
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033549765291754e-05,
+      "loss": 1.1085,
+      "step": 4988
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033474157225938e-05,
+      "loss": 1.107,
+      "step": 4989
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033398533551606e-05,
+      "loss": 1.133,
+      "step": 4990
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033322894270174e-05,
+      "loss": 1.1027,
+      "step": 4991
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033247239383059e-05,
+      "loss": 1.1438,
+      "step": 4992
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0331715688916773e-05,
+      "loss": 1.0938,
+      "step": 4993
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0330958827974463e-05,
+      "loss": 1.1028,
+      "step": 4994
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.033020181101783e-05,
+      "loss": 1.1184,
+      "step": 4995
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032944463806105e-05,
+      "loss": 1.1061,
+      "step": 4996
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0328687309118313e-05,
+      "loss": 1.1796,
+      "step": 4997
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032792982420379e-05,
+      "loss": 1.1337,
+      "step": 4998
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032717218333167e-05,
+      "loss": 1.1118,
+      "step": 4999
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0326414386516147e-05,
+      "loss": 1.0227,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "eval_text_loss": 0.9965324401855469,
+      "eval_text_runtime": 1.586,
+      "eval_text_samples_per_second": 46.658,
+      "eval_text_steps_per_second": 1.261,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "eval_svit_loss": 1.1787912845611572,
+      "eval_svit_runtime": 4.1248,
+      "eval_svit_samples_per_second": 165.1,
+      "eval_svit_steps_per_second": 2.667,
+      "step": 5000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0325656433771404e-05,
+      "loss": 1.134,
+      "step": 5001
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0324898325111635e-05,
+      "loss": 1.0682,
+      "step": 5002
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0324140060551048e-05,
+      "loss": 1.0987,
+      "step": 5003
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032338164010383e-05,
+      "loss": 1.2112,
+      "step": 5004
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0322623063784194e-05,
+      "loss": 1.136,
+      "step": 5005
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.032186433160634e-05,
+      "loss": 1.1004,
+      "step": 5006
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0321105443584474e-05,
+      "loss": 1.0864,
+      "step": 5007
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0320346399732817e-05,
+      "loss": 1.1144,
+      "step": 5008
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0319587200065576e-05,
+      "loss": 1.0466,
+      "step": 5009
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0318827844596975e-05,
+      "loss": 1.1257,
+      "step": 5010
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0318068333341228e-05,
+      "loss": 1.0512,
+      "step": 5011
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0317308666312564e-05,
+      "loss": 1.118,
+      "step": 5012
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0316548843525203e-05,
+      "loss": 1.1476,
+      "step": 5013
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031578886499338e-05,
+      "loss": 1.0999,
+      "step": 5014
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0315028730731327e-05,
+      "loss": 1.1222,
+      "step": 5015
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031426844075327e-05,
+      "loss": 1.1886,
+      "step": 5016
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0313507995073463e-05,
+      "loss": 1.1032,
+      "step": 5017
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0312747393706136e-05,
+      "loss": 1.0441,
+      "step": 5018
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.031198663666553e-05,
+      "loss": 1.2571,
+      "step": 5019
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0311225723965902e-05,
+      "loss": 1.0611,
+      "step": 5020
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0310464655621492e-05,
+      "loss": 1.1649,
+      "step": 5021
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0309703431646558e-05,
+      "loss": 1.1706,
+      "step": 5022
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0308942052055355e-05,
+      "loss": 1.1194,
+      "step": 5023
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0308180516862142e-05,
+      "loss": 1.1787,
+      "step": 5024
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0307418826081178e-05,
+      "loss": 1.1439,
+      "step": 5025
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0306656979726726e-05,
+      "loss": 1.1047,
+      "step": 5026
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0305894977813057e-05,
+      "loss": 1.087,
+      "step": 5027
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030513282035444e-05,
+      "loss": 1.166,
+      "step": 5028
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0304370507365146e-05,
+      "loss": 1.1841,
+      "step": 5029
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0303608038859454e-05,
+      "loss": 1.164,
+      "step": 5030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030284541485164e-05,
+      "loss": 1.1536,
+      "step": 5031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.030208263535599e-05,
+      "loss": 1.08,
+      "step": 5032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0301319700386777e-05,
+      "loss": 1.0647,
+      "step": 5033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0300556609958305e-05,
+      "loss": 1.1434,
+      "step": 5034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029979336408485e-05,
+      "loss": 1.1481,
+      "step": 5035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0299029962780717e-05,
+      "loss": 1.1201,
+      "step": 5036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0298266406060194e-05,
+      "loss": 1.1576,
+      "step": 5037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0297502693937585e-05,
+      "loss": 1.0891,
+      "step": 5038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029673882642719e-05,
+      "loss": 1.1433,
+      "step": 5039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0295974803543314e-05,
+      "loss": 1.1868,
+      "step": 5040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0295210625300266e-05,
+      "loss": 1.1,
+      "step": 5041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0294446291712356e-05,
+      "loss": 1.1073,
+      "step": 5042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.029368180279389e-05,
+      "loss": 1.0745,
+      "step": 5043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.02929171585592e-05,
+      "loss": 1.0202,
+      "step": 5044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0292152359022596e-05,
+      "loss": 1.0891,
+      "step": 5045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0291387404198405e-05,
+      "loss": 1.1186,
+      "step": 5046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0290622294100944e-05,
+      "loss": 1.0846,
+      "step": 5047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028985702874455e-05,
+      "loss": 1.0908,
+      "step": 5048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0289091608143548e-05,
+      "loss": 1.1104,
+      "step": 5049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028832603231228e-05,
+      "loss": 1.1312,
+      "step": 5050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0287560301265075e-05,
+      "loss": 1.1418,
+      "step": 5051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028679441501628e-05,
+      "loss": 1.0861,
+      "step": 5052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0286028373580227e-05,
+      "loss": 1.0742,
+      "step": 5053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0285262176971268e-05,
+      "loss": 1.1741,
+      "step": 5054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0284495825203758e-05,
+      "loss": 1.116,
+      "step": 5055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0283729318292037e-05,
+      "loss": 1.1207,
+      "step": 5056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028296265625047e-05,
+      "loss": 1.0357,
+      "step": 5057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0282195839093406e-05,
+      "loss": 1.145,
+      "step": 5058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.028142886683521e-05,
+      "loss": 1.1588,
+      "step": 5059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0280661739490243e-05,
+      "loss": 1.0685,
+      "step": 5060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0279894457072872e-05,
+      "loss": 1.1605,
+      "step": 5061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0279127019597467e-05,
+      "loss": 1.0922,
+      "step": 5062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0278359427078396e-05,
+      "loss": 1.114,
+      "step": 5063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0277591679530036e-05,
+      "loss": 1.1276,
+      "step": 5064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0276823776966766e-05,
+      "loss": 1.1244,
+      "step": 5065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0276055719402965e-05,
+      "loss": 1.084,
+      "step": 5066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0275287506853016e-05,
+      "loss": 1.1734,
+      "step": 5067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0274519139331308e-05,
+      "loss": 1.0764,
+      "step": 5068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0273750616852227e-05,
+      "loss": 1.0851,
+      "step": 5069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0272981939430166e-05,
+      "loss": 1.1317,
+      "step": 5070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0272213107079524e-05,
+      "loss": 1.1697,
+      "step": 5071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.027144411981469e-05,
+      "loss": 1.1189,
+      "step": 5072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0270674977650077e-05,
+      "loss": 1.083,
+      "step": 5073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0269905680600078e-05,
+      "loss": 1.0762,
+      "step": 5074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0269136228679106e-05,
+      "loss": 1.1532,
+      "step": 5075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026836662190157e-05,
+      "loss": 1.1218,
+      "step": 5076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0267596860281875e-05,
+      "loss": 1.1708,
+      "step": 5077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0266826943834446e-05,
+      "loss": 1.0284,
+      "step": 5078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0266056872573696e-05,
+      "loss": 1.0846,
+      "step": 5079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026528664651405e-05,
+      "loss": 1.1303,
+      "step": 5080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026451626566993e-05,
+      "loss": 1.0221,
+      "step": 5081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0263745730055763e-05,
+      "loss": 1.1551,
+      "step": 5082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0262975039685978e-05,
+      "loss": 1.0174,
+      "step": 5083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0262204194575006e-05,
+      "loss": 1.0684,
+      "step": 5084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026143319473729e-05,
+      "loss": 1.1089,
+      "step": 5085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.026066204018726e-05,
+      "loss": 1.1334,
+      "step": 5086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.0259890730939362e-05,
+      "loss": 1.1726,
+      "step": 5087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.025911926700804e-05,
+      "loss": 1.1901,
+      "step": 5088
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025834764840774e-05,
+      "loss": 1.1296,
+      "step": 5089
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0257575875152912e-05,
+      "loss": 1.1156,
+      "step": 5090
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0256803947258013e-05,
+      "loss": 1.1667,
+      "step": 5091
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0256031864737493e-05,
+      "loss": 1.033,
+      "step": 5092
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025525962760581e-05,
+      "loss": 1.1016,
+      "step": 5093
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0254487235877437e-05,
+      "loss": 1.0358,
+      "step": 5094
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0253714689566826e-05,
+      "loss": 1.1489,
+      "step": 5095
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0252941988688453e-05,
+      "loss": 1.1322,
+      "step": 5096
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0252169133256783e-05,
+      "loss": 1.1523,
+      "step": 5097
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.025139612328629e-05,
+      "loss": 1.0495,
+      "step": 5098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0250622958791453e-05,
+      "loss": 1.1432,
+      "step": 5099
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.024984963978675e-05,
+      "loss": 1.1007,
+      "step": 5100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0249076166286662e-05,
+      "loss": 1.078,
+      "step": 5101
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0248302538305673e-05,
+      "loss": 1.0482,
+      "step": 5102
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0247528755858272e-05,
+      "loss": 1.1255,
+      "step": 5103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0246754818958957e-05,
+      "loss": 1.0923,
+      "step": 5104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0245980727622204e-05,
+      "loss": 1.1448,
+      "step": 5105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0245206481862526e-05,
+      "loss": 1.1614,
+      "step": 5106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0244432081694417e-05,
+      "loss": 1.0399,
+      "step": 5107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0243657527132374e-05,
+      "loss": 1.1132,
+      "step": 5108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0242882818190906e-05,
+      "loss": 1.1333,
+      "step": 5109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0242107954884527e-05,
+      "loss": 1.1023,
+      "step": 5110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0241332937227736e-05,
+      "loss": 1.1193,
+      "step": 5111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0240557765235057e-05,
+      "loss": 1.0539,
+      "step": 5112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0239782438921004e-05,
+      "loss": 1.0791,
+      "step": 5113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0239006958300097e-05,
+      "loss": 1.1672,
+      "step": 5114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0238231323386853e-05,
+      "loss": 1.1006,
+      "step": 5115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0237455534195803e-05,
+      "loss": 1.0978,
+      "step": 5116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0236679590741474e-05,
+      "loss": 1.1537,
+      "step": 5117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0235903493038396e-05,
+      "loss": 1.1223,
+      "step": 5118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0235127241101104e-05,
+      "loss": 1.1241,
+      "step": 5119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0234350834944134e-05,
+      "loss": 1.1391,
+      "step": 5120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0233574274582024e-05,
+      "loss": 1.0848,
+      "step": 5121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0232797560029324e-05,
+      "loss": 1.1223,
+      "step": 5122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0232020691300573e-05,
+      "loss": 1.1269,
+      "step": 5123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023124366841032e-05,
+      "loss": 1.1674,
+      "step": 5124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.023046649137312e-05,
+      "loss": 1.0404,
+      "step": 5125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0229689160203523e-05,
+      "loss": 1.0469,
+      "step": 5126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022891167491609e-05,
+      "loss": 1.1164,
+      "step": 5127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0228134035525375e-05,
+      "loss": 1.1294,
+      "step": 5128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0227356242045948e-05,
+      "loss": 1.1043,
+      "step": 5129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022657829449237e-05,
+      "loss": 1.0633,
+      "step": 5130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0225800192879213e-05,
+      "loss": 1.1043,
+      "step": 5131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0225021937221046e-05,
+      "loss": 1.1436,
+      "step": 5132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022424352753244e-05,
+      "loss": 1.1238,
+      "step": 5133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.022346496382798e-05,
+      "loss": 1.065,
+      "step": 5134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0222686246122245e-05,
+      "loss": 1.0534,
+      "step": 5135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0221907374429816e-05,
+      "loss": 1.1409,
+      "step": 5136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0221128348765276e-05,
+      "loss": 1.1384,
+      "step": 5137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0220349169143217e-05,
+      "loss": 1.0827,
+      "step": 5138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021956983557823e-05,
+      "loss": 1.0423,
+      "step": 5139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0218790348084916e-05,
+      "loss": 1.0896,
+      "step": 5140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021801070667786e-05,
+      "loss": 1.1584,
+      "step": 5141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0217230911371668e-05,
+      "loss": 1.0945,
+      "step": 5142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021645096218095e-05,
+      "loss": 1.0587,
+      "step": 5143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0215670859120302e-05,
+      "loss": 1.1169,
+      "step": 5144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021489060220434e-05,
+      "loss": 1.0411,
+      "step": 5145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021411019144767e-05,
+      "loss": 1.0439,
+      "step": 5146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021332962686491e-05,
+      "loss": 1.0565,
+      "step": 5147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021254890847068e-05,
+      "loss": 1.0591,
+      "step": 5148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.02117680362796e-05,
+      "loss": 1.1711,
+      "step": 5149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021098701030629e-05,
+      "loss": 1.1333,
+      "step": 5150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.021020583056538e-05,
+      "loss": 1.155,
+      "step": 5151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0209424497071496e-05,
+      "loss": 1.1296,
+      "step": 5152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020864300983927e-05,
+      "loss": 1.1252,
+      "step": 5153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020786136888334e-05,
+      "loss": 1.1367,
+      "step": 5154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020707957421834e-05,
+      "loss": 1.1197,
+      "step": 5155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0206297625858914e-05,
+      "loss": 1.0949,
+      "step": 5156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0205515523819706e-05,
+      "loss": 1.0629,
+      "step": 5157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0204733268115357e-05,
+      "loss": 1.0744,
+      "step": 5158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0203950858760523e-05,
+      "loss": 1.0807,
+      "step": 5159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0203168295769854e-05,
+      "loss": 1.157,
+      "step": 5160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0202385579158002e-05,
+      "loss": 1.144,
+      "step": 5161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.020160270893963e-05,
+      "loss": 1.1157,
+      "step": 5162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0200819685129396e-05,
+      "loss": 1.0532,
+      "step": 5163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0200036507741965e-05,
+      "loss": 1.0646,
+      "step": 5164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0199253176792e-05,
+      "loss": 1.1019,
+      "step": 5165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019846969229418e-05,
+      "loss": 1.1819,
+      "step": 5166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0197686054263163e-05,
+      "loss": 1.0876,
+      "step": 5167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0196902262713633e-05,
+      "loss": 1.1252,
+      "step": 5168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019611831766027e-05,
+      "loss": 1.0891,
+      "step": 5169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019533421911775e-05,
+      "loss": 1.1446,
+      "step": 5170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019454996710076e-05,
+      "loss": 1.0838,
+      "step": 5171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0193765561623983e-05,
+      "loss": 1.0963,
+      "step": 5172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0192981002702112e-05,
+      "loss": 1.1201,
+      "step": 5173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019219629034984e-05,
+      "loss": 1.0835,
+      "step": 5174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.019141142458186e-05,
+      "loss": 1.1395,
+      "step": 5175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0190626405412873e-05,
+      "loss": 1.1053,
+      "step": 5176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0189841232857573e-05,
+      "loss": 1.1311,
+      "step": 5177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0189055906930675e-05,
+      "loss": 1.0844,
+      "step": 5178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0188270427646876e-05,
+      "loss": 1.022,
+      "step": 5179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018748479502089e-05,
+      "loss": 1.1045,
+      "step": 5180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0186699009067432e-05,
+      "loss": 1.1102,
+      "step": 5181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0185913069801215e-05,
+      "loss": 1.1217,
+      "step": 5182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0185126977236957e-05,
+      "loss": 1.1052,
+      "step": 5183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0184340731389377e-05,
+      "loss": 1.2135,
+      "step": 5184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0183554332273206e-05,
+      "loss": 1.0825,
+      "step": 5185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018276777990316e-05,
+      "loss": 1.0847,
+      "step": 5186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0181981074293977e-05,
+      "loss": 1.104,
+      "step": 5187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018119421546039e-05,
+      "loss": 1.0668,
+      "step": 5188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.018040720341713e-05,
+      "loss": 1.0871,
+      "step": 5189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017962003817894e-05,
+      "loss": 1.1033,
+      "step": 5190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017883271976056e-05,
+      "loss": 1.0868,
+      "step": 5191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0178045248176732e-05,
+      "loss": 1.1191,
+      "step": 5192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0177257623442202e-05,
+      "loss": 1.13,
+      "step": 5193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0176469845571725e-05,
+      "loss": 1.0982,
+      "step": 5194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017568191458005e-05,
+      "loss": 1.1304,
+      "step": 5195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0174893830481933e-05,
+      "loss": 1.0923,
+      "step": 5196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0174105593292133e-05,
+      "loss": 1.1044,
+      "step": 5197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0173317203025414e-05,
+      "loss": 1.1659,
+      "step": 5198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0172528659696534e-05,
+      "loss": 1.1586,
+      "step": 5199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0171739963320263e-05,
+      "loss": 1.0707,
+      "step": 5200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0170951113911376e-05,
+      "loss": 1.157,
+      "step": 5201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.017016211148464e-05,
+      "loss": 1.0707,
+      "step": 5202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0169372956054833e-05,
+      "loss": 1.071,
+      "step": 5203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0168583647636728e-05,
+      "loss": 1.1445,
+      "step": 5204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0167794186245117e-05,
+      "loss": 1.1243,
+      "step": 5205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0167004571894775e-05,
+      "loss": 1.1091,
+      "step": 5206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0166214804600495e-05,
+      "loss": 1.1411,
+      "step": 5207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0165424884377062e-05,
+      "loss": 1.1126,
+      "step": 5208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0164634811239273e-05,
+      "loss": 1.0952,
+      "step": 5209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0163844585201923e-05,
+      "loss": 1.09,
+      "step": 5210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0163054206279808e-05,
+      "loss": 1.1916,
+      "step": 5211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0162263674487732e-05,
+      "loss": 1.0805,
+      "step": 5212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.01614729898405e-05,
+      "loss": 1.044,
+      "step": 5213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0160682152352915e-05,
+      "loss": 1.0738,
+      "step": 5214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0159891162039793e-05,
+      "loss": 1.003,
+      "step": 5215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0159100018915945e-05,
+      "loss": 1.0998,
+      "step": 5216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0158308722996182e-05,
+      "loss": 1.0915,
+      "step": 5217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0157517274295324e-05,
+      "loss": 1.1712,
+      "step": 5218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0156725672828202e-05,
+      "loss": 1.0798,
+      "step": 5219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0155933918609628e-05,
+      "loss": 1.1144,
+      "step": 5220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0155142011654436e-05,
+      "loss": 1.0949,
+      "step": 5221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0154349951977455e-05,
+      "loss": 1.1019,
+      "step": 5222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0153557739593515e-05,
+      "loss": 1.0721,
+      "step": 5223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0152765374517456e-05,
+      "loss": 1.1208,
+      "step": 5224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0151972856764115e-05,
+      "loss": 1.125,
+      "step": 5225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0151180186348332e-05,
+      "loss": 1.1431,
+      "step": 5226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0150387363284954e-05,
+      "loss": 1.1068,
+      "step": 5227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0149594387588827e-05,
+      "loss": 1.0924,
+      "step": 5228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0148801259274802e-05,
+      "loss": 1.0903,
+      "step": 5229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0148007978357728e-05,
+      "loss": 1.0726,
+      "step": 5230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0147214544852467e-05,
+      "loss": 1.1284,
+      "step": 5231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014642095877387e-05,
+      "loss": 1.0975,
+      "step": 5232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.014562722013681e-05,
+      "loss": 1.115,
+      "step": 5233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0144833328956142e-05,
+      "loss": 1.1426,
+      "step": 5234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0144039285246733e-05,
+      "loss": 1.162,
+      "step": 5235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0143245089023454e-05,
+      "loss": 1.1266,
+      "step": 5236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0142450740301184e-05,
+      "loss": 1.0655,
+      "step": 5237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0141656239094796e-05,
+      "loss": 1.1551,
+      "step": 5238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0140861585419163e-05,
+      "loss": 1.0658,
+      "step": 5239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0140066779289174e-05,
+      "loss": 1.1333,
+      "step": 5240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0139271820719706e-05,
+      "loss": 1.0479,
+      "step": 5241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0138476709725652e-05,
+      "loss": 1.1311,
+      "step": 5242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0137681446321904e-05,
+      "loss": 1.0805,
+      "step": 5243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013688603052335e-05,
+      "loss": 1.0789,
+      "step": 5244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0136090462344885e-05,
+      "loss": 1.008,
+      "step": 5245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0135294741801413e-05,
+      "loss": 1.0817,
+      "step": 5246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0134498868907827e-05,
+      "loss": 1.1162,
+      "step": 5247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013370284367904e-05,
+      "loss": 1.1469,
+      "step": 5248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0132906666129957e-05,
+      "loss": 1.109,
+      "step": 5249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0132110336275486e-05,
+      "loss": 1.1032,
+      "step": 5250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.013131385413054e-05,
+      "loss": 1.0861,
+      "step": 5251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0130517219710036e-05,
+      "loss": 1.1841,
+      "step": 5252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0129720433028894e-05,
+      "loss": 1.1463,
+      "step": 5253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012892349410203e-05,
+      "loss": 1.1095,
+      "step": 5254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0128126402944373e-05,
+      "loss": 1.1252,
+      "step": 5255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0127329159570852e-05,
+      "loss": 1.1147,
+      "step": 5256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0126531763996392e-05,
+      "loss": 1.0458,
+      "step": 5257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012573421623593e-05,
+      "loss": 1.0839,
+      "step": 5258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.01249365163044e-05,
+      "loss": 1.094,
+      "step": 5259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0124138664216735e-05,
+      "loss": 1.0944,
+      "step": 5260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.012334065998789e-05,
+      "loss": 1.1497,
+      "step": 5261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0122542503632794e-05,
+      "loss": 1.0865,
+      "step": 5262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0121744195166404e-05,
+      "loss": 1.0407,
+      "step": 5263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0120945734603668e-05,
+      "loss": 1.1064,
+      "step": 5264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0120147121959537e-05,
+      "loss": 1.0931,
+      "step": 5265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011934835724897e-05,
+      "loss": 1.1077,
+      "step": 5266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011854944048692e-05,
+      "loss": 1.1208,
+      "step": 5267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0117750371688353e-05,
+      "loss": 1.0665,
+      "step": 5268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011695115086823e-05,
+      "loss": 1.1629,
+      "step": 5269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0116151778041522e-05,
+      "loss": 1.0841,
+      "step": 5270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0115352253223196e-05,
+      "loss": 1.0537,
+      "step": 5271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0114552576428226e-05,
+      "loss": 1.082,
+      "step": 5272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0113752747671587e-05,
+      "loss": 1.1166,
+      "step": 5273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0112952766968253e-05,
+      "loss": 1.1566,
+      "step": 5274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0112152634333215e-05,
+      "loss": 1.0495,
+      "step": 5275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.011135234978145e-05,
+      "loss": 1.1537,
+      "step": 5276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0110551913327945e-05,
+      "loss": 1.0922,
+      "step": 5277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0109751324987695e-05,
+      "loss": 1.1751,
+      "step": 5278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0108950584775687e-05,
+      "loss": 1.1388,
+      "step": 5279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010814969270692e-05,
+      "loss": 1.1596,
+      "step": 5280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010734864879639e-05,
+      "loss": 1.1107,
+      "step": 5281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0106547453059102e-05,
+      "loss": 1.1259,
+      "step": 5282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0105746105510055e-05,
+      "loss": 1.0039,
+      "step": 5283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010494460616426e-05,
+      "loss": 1.0907,
+      "step": 5284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0104142955036724e-05,
+      "loss": 1.0252,
+      "step": 5285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.010334115214246e-05,
+      "loss": 1.1388,
+      "step": 5286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0102539197496485e-05,
+      "loss": 1.105,
+      "step": 5287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0101737091113812e-05,
+      "loss": 1.1299,
+      "step": 5288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0100934833009472e-05,
+      "loss": 1.1209,
+      "step": 5289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0100132423198482e-05,
+      "loss": 1.1056,
+      "step": 5290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009932986169587e-05,
+      "loss": 1.0911,
+      "step": 5291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0098527148516664e-05,
+      "loss": 1.085,
+      "step": 5292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.00977242836759e-05,
+      "loss": 1.1138,
+      "step": 5293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0096921267188606e-05,
+      "loss": 1.0678,
+      "step": 5294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.009611809906983e-05,
+      "loss": 1.1223,
+      "step": 5295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0095314779334607e-05,
+      "loss": 1.0925,
+      "step": 5296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0094511307997984e-05,
+      "loss": 1.1599,
+      "step": 5297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0093707685075006e-05,
+      "loss": 1.1414,
+      "step": 5298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0092903910580724e-05,
+      "loss": 1.1565,
+      "step": 5299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0092099984530184e-05,
+      "loss": 1.107,
+      "step": 5300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0091295906938446e-05,
+      "loss": 1.0562,
+      "step": 5301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0090491677820575e-05,
+      "loss": 1.1121,
+      "step": 5302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008968729719162e-05,
+      "loss": 1.0705,
+      "step": 5303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0088882765066647e-05,
+      "loss": 1.1837,
+      "step": 5304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008807808146073e-05,
+      "loss": 1.1396,
+      "step": 5305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0087273246388928e-05,
+      "loss": 1.0782,
+      "step": 5306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008646825986632e-05,
+      "loss": 1.1101,
+      "step": 5307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0085663121907982e-05,
+      "loss": 1.0462,
+      "step": 5308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008485783252899e-05,
+      "loss": 1.1241,
+      "step": 5309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0084052391744422e-05,
+      "loss": 1.14,
+      "step": 5310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.008324679956936e-05,
+      "loss": 1.1112,
+      "step": 5311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0082441056018898e-05,
+      "loss": 1.0308,
+      "step": 5312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0081635161108126e-05,
+      "loss": 1.0628,
+      "step": 5313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0080829114852123e-05,
+      "loss": 1.1276,
+      "step": 5314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0080022917265996e-05,
+      "loss": 1.1106,
+      "step": 5315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007921656836484e-05,
+      "loss": 1.1607,
+      "step": 5316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007841006816375e-05,
+      "loss": 1.055,
+      "step": 5317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007760341667784e-05,
+      "loss": 1.0683,
+      "step": 5318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0076796613922206e-05,
+      "loss": 1.124,
+      "step": 5319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0075989659911957e-05,
+      "loss": 1.1762,
+      "step": 5320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0075182554662213e-05,
+      "loss": 1.1158,
+      "step": 5321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0074375298188087e-05,
+      "loss": 1.1354,
+      "step": 5322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007356789050469e-05,
+      "loss": 1.1445,
+      "step": 5323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.007276033162715e-05,
+      "loss": 1.0262,
+      "step": 5324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.0071952621570584e-05,
+      "loss": 1.1028,
+      "step": 5325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.007114476035012e-05,
+      "loss": 1.1698,
+      "step": 5326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0070336747980892e-05,
+      "loss": 1.0169,
+      "step": 5327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0069528584478024e-05,
+      "loss": 1.1223,
+      "step": 5328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0068720269856655e-05,
+      "loss": 1.1633,
+      "step": 5329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006791180413192e-05,
+      "loss": 1.2049,
+      "step": 5330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006710318731896e-05,
+      "loss": 1.0987,
+      "step": 5331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0066294419432916e-05,
+      "loss": 1.1039,
+      "step": 5332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006548550048894e-05,
+      "loss": 1.0363,
+      "step": 5333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0064676430502173e-05,
+      "loss": 1.1009,
+      "step": 5334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0063867209487774e-05,
+      "loss": 1.1351,
+      "step": 5335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0063057837460896e-05,
+      "loss": 1.0594,
+      "step": 5336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0062248314436686e-05,
+      "loss": 1.1439,
+      "step": 5337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.006143864043032e-05,
+      "loss": 1.1426,
+      "step": 5338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0060628815456947e-05,
+      "loss": 1.1523,
+      "step": 5339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0059818839531742e-05,
+      "loss": 1.1463,
+      "step": 5340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0059008712669867e-05,
+      "loss": 1.042,
+      "step": 5341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0058198434886493e-05,
+      "loss": 1.0713,
+      "step": 5342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0057388006196804e-05,
+      "loss": 1.1173,
+      "step": 5343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0056577426615964e-05,
+      "loss": 1.0739,
+      "step": 5344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0055766696159163e-05,
+      "loss": 1.0918,
+      "step": 5345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005495581484158e-05,
+      "loss": 1.0933,
+      "step": 5346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0054144782678398e-05,
+      "loss": 1.1191,
+      "step": 5347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0053333599684805e-05,
+      "loss": 1.0617,
+      "step": 5348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0052522265875993e-05,
+      "loss": 1.101,
+      "step": 5349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005171078126716e-05,
+      "loss": 1.0113,
+      "step": 5350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0050899145873502e-05,
+      "loss": 1.1334,
+      "step": 5351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.005008735971021e-05,
+      "loss": 1.0881,
+      "step": 5352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00492754227925e-05,
+      "loss": 1.0936,
+      "step": 5353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0048463335135567e-05,
+      "loss": 1.0318,
+      "step": 5354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004765109675462e-05,
+      "loss": 1.0666,
+      "step": 5355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0046838707664873e-05,
+      "loss": 1.0629,
+      "step": 5356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004602616788154e-05,
+      "loss": 1.1122,
+      "step": 5357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0045213477419832e-05,
+      "loss": 1.1377,
+      "step": 5358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0044400636294976e-05,
+      "loss": 1.1495,
+      "step": 5359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.004358764452219e-05,
+      "loss": 1.1235,
+      "step": 5360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0042774502116696e-05,
+      "loss": 1.152,
+      "step": 5361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0041961209093724e-05,
+      "loss": 1.0675,
+      "step": 5362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0041147765468508e-05,
+      "loss": 1.1296,
+      "step": 5363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0040334171256277e-05,
+      "loss": 1.0694,
+      "step": 5364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0039520426472275e-05,
+      "loss": 1.0622,
+      "step": 5365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003870653113173e-05,
+      "loss": 1.1109,
+      "step": 5366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003789248524989e-05,
+      "loss": 1.0859,
+      "step": 5367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0037078288841997e-05,
+      "loss": 1.0837,
+      "step": 5368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0036263941923303e-05,
+      "loss": 1.1062,
+      "step": 5369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0035449444509052e-05,
+      "loss": 1.0862,
+      "step": 5370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0034634796614506e-05,
+      "loss": 1.0696,
+      "step": 5371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003381999825491e-05,
+      "loss": 1.0629,
+      "step": 5372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003300504944553e-05,
+      "loss": 1.0929,
+      "step": 5373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0032189950201632e-05,
+      "loss": 1.0845,
+      "step": 5374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.003137470053847e-05,
+      "loss": 1.1446,
+      "step": 5375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0030559300471313e-05,
+      "loss": 1.1636,
+      "step": 5376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0029743750015437e-05,
+      "loss": 1.0415,
+      "step": 5377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002892804918611e-05,
+      "loss": 1.0676,
+      "step": 5378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002811219799861e-05,
+      "loss": 1.0735,
+      "step": 5379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0027296196468214e-05,
+      "loss": 1.1427,
+      "step": 5380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0026480044610204e-05,
+      "loss": 1.137,
+      "step": 5381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0025663742439865e-05,
+      "loss": 1.1425,
+      "step": 5382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002484728997248e-05,
+      "loss": 1.0532,
+      "step": 5383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0024030687223348e-05,
+      "loss": 1.1219,
+      "step": 5384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002321393420775e-05,
+      "loss": 1.1482,
+      "step": 5385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0022397030940987e-05,
+      "loss": 1.1636,
+      "step": 5386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.002157997743836e-05,
+      "loss": 1.073,
+      "step": 5387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0020762773715164e-05,
+      "loss": 1.1151,
+      "step": 5388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0019945419786707e-05,
+      "loss": 1.1211,
+      "step": 5389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0019127915668295e-05,
+      "loss": 1.0981,
+      "step": 5390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0018310261375233e-05,
+      "loss": 1.1509,
+      "step": 5391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001749245692284e-05,
+      "loss": 1.0453,
+      "step": 5392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001667450232643e-05,
+      "loss": 1.1172,
+      "step": 5393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.001585639760132e-05,
+      "loss": 1.0844,
+      "step": 5394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0015038142762824e-05,
+      "loss": 1.1014,
+      "step": 5395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0014219737826278e-05,
+      "loss": 1.1352,
+      "step": 5396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0013401182806996e-05,
+      "loss": 1.1649,
+      "step": 5397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0012582477720314e-05,
+      "loss": 1.057,
+      "step": 5398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0011763622581565e-05,
+      "loss": 1.1317,
+      "step": 5399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0010944617406078e-05,
+      "loss": 1.0195,
+      "step": 5400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0010125462209195e-05,
+      "loss": 1.183,
+      "step": 5401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0009306157006256e-05,
+      "loss": 1.1094,
+      "step": 5402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0008486701812603e-05,
+      "loss": 1.0506,
+      "step": 5403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0007667096643584e-05,
+      "loss": 1.0937,
+      "step": 5404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0006847341514545e-05,
+      "loss": 1.0588,
+      "step": 5405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000602743644084e-05,
+      "loss": 1.0882,
+      "step": 5406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000520738143782e-05,
+      "loss": 1.0385,
+      "step": 5407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0004387176520845e-05,
+      "loss": 1.0523,
+      "step": 5408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0003566821705273e-05,
+      "loss": 1.0949,
+      "step": 5409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.000274631700647e-05,
+      "loss": 1.0721,
+      "step": 5410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.00019256624398e-05,
+      "loss": 1.0843,
+      "step": 5411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0001104858020633e-05,
+      "loss": 1.0628,
+      "step": 5412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.0000283903764337e-05,
+      "loss": 1.152,
+      "step": 5413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999946279968629e-05,
+      "loss": 1.1346,
+      "step": 5414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9998641545801862e-05,
+      "loss": 1.1238,
+      "step": 5415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9997820142126438e-05,
+      "loss": 1.0597,
+      "step": 5416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99969985886754e-05,
+      "loss": 1.0397,
+      "step": 5417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9996176885464134e-05,
+      "loss": 1.145,
+      "step": 5418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999535503250803e-05,
+      "loss": 1.1802,
+      "step": 5419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999453302982247e-05,
+      "loss": 1.1222,
+      "step": 5420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999371087742286e-05,
+      "loss": 1.1054,
+      "step": 5421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9992888575324585e-05,
+      "loss": 1.0211,
+      "step": 5422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999206612354305e-05,
+      "loss": 1.1653,
+      "step": 5423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999124352209366e-05,
+      "loss": 1.0912,
+      "step": 5424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.999042077099181e-05,
+      "loss": 1.0814,
+      "step": 5425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998959787025292e-05,
+      "loss": 1.1432,
+      "step": 5426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9988774819892392e-05,
+      "loss": 1.162,
+      "step": 5427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998795161992564e-05,
+      "loss": 1.1327,
+      "step": 5428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987128270368086e-05,
+      "loss": 1.0632,
+      "step": 5429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986304771235143e-05,
+      "loss": 1.0688,
+      "step": 5430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9985481122542237e-05,
+      "loss": 1.1141,
+      "step": 5431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9984657324304786e-05,
+      "loss": 1.0941,
+      "step": 5432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9983833376538224e-05,
+      "loss": 1.0361,
+      "step": 5433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998300927925798e-05,
+      "loss": 1.1437,
+      "step": 5434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9982185032479484e-05,
+      "loss": 1.1213,
+      "step": 5435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9981360636218175e-05,
+      "loss": 1.1639,
+      "step": 5436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9980536090489487e-05,
+      "loss": 1.075,
+      "step": 5437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997971139530886e-05,
+      "loss": 1.1143,
+      "step": 5438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997888655069175e-05,
+      "loss": 1.1552,
+      "step": 5439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9978061556653594e-05,
+      "loss": 1.1659,
+      "step": 5440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9977236413209842e-05,
+      "loss": 1.1573,
+      "step": 5441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997641112037595e-05,
+      "loss": 1.1341,
+      "step": 5442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997558567816737e-05,
+      "loss": 1.1013,
+      "step": 5443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9974760086599562e-05,
+      "loss": 1.1285,
+      "step": 5444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9973934345687988e-05,
+      "loss": 1.1551,
+      "step": 5445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9973108455448107e-05,
+      "loss": 1.0772,
+      "step": 5446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9972282415895392e-05,
+      "loss": 1.1584,
+      "step": 5447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9971456227045308e-05,
+      "loss": 1.1083,
+      "step": 5448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.997062988891333e-05,
+      "loss": 1.0972,
+      "step": 5449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996980340151493e-05,
+      "loss": 1.1266,
+      "step": 5450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9968976764865587e-05,
+      "loss": 1.0867,
+      "step": 5451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9968149978980775e-05,
+      "loss": 1.1049,
+      "step": 5452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9967323043875992e-05,
+      "loss": 1.1041,
+      "step": 5453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996649595956671e-05,
+      "loss": 1.1039,
+      "step": 5454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9965668726068426e-05,
+      "loss": 1.1567,
+      "step": 5455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996484134339663e-05,
+      "loss": 1.11,
+      "step": 5456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9964013811566817e-05,
+      "loss": 1.0843,
+      "step": 5457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963186130594477e-05,
+      "loss": 1.0929,
+      "step": 5458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962358300495118e-05,
+      "loss": 1.2049,
+      "step": 5459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996153032128424e-05,
+      "loss": 1.2245,
+      "step": 5460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996070219297735e-05,
+      "loss": 1.0447,
+      "step": 5461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959873915589956e-05,
+      "loss": 1.0904,
+      "step": 5462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959045489137566e-05,
+      "loss": 1.1234,
+      "step": 5463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958216913635702e-05,
+      "loss": 1.132,
+      "step": 5464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957388189099872e-05,
+      "loss": 1.0845,
+      "step": 5465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99565593155456e-05,
+      "loss": 1.1107,
+      "step": 5466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9955730292988407e-05,
+      "loss": 1.0381,
+      "step": 5467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9954901121443817e-05,
+      "loss": 1.1276,
+      "step": 5468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995407180092736e-05,
+      "loss": 1.1357,
+      "step": 5469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9953242331454565e-05,
+      "loss": 1.1012,
+      "step": 5470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995241271304097e-05,
+      "loss": 1.1294,
+      "step": 5471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995158294570211e-05,
+      "loss": 1.075,
+      "step": 5472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9950753029453518e-05,
+      "loss": 1.0419,
+      "step": 5473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994992296431074e-05,
+      "loss": 1.0821,
+      "step": 5474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9949092750289323e-05,
+      "loss": 1.0633,
+      "step": 5475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994826238740481e-05,
+      "loss": 1.0928,
+      "step": 5476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9947431875672754e-05,
+      "loss": 1.07,
+      "step": 5477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994660121510871e-05,
+      "loss": 1.0568,
+      "step": 5478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994577040572823e-05,
+      "loss": 1.0354,
+      "step": 5479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9944939447546872e-05,
+      "loss": 1.0521,
+      "step": 5480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99441083405802e-05,
+      "loss": 1.0228,
+      "step": 5481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994327708484378e-05,
+      "loss": 1.1246,
+      "step": 5482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.994244568035317e-05,
+      "loss": 1.1352,
+      "step": 5483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9941614127123956e-05,
+      "loss": 1.1133,
+      "step": 5484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9940782425171693e-05,
+      "loss": 1.1216,
+      "step": 5485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9939950574511967e-05,
+      "loss": 1.0803,
+      "step": 5486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9939118575160355e-05,
+      "loss": 1.1254,
+      "step": 5487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9938286427132434e-05,
+      "loss": 1.132,
+      "step": 5488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993745413044379e-05,
+      "loss": 1.0991,
+      "step": 5489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9936621685110008e-05,
+      "loss": 1.1073,
+      "step": 5490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993578909114668e-05,
+      "loss": 1.1257,
+      "step": 5491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934956348569397e-05,
+      "loss": 1.1193,
+      "step": 5492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9934123457393752e-05,
+      "loss": 1.0975,
+      "step": 5493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993329041763534e-05,
+      "loss": 1.0891,
+      "step": 5494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9932457229309774e-05,
+      "loss": 1.0925,
+      "step": 5495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9931623892432643e-05,
+      "loss": 1.0719,
+      "step": 5496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.993079040701956e-05,
+      "loss": 1.1299,
+      "step": 5497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9929956773086134e-05,
+      "loss": 1.1142,
+      "step": 5498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9929122990647972e-05,
+      "loss": 1.077,
+      "step": 5499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9928289059720692e-05,
+      "loss": 1.1175,
+      "step": 5500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992745498031991e-05,
+      "loss": 1.0922,
+      "step": 5501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9926620752461242e-05,
+      "loss": 1.1632,
+      "step": 5502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9925786376160322e-05,
+      "loss": 1.1007,
+      "step": 5503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924951851432764e-05,
+      "loss": 1.0779,
+      "step": 5504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9924117178294196e-05,
+      "loss": 1.0042,
+      "step": 5505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992328235676026e-05,
+      "loss": 1.0968,
+      "step": 5506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992244738684658e-05,
+      "loss": 1.0199,
+      "step": 5507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9921612268568793e-05,
+      "loss": 1.1066,
+      "step": 5508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.992077700194255e-05,
+      "loss": 1.0835,
+      "step": 5509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9919941586983473e-05,
+      "loss": 1.1671,
+      "step": 5510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9919106023707222e-05,
+      "loss": 1.1163,
+      "step": 5511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991827031212944e-05,
+      "loss": 1.0516,
+      "step": 5512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9917434452265775e-05,
+      "loss": 1.0822,
+      "step": 5513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9916598444131886e-05,
+      "loss": 1.0404,
+      "step": 5514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9915762287743427e-05,
+      "loss": 1.1125,
+      "step": 5515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9914925983116054e-05,
+      "loss": 1.0163,
+      "step": 5516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991408953026543e-05,
+      "loss": 1.025,
+      "step": 5517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991325292920722e-05,
+      "loss": 1.1347,
+      "step": 5518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991241617995709e-05,
+      "loss": 1.1136,
+      "step": 5519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.991157928253071e-05,
+      "loss": 1.0939,
+      "step": 5520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9910742236943752e-05,
+      "loss": 1.1592,
+      "step": 5521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9909905043211895e-05,
+      "loss": 1.1428,
+      "step": 5522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990906770135081e-05,
+      "loss": 1.1178,
+      "step": 5523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9908230211376187e-05,
+      "loss": 1.099,
+      "step": 5524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9907392573303703e-05,
+      "loss": 1.1028,
+      "step": 5525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9906554787149047e-05,
+      "loss": 1.1417,
+      "step": 5526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9905716852927906e-05,
+      "loss": 1.1286,
+      "step": 5527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.990487877065598e-05,
+      "loss": 1.1048,
+      "step": 5528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9904040540348952e-05,
+      "loss": 1.1652,
+      "step": 5529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9903202162022527e-05,
+      "loss": 1.0798,
+      "step": 5530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9902363635692406e-05,
+      "loss": 1.0887,
+      "step": 5531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9901524961374285e-05,
+      "loss": 1.0908,
+      "step": 5532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9900686139083883e-05,
+      "loss": 1.0625,
+      "step": 5533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9899847168836893e-05,
+      "loss": 1.1384,
+      "step": 5534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989900805064904e-05,
+      "loss": 1.1318,
+      "step": 5535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9898168784536025e-05,
+      "loss": 1.1254,
+      "step": 5536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9897329370513576e-05,
+      "loss": 1.141,
+      "step": 5537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9896489808597406e-05,
+      "loss": 1.0867,
+      "step": 5538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9895650098803246e-05,
+      "loss": 1.1242,
+      "step": 5539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989481024114681e-05,
+      "loss": 1.1082,
+      "step": 5540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989397023564383e-05,
+      "loss": 1.1029,
+      "step": 5541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893130082310044e-05,
+      "loss": 1.1268,
+      "step": 5542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892289781161183e-05,
+      "loss": 1.0639,
+      "step": 5543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989144933221297e-05,
+      "loss": 1.1973,
+      "step": 5544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890608735481165e-05,
+      "loss": 1.1147,
+      "step": 5545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889767990981498e-05,
+      "loss": 1.0873,
+      "step": 5546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988892709872971e-05,
+      "loss": 1.1144,
+      "step": 5547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888086058741554e-05,
+      "loss": 1.0453,
+      "step": 5548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887244871032784e-05,
+      "loss": 1.1179,
+      "step": 5549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886403535619148e-05,
+      "loss": 1.0669,
+      "step": 5550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98855620525164e-05,
+      "loss": 1.1093,
+      "step": 5551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98847204217403e-05,
+      "loss": 1.069,
+      "step": 5552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883878643306613e-05,
+      "loss": 1.083,
+      "step": 5553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883036717231102e-05,
+      "loss": 1.0638,
+      "step": 5554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988219464352953e-05,
+      "loss": 1.0983,
+      "step": 5555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988135242221767e-05,
+      "loss": 1.095,
+      "step": 5556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988051005331129e-05,
+      "loss": 1.1115,
+      "step": 5557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987966753682617e-05,
+      "loss": 1.0911,
+      "step": 5558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878824872778086e-05,
+      "loss": 1.1178,
+      "step": 5559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987798206118282e-05,
+      "loss": 1.1341,
+      "step": 5560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987713910205615e-05,
+      "loss": 1.0679,
+      "step": 5561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987629599541387e-05,
+      "loss": 1.0752,
+      "step": 5562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987545274127176e-05,
+      "loss": 1.1184,
+      "step": 5563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987460933964562e-05,
+      "loss": 1.0985,
+      "step": 5564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873765790551246e-05,
+      "loss": 1.0695,
+      "step": 5565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872922094004426e-05,
+      "loss": 1.1016,
+      "step": 5566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872078250020964e-05,
+      "loss": 1.0761,
+      "step": 5567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987123425861666e-05,
+      "loss": 1.0473,
+      "step": 5568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870390119807326e-05,
+      "loss": 1.1407,
+      "step": 5569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869545833608766e-05,
+      "loss": 1.1193,
+      "step": 5570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868701400036793e-05,
+      "loss": 1.1256,
+      "step": 5571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867856819107216e-05,
+      "loss": 1.1286,
+      "step": 5572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867012090835858e-05,
+      "loss": 1.0368,
+      "step": 5573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866167215238532e-05,
+      "loss": 1.0344,
+      "step": 5574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865322192331063e-05,
+      "loss": 1.1116,
+      "step": 5575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986447702212928e-05,
+      "loss": 1.1459,
+      "step": 5576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863631704649005e-05,
+      "loss": 1.1555,
+      "step": 5577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986278623990607e-05,
+      "loss": 1.1219,
+      "step": 5578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861940627916308e-05,
+      "loss": 1.0571,
+      "step": 5579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861094868695553e-05,
+      "loss": 1.0835,
+      "step": 5580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860248962259648e-05,
+      "loss": 1.0796,
+      "step": 5581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985940290862443e-05,
+      "loss": 0.9256,
+      "step": 5582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9858556707805745e-05,
+      "loss": 1.0771,
+      "step": 5583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985771035981944e-05,
+      "loss": 1.0866,
+      "step": 5584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985686386468136e-05,
+      "loss": 1.0758,
+      "step": 5585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985601722240737e-05,
+      "loss": 1.1093,
+      "step": 5586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985517043301331e-05,
+      "loss": 1.0993,
+      "step": 5587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9854323496515047e-05,
+      "loss": 1.1526,
+      "step": 5588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9853476412928436e-05,
+      "loss": 1.0761,
+      "step": 5589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985262918226935e-05,
+      "loss": 1.0787,
+      "step": 5590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9851781804553645e-05,
+      "loss": 1.0471,
+      "step": 5591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9850934279797195e-05,
+      "loss": 1.0996,
+      "step": 5592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.985008660801587e-05,
+      "loss": 1.1204,
+      "step": 5593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984923878922554e-05,
+      "loss": 1.2097,
+      "step": 5594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9848390823442088e-05,
+      "loss": 1.0855,
+      "step": 5595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9847542710681396e-05,
+      "loss": 1.0843,
+      "step": 5596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9846694450959343e-05,
+      "loss": 1.1147,
+      "step": 5597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9845846044291815e-05,
+      "loss": 1.1233,
+      "step": 5598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98449974906947e-05,
+      "loss": 1.1301,
+      "step": 5599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9844148790183887e-05,
+      "loss": 1.1385,
+      "step": 5600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9843299942775274e-05,
+      "loss": 1.1257,
+      "step": 5601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9842450948484748e-05,
+      "loss": 1.117,
+      "step": 5602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9841601807328223e-05,
+      "loss": 1.1453,
+      "step": 5603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.984075251932159e-05,
+      "loss": 1.0654,
+      "step": 5604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9839903084480757e-05,
+      "loss": 1.0734,
+      "step": 5605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983905350282163e-05,
+      "loss": 1.0718,
+      "step": 5606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9838203774360123e-05,
+      "loss": 1.1357,
+      "step": 5607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9837353899112144e-05,
+      "loss": 1.1537,
+      "step": 5608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983650387709361e-05,
+      "loss": 1.1324,
+      "step": 5609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9835653708320442e-05,
+      "loss": 1.0777,
+      "step": 5610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9834803392808555e-05,
+      "loss": 1.076,
+      "step": 5611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983395293057388e-05,
+      "loss": 1.1141,
+      "step": 5612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9833102321632342e-05,
+      "loss": 1.073,
+      "step": 5613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.983225156599987e-05,
+      "loss": 1.1773,
+      "step": 5614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9831400663692393e-05,
+      "loss": 1.0817,
+      "step": 5615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9830549614725854e-05,
+      "loss": 1.1023,
+      "step": 5616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9829698419116177e-05,
+      "loss": 1.0946,
+      "step": 5617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9828847076879313e-05,
+      "loss": 1.1225,
+      "step": 5618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9827995588031204e-05,
+      "loss": 1.1613,
+      "step": 5619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9827143952587794e-05,
+      "loss": 1.0757,
+      "step": 5620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9826292170565034e-05,
+      "loss": 1.094,
+      "step": 5621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982544024197887e-05,
+      "loss": 1.1473,
+      "step": 5622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982458816684526e-05,
+      "loss": 1.0716,
+      "step": 5623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982373594518016e-05,
+      "loss": 1.1737,
+      "step": 5624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.982288357699953e-05,
+      "loss": 1.0866,
+      "step": 5625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9822031062319333e-05,
+      "loss": 1.067,
+      "step": 5626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9821178401155535e-05,
+      "loss": 1.1956,
+      "step": 5627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98203255935241e-05,
+      "loss": 1.0062,
+      "step": 5628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9819472639441e-05,
+      "loss": 1.1025,
+      "step": 5629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981861953892221e-05,
+      "loss": 1.0701,
+      "step": 5630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9817766291983704e-05,
+      "loss": 1.0589,
+      "step": 5631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9816912898641465e-05,
+      "loss": 1.0744,
+      "step": 5632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981605935891147e-05,
+      "loss": 1.0926,
+      "step": 5633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9815205672809705e-05,
+      "loss": 1.146,
+      "step": 5634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.981435184035216e-05,
+      "loss": 1.1056,
+      "step": 5635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9813497861554817e-05,
+      "loss": 1.0996,
+      "step": 5636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9812643736433675e-05,
+      "loss": 1.133,
+      "step": 5637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9811789465004722e-05,
+      "loss": 1.0771,
+      "step": 5638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9810935047283967e-05,
+      "loss": 1.1252,
+      "step": 5639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98100804832874e-05,
+      "loss": 1.0176,
+      "step": 5640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9809225773031038e-05,
+      "loss": 1.123,
+      "step": 5641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9808370916530875e-05,
+      "loss": 1.1646,
+      "step": 5642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9807515913802922e-05,
+      "loss": 1.0897,
+      "step": 5643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9806660764863194e-05,
+      "loss": 1.0974,
+      "step": 5644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98058054697277e-05,
+      "loss": 1.0824,
+      "step": 5645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9804950028412464e-05,
+      "loss": 1.1173,
+      "step": 5646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9804094440933507e-05,
+      "loss": 1.1122,
+      "step": 5647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.980323870730684e-05,
+      "loss": 1.1165,
+      "step": 5648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98023828275485e-05,
+      "loss": 1.12,
+      "step": 5649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9801526801674507e-05,
+      "loss": 1.1708,
+      "step": 5650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9800670629700898e-05,
+      "loss": 1.1506,
+      "step": 5651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9799814311643707e-05,
+      "loss": 1.179,
+      "step": 5652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9798957847518965e-05,
+      "loss": 1.0964,
+      "step": 5653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9798101237342713e-05,
+      "loss": 1.1158,
+      "step": 5654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9797244481130995e-05,
+      "loss": 1.1405,
+      "step": 5655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9796387578899857e-05,
+      "loss": 1.1219,
+      "step": 5656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9795530530665338e-05,
+      "loss": 1.135,
+      "step": 5657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9794673336443497e-05,
+      "loss": 1.0903,
+      "step": 5658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979381599625038e-05,
+      "loss": 1.1188,
+      "step": 5659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9792958510102045e-05,
+      "loss": 1.1071,
+      "step": 5660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9792100878014556e-05,
+      "loss": 1.1046,
+      "step": 5661
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979124310000396e-05,
+      "loss": 1.0686,
+      "step": 5662
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.979038517608634e-05,
+      "loss": 1.1378,
+      "step": 5663
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9789527106277744e-05,
+      "loss": 1.0668,
+      "step": 5664
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9788668890594247e-05,
+      "loss": 1.0766,
+      "step": 5665
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9787810529051928e-05,
+      "loss": 1.1027,
+      "step": 5666
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9786952021666856e-05,
+      "loss": 1.0864,
+      "step": 5667
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9786093368455105e-05,
+      "loss": 1.0739,
+      "step": 5668
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9785234569432764e-05,
+      "loss": 1.1027,
+      "step": 5669
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9784375624615903e-05,
+      "loss": 1.0099,
+      "step": 5670
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9783516534020617e-05,
+      "loss": 1.0843,
+      "step": 5671
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9782657297662995e-05,
+      "loss": 1.0974,
+      "step": 5672
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9781797915559125e-05,
+      "loss": 1.0783,
+      "step": 5673
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9780938387725096e-05,
+      "loss": 1.1149,
+      "step": 5674
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9780078714177012e-05,
+      "loss": 1.1158,
+      "step": 5675
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977921889493097e-05,
+      "loss": 1.122,
+      "step": 5676
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9778358930003072e-05,
+      "loss": 1.0981,
+      "step": 5677
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977749881940942e-05,
+      "loss": 1.1049,
+      "step": 5678
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9776638563166122e-05,
+      "loss": 1.1559,
+      "step": 5679
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977577816128929e-05,
+      "loss": 1.1638,
+      "step": 5680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9774917613795038e-05,
+      "loss": 1.1046,
+      "step": 5681
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9774056920699478e-05,
+      "loss": 1.1611,
+      "step": 5682
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977319608201873e-05,
+      "loss": 1.1324,
+      "step": 5683
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9772335097768916e-05,
+      "loss": 1.0818,
+      "step": 5684
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9771473967966157e-05,
+      "loss": 1.1384,
+      "step": 5685
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.977061269262658e-05,
+      "loss": 1.1153,
+      "step": 5686
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9769751271766314e-05,
+      "loss": 1.0741,
+      "step": 5687
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9768889705401493e-05,
+      "loss": 1.0412,
+      "step": 5688
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9768027993548253e-05,
+      "loss": 1.0678,
+      "step": 5689
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976716613622273e-05,
+      "loss": 1.0533,
+      "step": 5690
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9766304133441057e-05,
+      "loss": 0.9926,
+      "step": 5691
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9765441985219387e-05,
+      "loss": 1.0725,
+      "step": 5692
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9764579691573856e-05,
+      "loss": 1.1328,
+      "step": 5693
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976371725252062e-05,
+      "loss": 1.119,
+      "step": 5694
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.976285466807583e-05,
+      "loss": 1.0779,
+      "step": 5695
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9761991938255635e-05,
+      "loss": 1.1208,
+      "step": 5696
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9761129063076194e-05,
+      "loss": 1.0713,
+      "step": 5697
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9760266042553663e-05,
+      "loss": 1.1367,
+      "step": 5698
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9759402876704208e-05,
+      "loss": 1.0849,
+      "step": 5699
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975853956554399e-05,
+      "loss": 1.0832,
+      "step": 5700
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975767610908918e-05,
+      "loss": 1.0441,
+      "step": 5701
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9756812507355947e-05,
+      "loss": 1.1326,
+      "step": 5702
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975594876036046e-05,
+      "loss": 1.1308,
+      "step": 5703
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97550848681189e-05,
+      "loss": 1.1161,
+      "step": 5704
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9754220830647443e-05,
+      "loss": 1.154,
+      "step": 5705
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9753356647962267e-05,
+      "loss": 1.0605,
+      "step": 5706
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9752492320079557e-05,
+      "loss": 1.1089,
+      "step": 5707
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.97516278470155e-05,
+      "loss": 1.0751,
+      "step": 5708
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.975076322878629e-05,
+      "loss": 1.106,
+      "step": 5709
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974989846540811e-05,
+      "loss": 1.0435,
+      "step": 5710
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974903355689716e-05,
+      "loss": 1.2337,
+      "step": 5711
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9748168503269634e-05,
+      "loss": 1.0836,
+      "step": 5712
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9747303304541736e-05,
+      "loss": 1.1085,
+      "step": 5713
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9746437960729664e-05,
+      "loss": 1.1385,
+      "step": 5714
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9745572471849623e-05,
+      "loss": 1.1503,
+      "step": 5715
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974470683791783e-05,
+      "loss": 1.1534,
+      "step": 5716
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9743841058950486e-05,
+      "loss": 1.1713,
+      "step": 5717
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9742975134963806e-05,
+      "loss": 1.1862,
+      "step": 5718
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.974210906597401e-05,
+      "loss": 1.1449,
+      "step": 5719
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9741242851997314e-05,
+      "loss": 1.1231,
+      "step": 5720
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9740376493049943e-05,
+      "loss": 1.0666,
+      "step": 5721
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9739509989148116e-05,
+      "loss": 1.0587,
+      "step": 5722
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9738643340308065e-05,
+      "loss": 1.0884,
+      "step": 5723
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9737776546546017e-05,
+      "loss": 1.0754,
+      "step": 5724
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9736909607878204e-05,
+      "loss": 1.0946,
+      "step": 5725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9736042524320862e-05,
+      "loss": 1.0917,
+      "step": 5726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9735175295890234e-05,
+      "loss": 1.0859,
+      "step": 5727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9734307922602554e-05,
+      "loss": 1.1701,
+      "step": 5728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9733440404474068e-05,
+      "loss": 1.0374,
+      "step": 5729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973257274152102e-05,
+      "loss": 1.0741,
+      "step": 5730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973170493375966e-05,
+      "loss": 1.0735,
+      "step": 5731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.973083698120624e-05,
+      "loss": 1.0742,
+      "step": 5732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9729968883877012e-05,
+      "loss": 1.0675,
+      "step": 5733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972910064178824e-05,
+      "loss": 1.1253,
+      "step": 5734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972823225495617e-05,
+      "loss": 1.0909,
+      "step": 5735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972736372339708e-05,
+      "loss": 1.0598,
+      "step": 5736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9726495047127224e-05,
+      "loss": 1.1128,
+      "step": 5737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9725626226162877e-05,
+      "loss": 1.1153,
+      "step": 5738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9724757260520304e-05,
+      "loss": 1.159,
+      "step": 5739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972388815021578e-05,
+      "loss": 1.0972,
+      "step": 5740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.972301889526558e-05,
+      "loss": 1.1453,
+      "step": 5741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9722149495685988e-05,
+      "loss": 1.1485,
+      "step": 5742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9721279951493275e-05,
+      "loss": 1.1582,
+      "step": 5743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9720410262703734e-05,
+      "loss": 1.1004,
+      "step": 5744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9719540429333648e-05,
+      "loss": 1.0708,
+      "step": 5745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971867045139931e-05,
+      "loss": 1.0464,
+      "step": 5746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9717800328917008e-05,
+      "loss": 1.1117,
+      "step": 5747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9716930061903038e-05,
+      "loss": 1.0739,
+      "step": 5748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9716059650373693e-05,
+      "loss": 1.1178,
+      "step": 5749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971518909434528e-05,
+      "loss": 1.0485,
+      "step": 5750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9714318393834103e-05,
+      "loss": 1.0594,
+      "step": 5751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.971344754885646e-05,
+      "loss": 1.1272,
+      "step": 5752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9712576559428666e-05,
+      "loss": 1.1708,
+      "step": 5753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9711705425567032e-05,
+      "loss": 1.0904,
+      "step": 5754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9710834147287866e-05,
+      "loss": 1.1746,
+      "step": 5755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970996272460749e-05,
+      "loss": 1.1336,
+      "step": 5756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970909115754222e-05,
+      "loss": 1.1277,
+      "step": 5757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970821944610838e-05,
+      "loss": 1.1355,
+      "step": 5758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970734759032229e-05,
+      "loss": 1.1221,
+      "step": 5759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9706475590200286e-05,
+      "loss": 1.0591,
+      "step": 5760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9705603445758693e-05,
+      "loss": 1.1501,
+      "step": 5761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970473115701384e-05,
+      "loss": 1.0975,
+      "step": 5762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970385872398207e-05,
+      "loss": 0.9625,
+      "step": 5763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9702986146679712e-05,
+      "loss": 1.1131,
+      "step": 5764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970211342512312e-05,
+      "loss": 1.0685,
+      "step": 5765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.970124055932862e-05,
+      "loss": 1.0494,
+      "step": 5766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9700367549312574e-05,
+      "loss": 1.122,
+      "step": 5767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9699494395091326e-05,
+      "loss": 1.0896,
+      "step": 5768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9698621096681224e-05,
+      "loss": 1.0576,
+      "step": 5769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969774765409862e-05,
+      "loss": 1.0396,
+      "step": 5770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9696874067359882e-05,
+      "loss": 1.0935,
+      "step": 5771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9696000336481357e-05,
+      "loss": 1.0993,
+      "step": 5772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9695126461479418e-05,
+      "loss": 1.1205,
+      "step": 5773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9694252442370425e-05,
+      "loss": 1.1437,
+      "step": 5774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969337827917075e-05,
+      "loss": 1.0373,
+      "step": 5775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9692503971896754e-05,
+      "loss": 1.0329,
+      "step": 5776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9691629520564822e-05,
+      "loss": 1.1283,
+      "step": 5777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.969075492519132e-05,
+      "loss": 1.0612,
+      "step": 5778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968988018579263e-05,
+      "loss": 1.0581,
+      "step": 5779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9689005302385136e-05,
+      "loss": 1.0493,
+      "step": 5780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9688130274985222e-05,
+      "loss": 1.0263,
+      "step": 5781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9687255103609267e-05,
+      "loss": 1.1322,
+      "step": 5782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968637978827367e-05,
+      "loss": 1.0951,
+      "step": 5783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968550432899482e-05,
+      "loss": 1.1509,
+      "step": 5784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9684628725789113e-05,
+      "loss": 1.1052,
+      "step": 5785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9683752978672938e-05,
+      "loss": 1.0666,
+      "step": 5786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9682877087662704e-05,
+      "loss": 1.1098,
+      "step": 5787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9682001052774813e-05,
+      "loss": 1.1305,
+      "step": 5788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968112487402567e-05,
+      "loss": 1.116,
+      "step": 5789
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.968024855143168e-05,
+      "loss": 1.0664,
+      "step": 5790
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9679372085009257e-05,
+      "loss": 1.043,
+      "step": 5791
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9678495474774812e-05,
+      "loss": 1.1557,
+      "step": 5792
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9677618720744765e-05,
+      "loss": 1.1636,
+      "step": 5793
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.967674182293553e-05,
+      "loss": 1.0579,
+      "step": 5794
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9675864781363534e-05,
+      "loss": 1.0939,
+      "step": 5795
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.96749875960452e-05,
+      "loss": 1.1408,
+      "step": 5796
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9674110266996958e-05,
+      "loss": 1.0958,
+      "step": 5797
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.967323279423523e-05,
+      "loss": 1.0193,
+      "step": 5798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9672355177776452e-05,
+      "loss": 1.1522,
+      "step": 5799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9671477417637062e-05,
+      "loss": 1.081,
+      "step": 5800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9670599513833494e-05,
+      "loss": 1.1175,
+      "step": 5801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9669721466382192e-05,
+      "loss": 1.1137,
+      "step": 5802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9668843275299594e-05,
+      "loss": 1.0505,
+      "step": 5803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9667964940602153e-05,
+      "loss": 1.0857,
+      "step": 5804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9667086462306315e-05,
+      "loss": 1.1326,
+      "step": 5805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9666207840428527e-05,
+      "loss": 1.0237,
+      "step": 5806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9665329074985247e-05,
+      "loss": 1.1292,
+      "step": 5807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9664450165992933e-05,
+      "loss": 1.0175,
+      "step": 5808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966357111346804e-05,
+      "loss": 1.1703,
+      "step": 5809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9662691917427033e-05,
+      "loss": 1.1387,
+      "step": 5810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9661812577886377e-05,
+      "loss": 1.0387,
+      "step": 5811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.966093309486254e-05,
+      "loss": 1.043,
+      "step": 5812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9660053468371987e-05,
+      "loss": 1.1761,
+      "step": 5813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9659173698431194e-05,
+      "loss": 1.0178,
+      "step": 5814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965829378505664e-05,
+      "loss": 1.0575,
+      "step": 5815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96574137282648e-05,
+      "loss": 1.0642,
+      "step": 5816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9656533528072155e-05,
+      "loss": 1.0459,
+      "step": 5817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965565318449519e-05,
+      "loss": 1.0679,
+      "step": 5818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.965477269755039e-05,
+      "loss": 1.058,
+      "step": 5819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9653892067254243e-05,
+      "loss": 1.0878,
+      "step": 5820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9653011293623243e-05,
+      "loss": 1.0451,
+      "step": 5821
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9652130376673878e-05,
+      "loss": 1.0705,
+      "step": 5822
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9651249316422656e-05,
+      "loss": 1.135,
+      "step": 5823
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9650368112886066e-05,
+      "loss": 1.0323,
+      "step": 5824
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9649486766080613e-05,
+      "loss": 1.0896,
+      "step": 5825
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964860527602281e-05,
+      "loss": 1.147,
+      "step": 5826
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9647723642729154e-05,
+      "loss": 1.0992,
+      "step": 5827
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964684186621616e-05,
+      "loss": 1.0761,
+      "step": 5828
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9645959946500344e-05,
+      "loss": 1.1415,
+      "step": 5829
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9645077883598218e-05,
+      "loss": 1.0172,
+      "step": 5830
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96441956775263e-05,
+      "loss": 1.0739,
+      "step": 5831
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964331332830111e-05,
+      "loss": 1.0939,
+      "step": 5832
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.964243083593918e-05,
+      "loss": 1.134,
+      "step": 5833
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9641548200457028e-05,
+      "loss": 1.0816,
+      "step": 5834
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9640665421871184e-05,
+      "loss": 1.1187,
+      "step": 5835
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9639782500198182e-05,
+      "loss": 1.084,
+      "step": 5836
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9638899435454556e-05,
+      "loss": 1.0938,
+      "step": 5837
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9638016227656846e-05,
+      "loss": 1.083,
+      "step": 5838
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9637132876821588e-05,
+      "loss": 1.0993,
+      "step": 5839
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9636249382965324e-05,
+      "loss": 1.05,
+      "step": 5840
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96353657461046e-05,
+      "loss": 1.1566,
+      "step": 5841
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963448196625597e-05,
+      "loss": 1.1187,
+      "step": 5842
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9633598043435977e-05,
+      "loss": 1.1,
+      "step": 5843
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9632713977661177e-05,
+      "loss": 1.1276,
+      "step": 5844
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9631829768948124e-05,
+      "loss": 1.046,
+      "step": 5845
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.963094541731338e-05,
+      "loss": 1.1263,
+      "step": 5846
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9630060922773503e-05,
+      "loss": 1.1105,
+      "step": 5847
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962917628534506e-05,
+      "loss": 1.0586,
+      "step": 5848
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9628291505044616e-05,
+      "loss": 1.2297,
+      "step": 5849
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9627406581888742e-05,
+      "loss": 1.1247,
+      "step": 5850
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962652151589401e-05,
+      "loss": 1.0232,
+      "step": 5851
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9625636307076988e-05,
+      "loss": 1.0214,
+      "step": 5852
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.962475095545426e-05,
+      "loss": 1.1334,
+      "step": 5853
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9623865461042406e-05,
+      "loss": 1.1659,
+      "step": 5854
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9622979823858008e-05,
+      "loss": 1.0772,
+      "step": 5855
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9622094043917652e-05,
+      "loss": 1.1443,
+      "step": 5856
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9621208121237923e-05,
+      "loss": 1.1525,
+      "step": 5857
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9620322055835412e-05,
+      "loss": 1.1726,
+      "step": 5858
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9619435847726714e-05,
+      "loss": 1.1069,
+      "step": 5859
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961854949692843e-05,
+      "loss": 1.0742,
+      "step": 5860
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961766300345715e-05,
+      "loss": 1.0825,
+      "step": 5861
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9616776367329477e-05,
+      "loss": 1.0706,
+      "step": 5862
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961588958856202e-05,
+      "loss": 1.1262,
+      "step": 5863
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9615002667171383e-05,
+      "loss": 1.1063,
+      "step": 5864
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9614115603174174e-05,
+      "loss": 1.1225,
+      "step": 5865
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961322839658701e-05,
+      "loss": 1.0773,
+      "step": 5866
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96123410474265e-05,
+      "loss": 1.1746,
+      "step": 5867
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9611453555709262e-05,
+      "loss": 1.0204,
+      "step": 5868
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.961056592145192e-05,
+      "loss": 1.079,
+      "step": 5869
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9609678144671096e-05,
+      "loss": 1.0981,
+      "step": 5870
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960879022538341e-05,
+      "loss": 1.0544,
+      "step": 5871
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9607902163605496e-05,
+      "loss": 1.1018,
+      "step": 5872
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9607013959353984e-05,
+      "loss": 1.1155,
+      "step": 5873
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9606125612645504e-05,
+      "loss": 1.065,
+      "step": 5874
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9605237123496697e-05,
+      "loss": 1.0661,
+      "step": 5875
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.96043484919242e-05,
+      "loss": 1.1122,
+      "step": 5876
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.960345971794465e-05,
+      "loss": 1.0911,
+      "step": 5877
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9602570801574697e-05,
+      "loss": 1.0819,
+      "step": 5878
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9601681742830985e-05,
+      "loss": 1.1371,
+      "step": 5879
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9600792541730166e-05,
+      "loss": 0.9927,
+      "step": 5880
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959990319828889e-05,
+      "loss": 1.0497,
+      "step": 5881
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959901371252381e-05,
+      "loss": 1.0867,
+      "step": 5882
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9598124084451584e-05,
+      "loss": 1.132,
+      "step": 5883
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9597234314088876e-05,
+      "loss": 1.0615,
+      "step": 5884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9596344401452347e-05,
+      "loss": 1.1021,
+      "step": 5885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959545434655866e-05,
+      "loss": 1.1408,
+      "step": 5886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9594564149424488e-05,
+      "loss": 1.103,
+      "step": 5887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9593673810066496e-05,
+      "loss": 1.0872,
+      "step": 5888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.959278332850136e-05,
+      "loss": 1.143,
+      "step": 5889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9591892704745757e-05,
+      "loss": 1.0634,
+      "step": 5890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9591001938816365e-05,
+      "loss": 1.0457,
+      "step": 5891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9590111030729865e-05,
+      "loss": 1.1458,
+      "step": 5892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958921998050294e-05,
+      "loss": 1.1588,
+      "step": 5893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9588328788152282e-05,
+      "loss": 1.0993,
+      "step": 5894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9587437453694576e-05,
+      "loss": 1.0859,
+      "step": 5895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958654597714651e-05,
+      "loss": 1.1072,
+      "step": 5896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9585654358524786e-05,
+      "loss": 1.0262,
+      "step": 5897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95847625978461e-05,
+      "loss": 1.1415,
+      "step": 5898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.958387069512715e-05,
+      "loss": 1.1379,
+      "step": 5899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9582978650384642e-05,
+      "loss": 1.0938,
+      "step": 5900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9582086463635276e-05,
+      "loss": 1.0977,
+      "step": 5901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9581194134895765e-05,
+      "loss": 1.0669,
+      "step": 5902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9580301664182816e-05,
+      "loss": 1.1041,
+      "step": 5903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957940905151314e-05,
+      "loss": 1.0607,
+      "step": 5904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9578516296903465e-05,
+      "loss": 1.1428,
+      "step": 5905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9577623400370495e-05,
+      "loss": 1.1062,
+      "step": 5906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957673036193096e-05,
+      "loss": 1.1735,
+      "step": 5907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9575837181601585e-05,
+      "loss": 1.0936,
+      "step": 5908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9574943859399092e-05,
+      "loss": 1.0927,
+      "step": 5909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957405039534021e-05,
+      "loss": 1.1411,
+      "step": 5910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9573156789441677e-05,
+      "loss": 1.1433,
+      "step": 5911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.957226304172022e-05,
+      "loss": 1.0923,
+      "step": 5912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9571369152192578e-05,
+      "loss": 1.1471,
+      "step": 5913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9570475120875495e-05,
+      "loss": 1.1384,
+      "step": 5914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956958094778571e-05,
+      "loss": 1.0749,
+      "step": 5915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9568686632939972e-05,
+      "loss": 1.0776,
+      "step": 5916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9567792176355025e-05,
+      "loss": 1.065,
+      "step": 5917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9566897578047622e-05,
+      "loss": 1.0414,
+      "step": 5918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9566002838034514e-05,
+      "loss": 1.1282,
+      "step": 5919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9565107956332455e-05,
+      "loss": 1.1678,
+      "step": 5920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956421293295821e-05,
+      "loss": 1.1025,
+      "step": 5921
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9563317767928536e-05,
+      "loss": 1.0922,
+      "step": 5922
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95624224612602e-05,
+      "loss": 1.0292,
+      "step": 5923
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.956152701296996e-05,
+      "loss": 1.0909,
+      "step": 5924
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9560631423074592e-05,
+      "loss": 1.1163,
+      "step": 5925
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955973569159087e-05,
+      "loss": 1.1,
+      "step": 5926
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9558839818535564e-05,
+      "loss": 1.069,
+      "step": 5927
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955794380392545e-05,
+      "loss": 1.0251,
+      "step": 5928
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9557047647777314e-05,
+      "loss": 1.0625,
+      "step": 5929
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955615135010793e-05,
+      "loss": 1.043,
+      "step": 5930
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.955525491093409e-05,
+      "loss": 1.0644,
+      "step": 5931
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9554358330272575e-05,
+      "loss": 1.1452,
+      "step": 5932
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9553461608140185e-05,
+      "loss": 1.0824,
+      "step": 5933
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9552564744553702e-05,
+      "loss": 1.0876,
+      "step": 5934
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9551667739529932e-05,
+      "loss": 1.0908,
+      "step": 5935
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9550770593085663e-05,
+      "loss": 1.132,
+      "step": 5936
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95498733052377e-05,
+      "loss": 1.0706,
+      "step": 5937
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548975876002853e-05,
+      "loss": 1.0936,
+      "step": 5938
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9548078305397922e-05,
+      "loss": 1.0561,
+      "step": 5939
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9547180593439713e-05,
+      "loss": 1.106,
+      "step": 5940
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9546282740145047e-05,
+      "loss": 1.1188,
+      "step": 5941
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9545384745530726e-05,
+      "loss": 1.1525,
+      "step": 5942
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9544486609613574e-05,
+      "loss": 1.1974,
+      "step": 5943
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.954358833241041e-05,
+      "loss": 1.0968,
+      "step": 5944
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9542689913938055e-05,
+      "loss": 1.1151,
+      "step": 5945
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9541791354213338e-05,
+      "loss": 1.1126,
+      "step": 5946
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9540892653253082e-05,
+      "loss": 1.1214,
+      "step": 5947
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9539993811074115e-05,
+      "loss": 1.0743,
+      "step": 5948
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9539094827693275e-05,
+      "loss": 1.0741,
+      "step": 5949
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953819570312739e-05,
+      "loss": 1.0002,
+      "step": 5950
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953729643739331e-05,
+      "loss": 1.0983,
+      "step": 5951
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953639703050786e-05,
+      "loss": 1.1206,
+      "step": 5952
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9535497482487898e-05,
+      "loss": 1.1354,
+      "step": 5953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953459779335026e-05,
+      "loss": 1.0976,
+      "step": 5954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9533697963111797e-05,
+      "loss": 1.1409,
+      "step": 5955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9532797991789366e-05,
+      "loss": 1.1153,
+      "step": 5956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.953189787939981e-05,
+      "loss": 1.0742,
+      "step": 5957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530997625959996e-05,
+      "loss": 1.068,
+      "step": 5958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9530097231486772e-05,
+      "loss": 1.0971,
+      "step": 5959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9529196695997013e-05,
+      "loss": 1.0614,
+      "step": 5960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9528296019507572e-05,
+      "loss": 1.0945,
+      "step": 5961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9527395202035327e-05,
+      "loss": 1.1099,
+      "step": 5962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.952649424359713e-05,
+      "loss": 1.1262,
+      "step": 5963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9525593144209876e-05,
+      "loss": 1.1143,
+      "step": 5964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9524691903890422e-05,
+      "loss": 1.1377,
+      "step": 5965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9523790522655653e-05,
+      "loss": 1.1158,
+      "step": 5966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9522889000522452e-05,
+      "loss": 1.1202,
+      "step": 5967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521987337507693e-05,
+      "loss": 1.0703,
+      "step": 5968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9521085533628267e-05,
+      "loss": 1.0695,
+      "step": 5969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9520183588901063e-05,
+      "loss": 1.1327,
+      "step": 5970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9519281503342972e-05,
+      "loss": 1.0799,
+      "step": 5971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9518379276970885e-05,
+      "loss": 1.1012,
+      "step": 5972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.95174769098017e-05,
+      "loss": 1.1517,
+      "step": 5973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951657440185231e-05,
+      "loss": 1.0947,
+      "step": 5974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9515671753139627e-05,
+      "loss": 1.1165,
+      "step": 5975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9514768963680545e-05,
+      "loss": 1.1048,
+      "step": 5976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9513866033491974e-05,
+      "loss": 1.0392,
+      "step": 5977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9512962962590824e-05,
+      "loss": 1.0423,
+      "step": 5978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951205975099401e-05,
+      "loss": 1.1074,
+      "step": 5979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.951115639871844e-05,
+      "loss": 1.089,
+      "step": 5980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9510252905781035e-05,
+      "loss": 1.1596,
+      "step": 5981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9509349272198714e-05,
+      "loss": 1.1305,
+      "step": 5982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9508445497988397e-05,
+      "loss": 1.1263,
+      "step": 5983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9507541583167013e-05,
+      "loss": 1.0695,
+      "step": 5984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950663752775149e-05,
+      "loss": 1.0933,
+      "step": 5985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9505733331758752e-05,
+      "loss": 1.1457,
+      "step": 5986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9504828995205742e-05,
+      "loss": 1.1061,
+      "step": 5987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9503924518109383e-05,
+      "loss": 1.1161,
+      "step": 5988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9503019900486626e-05,
+      "loss": 1.041,
+      "step": 5989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9502115142354405e-05,
+      "loss": 1.1417,
+      "step": 5990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.950121024372966e-05,
+      "loss": 1.0464,
+      "step": 5991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9500305204629344e-05,
+      "loss": 1.0928,
+      "step": 5992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.94994000250704e-05,
+      "loss": 1.1047,
+      "step": 5993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9498494705069788e-05,
+      "loss": 1.0608,
+      "step": 5994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9497589244644455e-05,
+      "loss": 1.1228,
+      "step": 5995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949668364381136e-05,
+      "loss": 1.0593,
+      "step": 5996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9495777902587457e-05,
+      "loss": 1.1153,
+      "step": 5997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9494872020989712e-05,
+      "loss": 1.1004,
+      "step": 5998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9493965999035094e-05,
+      "loss": 1.1146,
+      "step": 5999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9493059836740564e-05,
+      "loss": 1.0811,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "eval_text_loss": 0.9955954551696777,
+      "eval_text_runtime": 1.5746,
+      "eval_text_samples_per_second": 46.997,
+      "eval_text_steps_per_second": 1.27,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "eval_svit_loss": 1.1642725467681885,
+      "eval_svit_runtime": 4.2749,
+      "eval_svit_samples_per_second": 159.302,
+      "eval_svit_steps_per_second": 2.573,
+      "step": 6000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949215353412309e-05,
+      "loss": 1.0997,
+      "step": 6001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9491247091199654e-05,
+      "loss": 1.0992,
+      "step": 6002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.949034050798722e-05,
+      "loss": 1.147,
+      "step": 6003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9489433784502773e-05,
+      "loss": 1.1102,
+      "step": 6004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948852692076329e-05,
+      "loss": 1.0735,
+      "step": 6005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9487619916785753e-05,
+      "loss": 1.1805,
+      "step": 6006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948671277258715e-05,
+      "loss": 1.0886,
+      "step": 6007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948580548818447e-05,
+      "loss": 1.0774,
+      "step": 6008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.94848980635947e-05,
+      "loss": 1.1093,
+      "step": 6009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9483990498834838e-05,
+      "loss": 1.0961,
+      "step": 6010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9483082793921875e-05,
+      "loss": 1.105,
+      "step": 6011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948217494887281e-05,
+      "loss": 1.0647,
+      "step": 6012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.948126696370465e-05,
+      "loss": 1.1025,
+      "step": 6013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9480358838434396e-05,
+      "loss": 1.1407,
+      "step": 6014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9479450573079048e-05,
+      "loss": 1.1034,
+      "step": 6015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9478542167655628e-05,
+      "loss": 1.114,
+      "step": 6016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9477633622181138e-05,
+      "loss": 1.1394,
+      "step": 6017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9476724936672596e-05,
+      "loss": 1.0711,
+      "step": 6018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9475816111147017e-05,
+      "loss": 1.101,
+      "step": 6019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9474907145621418e-05,
+      "loss": 1.1145,
+      "step": 6020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.947399804011283e-05,
+      "loss": 1.0832,
+      "step": 6021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9473088794638273e-05,
+      "loss": 1.118,
+      "step": 6022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9472179409214772e-05,
+      "loss": 1.0912,
+      "step": 6023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9471269883859358e-05,
+      "loss": 1.049,
+      "step": 6024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9470360218589066e-05,
+      "loss": 1.1286,
+      "step": 6025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9469450413420932e-05,
+      "loss": 1.0975,
+      "step": 6026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9468540468371992e-05,
+      "loss": 1.0785,
+      "step": 6027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9467630383459285e-05,
+      "loss": 1.0903,
+      "step": 6028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946672015869986e-05,
+      "loss": 1.0722,
+      "step": 6029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9465809794110756e-05,
+      "loss": 1.0792,
+      "step": 6030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9464899289709027e-05,
+      "loss": 1.01,
+      "step": 6031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946398864551172e-05,
+      "loss": 1.1286,
+      "step": 6032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946307786153589e-05,
+      "loss": 1.1433,
+      "step": 6033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.946216693779859e-05,
+      "loss": 1.0295,
+      "step": 6034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9461255874316888e-05,
+      "loss": 0.9755,
+      "step": 6035
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.946034467110784e-05,
+      "loss": 1.0559,
+      "step": 6036
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9459433328188505e-05,
+      "loss": 1.0456,
+      "step": 6037
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945852184557596e-05,
+      "loss": 1.0833,
+      "step": 6038
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9457610223287265e-05,
+      "loss": 1.1348,
+      "step": 6039
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94566984613395e-05,
+      "loss": 1.0241,
+      "step": 6040
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9455786559749735e-05,
+      "loss": 1.0447,
+      "step": 6041
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945487451853505e-05,
+      "loss": 1.1282,
+      "step": 6042
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945396233771252e-05,
+      "loss": 1.1129,
+      "step": 6043
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945305001729923e-05,
+      "loss": 0.9895,
+      "step": 6044
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9452137557312267e-05,
+      "loss": 1.0393,
+      "step": 6045
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.945122495776872e-05,
+      "loss": 1.0644,
+      "step": 6046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9450312218685674e-05,
+      "loss": 1.0972,
+      "step": 6047
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9449399340080223e-05,
+      "loss": 1.0258,
+      "step": 6048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9448486321969465e-05,
+      "loss": 1.1126,
+      "step": 6049
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9447573164370497e-05,
+      "loss": 1.0765,
+      "step": 6050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944665986730042e-05,
+      "loss": 1.149,
+      "step": 6051
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9445746430776337e-05,
+      "loss": 1.07,
+      "step": 6052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9444832854815352e-05,
+      "loss": 1.0248,
+      "step": 6053
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.944391913943458e-05,
+      "loss": 1.0172,
+      "step": 6054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9443005284651128e-05,
+      "loss": 1.0888,
+      "step": 6055
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9442091290482106e-05,
+      "loss": 1.0863,
+      "step": 6056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9441177156944637e-05,
+      "loss": 1.1006,
+      "step": 6057
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9440262884055836e-05,
+      "loss": 1.0692,
+      "step": 6058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9439348471832825e-05,
+      "loss": 1.1005,
+      "step": 6059
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9438433920292732e-05,
+      "loss": 1.0292,
+      "step": 6060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9437519229452677e-05,
+      "loss": 1.1197,
+      "step": 6061
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9436604399329797e-05,
+      "loss": 1.1265,
+      "step": 6062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943568942994122e-05,
+      "loss": 1.0838,
+      "step": 6063
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943477432130408e-05,
+      "loss": 1.0811,
+      "step": 6064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9433859073435515e-05,
+      "loss": 1.0645,
+      "step": 6065
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9432943686352667e-05,
+      "loss": 1.0673,
+      "step": 6066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9432028160072676e-05,
+      "loss": 1.08,
+      "step": 6067
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.943111249461269e-05,
+      "loss": 1.0771,
+      "step": 6068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9430196689989852e-05,
+      "loss": 1.1131,
+      "step": 6069
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9429280746221318e-05,
+      "loss": 1.0808,
+      "step": 6070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9428364663324234e-05,
+      "loss": 1.1032,
+      "step": 6071
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942744844131576e-05,
+      "loss": 1.088,
+      "step": 6072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9426532080213056e-05,
+      "loss": 1.0603,
+      "step": 6073
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9425615580033282e-05,
+      "loss": 0.9901,
+      "step": 6074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9424698940793594e-05,
+      "loss": 1.0649,
+      "step": 6075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9423782162511168e-05,
+      "loss": 1.0499,
+      "step": 6076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9422865245203168e-05,
+      "loss": 1.0725,
+      "step": 6077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9421948188886764e-05,
+      "loss": 1.0463,
+      "step": 6078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9421030993579135e-05,
+      "loss": 1.0876,
+      "step": 6079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.942011365929745e-05,
+      "loss": 1.0901,
+      "step": 6080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941919618605889e-05,
+      "loss": 1.152,
+      "step": 6081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941827857388064e-05,
+      "loss": 1.1315,
+      "step": 6082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9417360822779882e-05,
+      "loss": 1.1278,
+      "step": 6083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9416442932773804e-05,
+      "loss": 1.0784,
+      "step": 6084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9415524903879595e-05,
+      "loss": 1.1597,
+      "step": 6085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9414606736114442e-05,
+      "loss": 1.0713,
+      "step": 6086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9413688429495548e-05,
+      "loss": 1.0389,
+      "step": 6087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9412769984040103e-05,
+      "loss": 1.0921,
+      "step": 6088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9411851399765313e-05,
+      "loss": 0.9963,
+      "step": 6089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.941093267668838e-05,
+      "loss": 1.1213,
+      "step": 6090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.94100138148265e-05,
+      "loss": 1.083,
+      "step": 6091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9409094814196892e-05,
+      "loss": 1.126,
+      "step": 6092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9408175674816756e-05,
+      "loss": 1.1209,
+      "step": 6093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940725639670331e-05,
+      "loss": 1.1117,
+      "step": 6094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9406336979873774e-05,
+      "loss": 1.0986,
+      "step": 6095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940541742434536e-05,
+      "loss": 1.121,
+      "step": 6096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9404497730135284e-05,
+      "loss": 1.0675,
+      "step": 6097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940357789726078e-05,
+      "loss": 1.0602,
+      "step": 6098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9402657925739068e-05,
+      "loss": 1.1261,
+      "step": 6099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9401737815587376e-05,
+      "loss": 1.1028,
+      "step": 6100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.940081756682294e-05,
+      "loss": 1.0898,
+      "step": 6101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9399897179462984e-05,
+      "loss": 1.1263,
+      "step": 6102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9398976653524753e-05,
+      "loss": 1.0659,
+      "step": 6103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9398055989025478e-05,
+      "loss": 1.0841,
+      "step": 6104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9397135185982408e-05,
+      "loss": 1.0939,
+      "step": 6105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9396214244412784e-05,
+      "loss": 1.0885,
+      "step": 6106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9395293164333853e-05,
+      "loss": 1.0166,
+      "step": 6107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9394371945762863e-05,
+      "loss": 1.1152,
+      "step": 6108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.939345058871706e-05,
+      "loss": 1.1313,
+      "step": 6109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9392529093213713e-05,
+      "loss": 1.0712,
+      "step": 6110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9391607459270068e-05,
+      "loss": 1.0755,
+      "step": 6111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9390685686903384e-05,
+      "loss": 1.0693,
+      "step": 6112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9389763776130923e-05,
+      "loss": 1.0412,
+      "step": 6113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9388841726969956e-05,
+      "loss": 1.0854,
+      "step": 6114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938791953943775e-05,
+      "loss": 1.1145,
+      "step": 6115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9386997213551566e-05,
+      "loss": 1.0526,
+      "step": 6116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9386074749328686e-05,
+      "loss": 1.06,
+      "step": 6117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938515214678638e-05,
+      "loss": 1.1825,
+      "step": 6118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9384229405941925e-05,
+      "loss": 1.0847,
+      "step": 6119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9383306526812603e-05,
+      "loss": 1.1253,
+      "step": 6120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9382383509415697e-05,
+      "loss": 1.0232,
+      "step": 6121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9381460353768492e-05,
+      "loss": 1.0707,
+      "step": 6122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.938053705988828e-05,
+      "loss": 1.0891,
+      "step": 6123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9379613627792342e-05,
+      "loss": 1.1203,
+      "step": 6124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9378690057497983e-05,
+      "loss": 1.0471,
+      "step": 6125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9377766349022487e-05,
+      "loss": 1.0665,
+      "step": 6126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9376842502383158e-05,
+      "loss": 1.0551,
+      "step": 6127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93759185175973e-05,
+      "loss": 1.0746,
+      "step": 6128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9374994394682212e-05,
+      "loss": 1.0721,
+      "step": 6129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93740701336552e-05,
+      "loss": 1.071,
+      "step": 6130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9373145734533574e-05,
+      "loss": 1.1808,
+      "step": 6131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9372221197334646e-05,
+      "loss": 1.0967,
+      "step": 6132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.937129652207573e-05,
+      "loss": 1.122,
+      "step": 6133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9370371708774145e-05,
+      "loss": 1.05,
+      "step": 6134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9369446757447203e-05,
+      "loss": 1.0865,
+      "step": 6135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9368521668112226e-05,
+      "loss": 1.1342,
+      "step": 6136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9367596440786547e-05,
+      "loss": 1.0817,
+      "step": 6137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9366671075487484e-05,
+      "loss": 1.151,
+      "step": 6138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9365745572232376e-05,
+      "loss": 1.156,
+      "step": 6139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936481993103854e-05,
+      "loss": 0.9873,
+      "step": 6140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9363894151923324e-05,
+      "loss": 1.1152,
+      "step": 6141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9362968234904063e-05,
+      "loss": 1.1026,
+      "step": 6142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936204217999809e-05,
+      "loss": 1.078,
+      "step": 6143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936111598722275e-05,
+      "loss": 1.0687,
+      "step": 6144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.936018965659539e-05,
+      "loss": 1.0182,
+      "step": 6145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9359263188133358e-05,
+      "loss": 1.0469,
+      "step": 6146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9358336581854002e-05,
+      "loss": 1.0643,
+      "step": 6147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9357409837774677e-05,
+      "loss": 1.1471,
+      "step": 6148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9356482955912735e-05,
+      "loss": 1.0242,
+      "step": 6149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9355555936285532e-05,
+      "loss": 1.0495,
+      "step": 6150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9354628778910434e-05,
+      "loss": 1.2219,
+      "step": 6151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9353701483804803e-05,
+      "loss": 1.0958,
+      "step": 6152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9352774050986e-05,
+      "loss": 1.0837,
+      "step": 6153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9351846480471402e-05,
+      "loss": 1.0816,
+      "step": 6154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9350918772278367e-05,
+      "loss": 1.1696,
+      "step": 6155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9349990926424273e-05,
+      "loss": 1.0538,
+      "step": 6156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9349062942926506e-05,
+      "loss": 1.135,
+      "step": 6157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.934813482180243e-05,
+      "loss": 1.0966,
+      "step": 6158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9347206563069434e-05,
+      "loss": 1.1353,
+      "step": 6159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93462781667449e-05,
+      "loss": 1.0128,
+      "step": 6160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9345349632846216e-05,
+      "loss": 1.1018,
+      "step": 6161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9344420961390766e-05,
+      "loss": 1.1656,
+      "step": 6162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9343492152395942e-05,
+      "loss": 1.1423,
+      "step": 6163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9342563205879146e-05,
+      "loss": 1.116,
+      "step": 6164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9341634121857762e-05,
+      "loss": 1.0137,
+      "step": 6165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9340704900349196e-05,
+      "loss": 1.1716,
+      "step": 6166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933977554137085e-05,
+      "loss": 1.0658,
+      "step": 6167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9338846044940127e-05,
+      "loss": 1.1379,
+      "step": 6168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9337916411074434e-05,
+      "loss": 1.1001,
+      "step": 6169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9336986639791178e-05,
+      "loss": 1.0716,
+      "step": 6170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9336056731107773e-05,
+      "loss": 1.1134,
+      "step": 6171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9335126685041635e-05,
+      "loss": 1.0784,
+      "step": 6172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9334196501610175e-05,
+      "loss": 1.0473,
+      "step": 6173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933326618083082e-05,
+      "loss": 1.0771,
+      "step": 6174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.933233572272099e-05,
+      "loss": 1.0967,
+      "step": 6175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9331405127298108e-05,
+      "loss": 1.0877,
+      "step": 6176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93304743945796e-05,
+      "loss": 1.1204,
+      "step": 6177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93295435245829e-05,
+      "loss": 1.1237,
+      "step": 6178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9328612517325436e-05,
+      "loss": 1.0913,
+      "step": 6179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9327681372824648e-05,
+      "loss": 1.0944,
+      "step": 6180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9326750091097973e-05,
+      "loss": 1.0415,
+      "step": 6181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9325818672162846e-05,
+      "loss": 1.0742,
+      "step": 6182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932488711603671e-05,
+      "loss": 1.1017,
+      "step": 6183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9323955422737017e-05,
+      "loss": 1.1297,
+      "step": 6184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932302359228121e-05,
+      "loss": 1.139,
+      "step": 6185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9322091624686742e-05,
+      "loss": 1.0961,
+      "step": 6186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932115951997106e-05,
+      "loss": 1.0526,
+      "step": 6187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.932022727815163e-05,
+      "loss": 1.0314,
+      "step": 6188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93192948992459e-05,
+      "loss": 1.0672,
+      "step": 6189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9318362383271337e-05,
+      "loss": 1.1018,
+      "step": 6190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9317429730245405e-05,
+      "loss": 1.0978,
+      "step": 6191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931649694018556e-05,
+      "loss": 1.0283,
+      "step": 6192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9315564013109285e-05,
+      "loss": 1.095,
+      "step": 6193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931463094903404e-05,
+      "loss": 1.0585,
+      "step": 6194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93136977479773e-05,
+      "loss": 1.0585,
+      "step": 6195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9312764409956544e-05,
+      "loss": 1.0198,
+      "step": 6196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.931183093498925e-05,
+      "loss": 1.0712,
+      "step": 6197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.93108973230929e-05,
+      "loss": 1.0786,
+      "step": 6198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9309963574284976e-05,
+      "loss": 0.9888,
+      "step": 6199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9309029688582963e-05,
+      "loss": 1.151,
+      "step": 6200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9308095666004356e-05,
+      "loss": 1.0596,
+      "step": 6201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9307161506566645e-05,
+      "loss": 1.0482,
+      "step": 6202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9306227210287315e-05,
+      "loss": 1.1077,
+      "step": 6203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9305292777183874e-05,
+      "loss": 1.088,
+      "step": 6204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9304358207273813e-05,
+      "loss": 1.1369,
+      "step": 6205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930342350057464e-05,
+      "loss": 1.0444,
+      "step": 6206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9302488657103855e-05,
+      "loss": 1.1099,
+      "step": 6207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9301553676878966e-05,
+      "loss": 1.1295,
+      "step": 6208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.930061855991748e-05,
+      "loss": 1.1054,
+      "step": 6209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9299683306236916e-05,
+      "loss": 1.1151,
+      "step": 6210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9298747915854785e-05,
+      "loss": 1.082,
+      "step": 6211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.92978123887886e-05,
+      "loss": 1.1251,
+      "step": 6212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929687672505589e-05,
+      "loss": 1.094,
+      "step": 6213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929594092467416e-05,
+      "loss": 1.0714,
+      "step": 6214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9295004987660957e-05,
+      "loss": 1.0997,
+      "step": 6215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9294068914033793e-05,
+      "loss": 1.1698,
+      "step": 6216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9293132703810204e-05,
+      "loss": 1.1058,
+      "step": 6217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9292196357007718e-05,
+      "loss": 1.101,
+      "step": 6218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9291259873643877e-05,
+      "loss": 1.1755,
+      "step": 6219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.929032325373621e-05,
+      "loss": 1.1107,
+      "step": 6220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9289386497302263e-05,
+      "loss": 1.1609,
+      "step": 6221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9288449604359577e-05,
+      "loss": 1.0862,
+      "step": 6222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.92875125749257e-05,
+      "loss": 1.1121,
+      "step": 6223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9286575409018174e-05,
+      "loss": 1.1759,
+      "step": 6224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9285638106654554e-05,
+      "loss": 1.0825,
+      "step": 6225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9284700667852392e-05,
+      "loss": 1.1193,
+      "step": 6226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9283763092629245e-05,
+      "loss": 1.1696,
+      "step": 6227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9282825381002666e-05,
+      "loss": 1.075,
+      "step": 6228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9281887532990217e-05,
+      "loss": 1.1021,
+      "step": 6229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9280949548609465e-05,
+      "loss": 1.072,
+      "step": 6230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9280011427877973e-05,
+      "loss": 1.1003,
+      "step": 6231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927907317081331e-05,
+      "loss": 1.0827,
+      "step": 6232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9278134777433047e-05,
+      "loss": 1.1097,
+      "step": 6233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9277196247754757e-05,
+      "loss": 1.0543,
+      "step": 6234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9276257581796014e-05,
+      "loss": 1.1367,
+      "step": 6235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9275318779574397e-05,
+      "loss": 1.0885,
+      "step": 6236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927437984110749e-05,
+      "loss": 1.0768,
+      "step": 6237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9273440766412873e-05,
+      "loss": 1.1949,
+      "step": 6238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.927250155550814e-05,
+      "loss": 1.0954,
+      "step": 6239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9271562208410864e-05,
+      "loss": 1.0617,
+      "step": 6240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9270622725138653e-05,
+      "loss": 1.0524,
+      "step": 6241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926968310570909e-05,
+      "loss": 1.1377,
+      "step": 6242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9268743350139776e-05,
+      "loss": 1.0859,
+      "step": 6243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926780345844831e-05,
+      "loss": 1.0917,
+      "step": 6244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9266863430652288e-05,
+      "loss": 1.1208,
+      "step": 6245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926592326676932e-05,
+      "loss": 1.0591,
+      "step": 6246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926498296681701e-05,
+      "loss": 1.0752,
+      "step": 6247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9264042530812968e-05,
+      "loss": 1.1166,
+      "step": 6248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9263101958774806e-05,
+      "loss": 1.0652,
+      "step": 6249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9262161250720133e-05,
+      "loss": 1.1149,
+      "step": 6250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9261220406666575e-05,
+      "loss": 1.1114,
+      "step": 6251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.926027942663174e-05,
+      "loss": 1.053,
+      "step": 6252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9259338310633265e-05,
+      "loss": 1.0983,
+      "step": 6253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.925839705868876e-05,
+      "loss": 1.1148,
+      "step": 6254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9257455670815857e-05,
+      "loss": 1.0998,
+      "step": 6255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9256514147032183e-05,
+      "loss": 1.0963,
+      "step": 6256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9255572487355374e-05,
+      "loss": 1.0737,
+      "step": 6257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9254630691803066e-05,
+      "loss": 1.0771,
+      "step": 6258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.925368876039289e-05,
+      "loss": 1.0833,
+      "step": 6259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9252746693142492e-05,
+      "loss": 1.0762,
+      "step": 6260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.925180449006951e-05,
+      "loss": 1.1075,
+      "step": 6261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9250862151191588e-05,
+      "loss": 1.1635,
+      "step": 6262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9249919676526372e-05,
+      "loss": 1.0744,
+      "step": 6263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924897706609152e-05,
+      "loss": 1.1068,
+      "step": 6264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924803431990468e-05,
+      "loss": 1.104,
+      "step": 6265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9247091437983498e-05,
+      "loss": 1.1815,
+      "step": 6266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9246148420345643e-05,
+      "loss": 1.1709,
+      "step": 6267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924520526700877e-05,
+      "loss": 1.0776,
+      "step": 6268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9244261977990544e-05,
+      "loss": 1.1315,
+      "step": 6269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9243318553308628e-05,
+      "loss": 1.1009,
+      "step": 6270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.924237499298069e-05,
+      "loss": 1.0848,
+      "step": 6271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9241431297024405e-05,
+      "loss": 1.1491,
+      "step": 6272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9240487465457436e-05,
+      "loss": 1.0868,
+      "step": 6273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9239543498297464e-05,
+      "loss": 1.0453,
+      "step": 6274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9238599395562163e-05,
+      "loss": 1.1273,
+      "step": 6275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9237655157269218e-05,
+      "loss": 1.0566,
+      "step": 6276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923671078343631e-05,
+      "loss": 1.0599,
+      "step": 6277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9235766274081125e-05,
+      "loss": 1.108,
+      "step": 6278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923482162922135e-05,
+      "loss": 1.0586,
+      "step": 6279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923387684887467e-05,
+      "loss": 1.1152,
+      "step": 6280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.923293193305879e-05,
+      "loss": 1.1546,
+      "step": 6281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9231986881791398e-05,
+      "loss": 1.1002,
+      "step": 6282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9231041695090193e-05,
+      "loss": 1.1142,
+      "step": 6283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9230096372972876e-05,
+      "loss": 1.114,
+      "step": 6284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9229150915457145e-05,
+      "loss": 1.1123,
+      "step": 6285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9228205322560714e-05,
+      "loss": 1.0889,
+      "step": 6286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9227259594301287e-05,
+      "loss": 1.0423,
+      "step": 6287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9226313730696575e-05,
+      "loss": 1.1608,
+      "step": 6288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9225367731764293e-05,
+      "loss": 1.1265,
+      "step": 6289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9224421597522153e-05,
+      "loss": 1.1198,
+      "step": 6290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9223475327987876e-05,
+      "loss": 1.0421,
+      "step": 6291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922252892317918e-05,
+      "loss": 1.0472,
+      "step": 6292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9221582383113795e-05,
+      "loss": 1.0674,
+      "step": 6293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.922063570780944e-05,
+      "loss": 1.0951,
+      "step": 6294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9219688897283846e-05,
+      "loss": 1.0775,
+      "step": 6295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9218741951554744e-05,
+      "loss": 1.0196,
+      "step": 6296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921779487063987e-05,
+      "loss": 1.0169,
+      "step": 6297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9216847654556952e-05,
+      "loss": 1.0462,
+      "step": 6298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921590030332374e-05,
+      "loss": 1.0694,
+      "step": 6299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9214952816957962e-05,
+      "loss": 1.1408,
+      "step": 6300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9214005195477376e-05,
+      "loss": 1.1199,
+      "step": 6301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9213057438899717e-05,
+      "loss": 1.0589,
+      "step": 6302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.921210954724274e-05,
+      "loss": 1.0916,
+      "step": 6303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9211161520524192e-05,
+      "loss": 1.0917,
+      "step": 6304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9210213358761828e-05,
+      "loss": 1.0813,
+      "step": 6305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9209265061973406e-05,
+      "loss": 1.0034,
+      "step": 6306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9208316630176684e-05,
+      "loss": 1.0244,
+      "step": 6307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9207368063389423e-05,
+      "loss": 1.0268,
+      "step": 6308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9206419361629385e-05,
+      "loss": 1.0746,
+      "step": 6309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9205470524914343e-05,
+      "loss": 1.1114,
+      "step": 6310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920452155326206e-05,
+      "loss": 1.0526,
+      "step": 6311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9203572446690306e-05,
+      "loss": 1.1491,
+      "step": 6312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920262320521686e-05,
+      "loss": 1.1423,
+      "step": 6313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.920167382885949e-05,
+      "loss": 1.0497,
+      "step": 6314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9200724317635986e-05,
+      "loss": 1.0411,
+      "step": 6315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9199774671564127e-05,
+      "loss": 1.024,
+      "step": 6316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9198824890661692e-05,
+      "loss": 1.0634,
+      "step": 6317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919787497494647e-05,
+      "loss": 1.0364,
+      "step": 6318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919692492443625e-05,
+      "loss": 1.0764,
+      "step": 6319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9195974739148827e-05,
+      "loss": 1.1164,
+      "step": 6320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919502441910199e-05,
+      "loss": 1.1033,
+      "step": 6321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919407396431354e-05,
+      "loss": 1.1127,
+      "step": 6322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919312337480127e-05,
+      "loss": 1.0314,
+      "step": 6323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.919217265058299e-05,
+      "loss": 1.0497,
+      "step": 6324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91912217916765e-05,
+      "loss": 1.1489,
+      "step": 6325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9190270798099603e-05,
+      "loss": 1.096,
+      "step": 6326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9189319669870116e-05,
+      "loss": 1.0485,
+      "step": 6327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9188368407005846e-05,
+      "loss": 1.1634,
+      "step": 6328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918741700952461e-05,
+      "loss": 1.1534,
+      "step": 6329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9186465477444223e-05,
+      "loss": 1.0774,
+      "step": 6330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9185513810782502e-05,
+      "loss": 1.0047,
+      "step": 6331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9184562009557275e-05,
+      "loss": 1.122,
+      "step": 6332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918361007378636e-05,
+      "loss": 1.0581,
+      "step": 6333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918265800348759e-05,
+      "loss": 1.0951,
+      "step": 6334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.918170579867879e-05,
+      "loss": 1.0522,
+      "step": 6335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9180753459377795e-05,
+      "loss": 1.0263,
+      "step": 6336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9179800985602437e-05,
+      "loss": 1.1033,
+      "step": 6337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9178848377370554e-05,
+      "loss": 1.0811,
+      "step": 6338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9177895634699983e-05,
+      "loss": 1.0923,
+      "step": 6339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9176942757608573e-05,
+      "loss": 1.0954,
+      "step": 6340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.917598974611416e-05,
+      "loss": 1.1193,
+      "step": 6341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9175036600234598e-05,
+      "loss": 1.0652,
+      "step": 6342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9174083319987733e-05,
+      "loss": 1.0592,
+      "step": 6343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9173129905391424e-05,
+      "loss": 1.0533,
+      "step": 6344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9172176356463513e-05,
+      "loss": 1.0798,
+      "step": 6345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.917122267322186e-05,
+      "loss": 1.099,
+      "step": 6346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9170268855684336e-05,
+      "loss": 1.1282,
+      "step": 6347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9169314903868795e-05,
+      "loss": 1.0246,
+      "step": 6348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9168360817793104e-05,
+      "loss": 1.1277,
+      "step": 6349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9167406597475125e-05,
+      "loss": 1.0793,
+      "step": 6350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9166452242932734e-05,
+      "loss": 0.9997,
+      "step": 6351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9165497754183798e-05,
+      "loss": 1.07,
+      "step": 6352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9164543131246197e-05,
+      "loss": 1.0859,
+      "step": 6353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91635883741378e-05,
+      "loss": 1.0169,
+      "step": 6354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9162633482876497e-05,
+      "loss": 1.0077,
+      "step": 6355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9161678457480167e-05,
+      "loss": 1.1112,
+      "step": 6356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9160723297966693e-05,
+      "loss": 1.166,
+      "step": 6357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9159768004353964e-05,
+      "loss": 1.0207,
+      "step": 6358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9158812576659864e-05,
+      "loss": 1.0626,
+      "step": 6359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9157857014902295e-05,
+      "loss": 1.0933,
+      "step": 6360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9156901319099147e-05,
+      "loss": 1.093,
+      "step": 6361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9155945489268315e-05,
+      "loss": 1.0863,
+      "step": 6362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91549895254277e-05,
+      "loss": 1.1086,
+      "step": 6363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.915403342759521e-05,
+      "loss": 1.0931,
+      "step": 6364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9153077195788744e-05,
+      "loss": 1.0643,
+      "step": 6365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9152120830026207e-05,
+      "loss": 1.1085,
+      "step": 6366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9151164330325514e-05,
+      "loss": 1.0572,
+      "step": 6367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.915020769670458e-05,
+      "loss": 1.0689,
+      "step": 6368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9149250929181312e-05,
+      "loss": 1.1014,
+      "step": 6369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9148294027773636e-05,
+      "loss": 1.0782,
+      "step": 6370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9147336992499466e-05,
+      "loss": 1.0848,
+      "step": 6371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9146379823376726e-05,
+      "loss": 1.1676,
+      "step": 6372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9145422520423343e-05,
+      "loss": 1.1049,
+      "step": 6373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9144465083657236e-05,
+      "loss": 1.1058,
+      "step": 6374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9143507513096347e-05,
+      "loss": 1.0681,
+      "step": 6375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91425498087586e-05,
+      "loss": 1.0729,
+      "step": 6376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9141591970661934e-05,
+      "loss": 1.0983,
+      "step": 6377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9140633998824285e-05,
+      "loss": 1.0476,
+      "step": 6378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9139675893263596e-05,
+      "loss": 1.0604,
+      "step": 6379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9138717653997805e-05,
+      "loss": 1.0853,
+      "step": 6380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9137759281044857e-05,
+      "loss": 1.093,
+      "step": 6381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9136800774422704e-05,
+      "loss": 1.0508,
+      "step": 6382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9135842134149292e-05,
+      "loss": 1.1115,
+      "step": 6383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9134883360242576e-05,
+      "loss": 1.1259,
+      "step": 6384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913392445272051e-05,
+      "loss": 1.097,
+      "step": 6385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913296541160105e-05,
+      "loss": 0.9937,
+      "step": 6386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9132006236902158e-05,
+      "loss": 1.0765,
+      "step": 6387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9131046928641796e-05,
+      "loss": 1.1635,
+      "step": 6388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.913008748683793e-05,
+      "loss": 1.0499,
+      "step": 6389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9129127911508523e-05,
+      "loss": 1.1176,
+      "step": 6390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912816820267155e-05,
+      "loss": 1.1055,
+      "step": 6391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9127208360344984e-05,
+      "loss": 1.052,
+      "step": 6392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9126248384546796e-05,
+      "loss": 1.0805,
+      "step": 6393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9125288275294966e-05,
+      "loss": 1.1164,
+      "step": 6394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9124328032607474e-05,
+      "loss": 1.0488,
+      "step": 6395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91233676565023e-05,
+      "loss": 1.0294,
+      "step": 6396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9122407146997436e-05,
+      "loss": 1.1265,
+      "step": 6397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.912144650411086e-05,
+      "loss": 1.1693,
+      "step": 6398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9120485727860568e-05,
+      "loss": 1.1011,
+      "step": 6399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9119524818264555e-05,
+      "loss": 1.0348,
+      "step": 6400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9118563775340806e-05,
+      "loss": 1.1785,
+      "step": 6401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9117602599107325e-05,
+      "loss": 1.0763,
+      "step": 6402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9116641289582118e-05,
+      "loss": 1.1042,
+      "step": 6403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9115679846783174e-05,
+      "loss": 1.1139,
+      "step": 6404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911471827072851e-05,
+      "loss": 1.1366,
+      "step": 6405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9113756561436127e-05,
+      "loss": 1.0991,
+      "step": 6406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9112794718924037e-05,
+      "loss": 1.1536,
+      "step": 6407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.911183274321025e-05,
+      "loss": 1.1301,
+      "step": 6408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9110870634312784e-05,
+      "loss": 1.1212,
+      "step": 6409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9109908392249655e-05,
+      "loss": 1.0883,
+      "step": 6410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9108946017038886e-05,
+      "loss": 1.1078,
+      "step": 6411
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9107983508698495e-05,
+      "loss": 1.1229,
+      "step": 6412
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910702086724651e-05,
+      "loss": 1.0671,
+      "step": 6413
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9106058092700958e-05,
+      "loss": 1.1791,
+      "step": 6414
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9105095185079868e-05,
+      "loss": 1.1593,
+      "step": 6415
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910413214440127e-05,
+      "loss": 1.0848,
+      "step": 6416
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91031689706832e-05,
+      "loss": 1.064,
+      "step": 6417
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.91022056639437e-05,
+      "loss": 1.0888,
+      "step": 6418
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.910124222420081e-05,
+      "loss": 1.1183,
+      "step": 6419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9100278651472566e-05,
+      "loss": 1.1328,
+      "step": 6420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9099314945777013e-05,
+      "loss": 1.1462,
+      "step": 6421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9098351107132205e-05,
+      "loss": 1.1623,
+      "step": 6422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909738713555619e-05,
+      "loss": 1.09,
+      "step": 6423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909642303106702e-05,
+      "loss": 1.094,
+      "step": 6424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9095458793682742e-05,
+      "loss": 1.0807,
+      "step": 6425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9094494423421425e-05,
+      "loss": 1.1371,
+      "step": 6426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909352992030112e-05,
+      "loss": 1.0969,
+      "step": 6427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9092565284339895e-05,
+      "loss": 1.1136,
+      "step": 6428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9091600515555818e-05,
+      "loss": 1.0529,
+      "step": 6429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.909063561396694e-05,
+      "loss": 1.0419,
+      "step": 6430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908967057959135e-05,
+      "loss": 1.1162,
+      "step": 6431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908870541244711e-05,
+      "loss": 1.0945,
+      "step": 6432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9087740112552298e-05,
+      "loss": 1.1357,
+      "step": 6433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908677467992499e-05,
+      "loss": 1.0089,
+      "step": 6434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9085809114583264e-05,
+      "loss": 1.0623,
+      "step": 6435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9084843416545204e-05,
+      "loss": 1.0901,
+      "step": 6436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9083877585828897e-05,
+      "loss": 1.0832,
+      "step": 6437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9082911622452424e-05,
+      "loss": 1.1208,
+      "step": 6438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.908194552643388e-05,
+      "loss": 1.0732,
+      "step": 6439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9080979297791358e-05,
+      "loss": 1.0551,
+      "step": 6440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9080012936542947e-05,
+      "loss": 1.0528,
+      "step": 6441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9079046442706748e-05,
+      "loss": 1.085,
+      "step": 6442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9078079816300863e-05,
+      "loss": 1.1011,
+      "step": 6443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9077113057343388e-05,
+      "loss": 1.0993,
+      "step": 6444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907614616585243e-05,
+      "loss": 1.1073,
+      "step": 6445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9075179141846098e-05,
+      "loss": 1.1222,
+      "step": 6446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90742119853425e-05,
+      "loss": 1.1139,
+      "step": 6447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9073244696359748e-05,
+      "loss": 1.1135,
+      "step": 6448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907227727491595e-05,
+      "loss": 1.0867,
+      "step": 6449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.907130972102924e-05,
+      "loss": 1.0995,
+      "step": 6450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9070342034717723e-05,
+      "loss": 1.0643,
+      "step": 6451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9069374215999524e-05,
+      "loss": 1.0763,
+      "step": 6452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9068406264892768e-05,
+      "loss": 1.063,
+      "step": 6453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9067438181415582e-05,
+      "loss": 1.1401,
+      "step": 6454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9066469965586097e-05,
+      "loss": 1.0773,
+      "step": 6455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9065501617422443e-05,
+      "loss": 1.1031,
+      "step": 6456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906453313694275e-05,
+      "loss": 1.0413,
+      "step": 6457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9063564524165163e-05,
+      "loss": 1.0548,
+      "step": 6458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906259577910782e-05,
+      "loss": 1.099,
+      "step": 6459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906162690178886e-05,
+      "loss": 1.1508,
+      "step": 6460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.906065789222642e-05,
+      "loss": 1.0609,
+      "step": 6461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9059688750438666e-05,
+      "loss": 1.0392,
+      "step": 6462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9058719476443727e-05,
+      "loss": 1.1388,
+      "step": 6463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9057750070259766e-05,
+      "loss": 1.1105,
+      "step": 6464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9056780531904933e-05,
+      "loss": 1.1389,
+      "step": 6465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9055810861397388e-05,
+      "loss": 1.0852,
+      "step": 6466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9054841058755285e-05,
+      "loss": 1.1091,
+      "step": 6467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9053871123996792e-05,
+      "loss": 1.1481,
+      "step": 6468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9052901057140065e-05,
+      "loss": 1.0593,
+      "step": 6469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9051930858203277e-05,
+      "loss": 1.0398,
+      "step": 6470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9050960527204594e-05,
+      "loss": 1.144,
+      "step": 6471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.904999006416219e-05,
+      "loss": 0.9886,
+      "step": 6472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9049019469094238e-05,
+      "loss": 1.1663,
+      "step": 6473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9048048742018913e-05,
+      "loss": 1.082,
+      "step": 6474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9047077882954395e-05,
+      "loss": 1.0432,
+      "step": 6475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9046106891918863e-05,
+      "loss": 1.0922,
+      "step": 6476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9045135768930503e-05,
+      "loss": 1.0696,
+      "step": 6477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90441645140075e-05,
+      "loss": 1.1819,
+      "step": 6478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9043193127168046e-05,
+      "loss": 1.0882,
+      "step": 6479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9042221608430332e-05,
+      "loss": 1.1158,
+      "step": 6480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9041249957812545e-05,
+      "loss": 1.0412,
+      "step": 6481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9040278175332887e-05,
+      "loss": 1.0722,
+      "step": 6482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9039306261009556e-05,
+      "loss": 1.1089,
+      "step": 6483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9038334214860752e-05,
+      "loss": 1.0216,
+      "step": 6484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9037362036904678e-05,
+      "loss": 1.1156,
+      "step": 6485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9036389727159543e-05,
+      "loss": 1.0571,
+      "step": 6486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9035417285643553e-05,
+      "loss": 1.0318,
+      "step": 6487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9034444712374918e-05,
+      "loss": 1.0415,
+      "step": 6488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9033472007371853e-05,
+      "loss": 1.1056,
+      "step": 6489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9032499170652573e-05,
+      "loss": 1.1325,
+      "step": 6490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9031526202235303e-05,
+      "loss": 1.0087,
+      "step": 6491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.903055310213825e-05,
+      "loss": 1.1036,
+      "step": 6492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9029579870379653e-05,
+      "loss": 1.1525,
+      "step": 6493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9028606506977726e-05,
+      "loss": 1.073,
+      "step": 6494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.90276330119507e-05,
+      "loss": 1.1175,
+      "step": 6495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9026659385316812e-05,
+      "loss": 1.1135,
+      "step": 6496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9025685627094285e-05,
+      "loss": 1.1311,
+      "step": 6497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902471173730136e-05,
+      "loss": 1.1005,
+      "step": 6498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.902373771595628e-05,
+      "loss": 1.0976,
+      "step": 6499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9022763563077273e-05,
+      "loss": 1.0687,
+      "step": 6500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9021789278682596e-05,
+      "loss": 1.1037,
+      "step": 6501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9020814862790485e-05,
+      "loss": 1.0531,
+      "step": 6502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901984031541919e-05,
+      "loss": 1.1082,
+      "step": 6503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9018865636586964e-05,
+      "loss": 1.0623,
+      "step": 6504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.901789082631206e-05,
+      "loss": 1.0896,
+      "step": 6505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9016915884612728e-05,
+      "loss": 1.103,
+      "step": 6506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9015940811507228e-05,
+      "loss": 1.0901,
+      "step": 6507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9014965607013826e-05,
+      "loss": 1.105,
+      "step": 6508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9013990271150778e-05,
+      "loss": 1.0629,
+      "step": 6509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9013014803936352e-05,
+      "loss": 1.1148,
+      "step": 6510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9012039205388814e-05,
+      "loss": 1.1174,
+      "step": 6511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9011063475526435e-05,
+      "loss": 1.0732,
+      "step": 6512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9010087614367488e-05,
+      "loss": 1.0712,
+      "step": 6513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9009111621930246e-05,
+      "loss": 1.1292,
+      "step": 6514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9008135498232993e-05,
+      "loss": 1.1169,
+      "step": 6515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9007159243294003e-05,
+      "loss": 1.093,
+      "step": 6516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9006182857131554e-05,
+      "loss": 1.0985,
+      "step": 6517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9005206339763938e-05,
+      "loss": 1.0626,
+      "step": 6518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9004229691209442e-05,
+      "loss": 1.0849,
+      "step": 6519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9003252911486354e-05,
+      "loss": 1.0848,
+      "step": 6520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9002276000612968e-05,
+      "loss": 1.0587,
+      "step": 6521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9001298958607575e-05,
+      "loss": 1.0916,
+      "step": 6522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9000321785488478e-05,
+      "loss": 1.0656,
+      "step": 6523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8999344481273966e-05,
+      "loss": 1.1299,
+      "step": 6524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8998367045982353e-05,
+      "loss": 1.084,
+      "step": 6525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899738947963194e-05,
+      "loss": 1.1081,
+      "step": 6526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899641178224103e-05,
+      "loss": 1.1227,
+      "step": 6527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8995433953827933e-05,
+      "loss": 1.0695,
+      "step": 6528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8994455994410964e-05,
+      "loss": 1.1307,
+      "step": 6529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8993477904008433e-05,
+      "loss": 1.1108,
+      "step": 6530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8992499682638664e-05,
+      "loss": 1.096,
+      "step": 6531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.899152133031997e-05,
+      "loss": 1.0661,
+      "step": 6532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8990542847070673e-05,
+      "loss": 1.0694,
+      "step": 6533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89895642329091e-05,
+      "loss": 1.1527,
+      "step": 6534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8988585487853578e-05,
+      "loss": 1.0866,
+      "step": 6535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898760661192243e-05,
+      "loss": 1.0944,
+      "step": 6536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8986627605133994e-05,
+      "loss": 1.0878,
+      "step": 6537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89856484675066e-05,
+      "loss": 1.1135,
+      "step": 6538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898466919905859e-05,
+      "loss": 1.1017,
+      "step": 6539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8983689799808295e-05,
+      "loss": 1.1711,
+      "step": 6540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898271026977406e-05,
+      "loss": 1.0116,
+      "step": 6541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898173060897423e-05,
+      "loss": 1.1106,
+      "step": 6542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.898075081742715e-05,
+      "loss": 1.1444,
+      "step": 6543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8979770895151167e-05,
+      "loss": 1.0971,
+      "step": 6544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8978790842164633e-05,
+      "loss": 1.0717,
+      "step": 6545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8977810658485904e-05,
+      "loss": 1.1047,
+      "step": 6546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897683034413333e-05,
+      "loss": 1.1001,
+      "step": 6547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8975849899125277e-05,
+      "loss": 1.0771,
+      "step": 6548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8974869323480103e-05,
+      "loss": 1.031,
+      "step": 6549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897388861721617e-05,
+      "loss": 1.0847,
+      "step": 6550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8972907780351842e-05,
+      "loss": 1.0751,
+      "step": 6551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.897192681290549e-05,
+      "loss": 1.1491,
+      "step": 6552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8970945714895487e-05,
+      "loss": 1.0631,
+      "step": 6553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89699644863402e-05,
+      "loss": 1.1489,
+      "step": 6554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8968983127258007e-05,
+      "loss": 1.1244,
+      "step": 6555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.896800163766729e-05,
+      "loss": 1.123,
+      "step": 6556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8967020017586425e-05,
+      "loss": 1.0819,
+      "step": 6557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8966038267033796e-05,
+      "loss": 1.0852,
+      "step": 6558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8965056386027787e-05,
+      "loss": 1.1089,
+      "step": 6559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8964074374586784e-05,
+      "loss": 1.1546,
+      "step": 6560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8963092232729183e-05,
+      "loss": 1.0707,
+      "step": 6561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8962109960473377e-05,
+      "loss": 1.063,
+      "step": 6562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8961127557837753e-05,
+      "loss": 1.0973,
+      "step": 6563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8960145024840713e-05,
+      "loss": 1.0901,
+      "step": 6564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895916236150066e-05,
+      "loss": 1.1317,
+      "step": 6565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895817956783599e-05,
+      "loss": 1.1342,
+      "step": 6566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8957196643865116e-05,
+      "loss": 1.0377,
+      "step": 6567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8956213589606437e-05,
+      "loss": 1.0329,
+      "step": 6568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895523040507837e-05,
+      "loss": 1.0659,
+      "step": 6569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8954247090299324e-05,
+      "loss": 1.0873,
+      "step": 6570
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.895326364528771e-05,
+      "loss": 1.0664,
+      "step": 6571
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8952280070061947e-05,
+      "loss": 1.0885,
+      "step": 6572
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8951296364640462e-05,
+      "loss": 1.1216,
+      "step": 6573
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8950312529041664e-05,
+      "loss": 1.0511,
+      "step": 6574
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894932856328399e-05,
+      "loss": 1.0545,
+      "step": 6575
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8948344467385855e-05,
+      "loss": 1.0266,
+      "step": 6576
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8947360241365696e-05,
+      "loss": 1.012,
+      "step": 6577
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8946375885241943e-05,
+      "loss": 1.0829,
+      "step": 6578
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894539139903303e-05,
+      "loss": 1.1182,
+      "step": 6579
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.894440678275739e-05,
+      "loss": 1.0212,
+      "step": 6580
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8943422036433468e-05,
+      "loss": 1.0873,
+      "step": 6581
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89424371600797e-05,
+      "loss": 1.0877,
+      "step": 6582
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8941452153714535e-05,
+      "loss": 1.1007,
+      "step": 6583
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8940467017356416e-05,
+      "loss": 1.1165,
+      "step": 6584
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8939481751023794e-05,
+      "loss": 1.0478,
+      "step": 6585
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8938496354735115e-05,
+      "loss": 1.1274,
+      "step": 6586
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893751082850884e-05,
+      "loss": 1.1413,
+      "step": 6587
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8936525172363418e-05,
+      "loss": 1.0583,
+      "step": 6588
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8935539386317313e-05,
+      "loss": 1.1,
+      "step": 6589
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893455347038898e-05,
+      "loss": 1.0577,
+      "step": 6590
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.893356742459689e-05,
+      "loss": 1.1237,
+      "step": 6591
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.89325812489595e-05,
+      "loss": 1.0414,
+      "step": 6592
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8931594943495286e-05,
+      "loss": 1.0937,
+      "step": 6593
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8930608508222713e-05,
+      "loss": 1.0733,
+      "step": 6594
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8929621943160257e-05,
+      "loss": 1.0644,
+      "step": 6595
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892863524832639e-05,
+      "loss": 1.1093,
+      "step": 6596
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8927648423739595e-05,
+      "loss": 1.1209,
+      "step": 6597
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8926661469418354e-05,
+      "loss": 1.0945,
+      "step": 6598
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892567438538114e-05,
+      "loss": 1.1093,
+      "step": 6599
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8924687171646446e-05,
+      "loss": 1.2142,
+      "step": 6600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8923699828232757e-05,
+      "loss": 1.1093,
+      "step": 6601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8922712355158564e-05,
+      "loss": 1.069,
+      "step": 6602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8921724752442357e-05,
+      "loss": 1.0873,
+      "step": 6603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.892073702010263e-05,
+      "loss": 1.1216,
+      "step": 6604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891974915815789e-05,
+      "loss": 1.1222,
+      "step": 6605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8918761166626626e-05,
+      "loss": 1.0581,
+      "step": 6606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8917773045527347e-05,
+      "loss": 1.1343,
+      "step": 6607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891678479487855e-05,
+      "loss": 1.0937,
+      "step": 6608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891579641469875e-05,
+      "loss": 1.0574,
+      "step": 6609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.891480790500645e-05,
+      "loss": 1.0844,
+      "step": 6610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8913819265820166e-05,
+      "loss": 1.0819,
+      "step": 6611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8912830497158414e-05,
+      "loss": 1.0943,
+      "step": 6612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8911841599039704e-05,
+      "loss": 1.1007,
+      "step": 6613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8910852571482566e-05,
+      "loss": 1.1227,
+      "step": 6614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890986341450551e-05,
+      "loss": 1.1099,
+      "step": 6615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8908874128127062e-05,
+      "loss": 1.0719,
+      "step": 6616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8907884712365753e-05,
+      "loss": 1.0551,
+      "step": 6617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8906895167240113e-05,
+      "loss": 1.1254,
+      "step": 6618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8905905492768666e-05,
+      "loss": 1.0871,
+      "step": 6619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8904915688969952e-05,
+      "loss": 1.1274,
+      "step": 6620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8903925755862504e-05,
+      "loss": 1.0676,
+      "step": 6621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.890293569346486e-05,
+      "loss": 1.123,
+      "step": 6622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8901945501795568e-05,
+      "loss": 1.0953,
+      "step": 6623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8900955180873163e-05,
+      "loss": 1.1231,
+      "step": 6624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8899964730716194e-05,
+      "loss": 1.1001,
+      "step": 6625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8898974151343207e-05,
+      "loss": 1.1135,
+      "step": 6626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8897983442772756e-05,
+      "loss": 1.0,
+      "step": 6627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8896992605023393e-05,
+      "loss": 1.049,
+      "step": 6628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8896001638113673e-05,
+      "loss": 1.0725,
+      "step": 6629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8895010542062152e-05,
+      "loss": 1.1127,
+      "step": 6630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8894019316887393e-05,
+      "loss": 1.1195,
+      "step": 6631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8893027962607958e-05,
+      "loss": 1.0857,
+      "step": 6632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8892036479242415e-05,
+      "loss": 1.0852,
+      "step": 6633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8891044866809326e-05,
+      "loss": 1.1148,
+      "step": 6634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889005312532726e-05,
+      "loss": 1.0474,
+      "step": 6635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88890612548148e-05,
+      "loss": 1.1143,
+      "step": 6636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888806925529051e-05,
+      "loss": 1.1189,
+      "step": 6637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8887077126772973e-05,
+      "loss": 0.9885,
+      "step": 6638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8886084869280767e-05,
+      "loss": 1.0149,
+      "step": 6639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888509248283247e-05,
+      "loss": 1.1281,
+      "step": 6640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8884099967446675e-05,
+      "loss": 1.0819,
+      "step": 6641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8883107323141963e-05,
+      "loss": 1.0891,
+      "step": 6642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8882114549936922e-05,
+      "loss": 1.0499,
+      "step": 6643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8881121647850146e-05,
+      "loss": 0.9552,
+      "step": 6644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8880128616900234e-05,
+      "loss": 1.0953,
+      "step": 6645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8879135457105774e-05,
+      "loss": 1.0625,
+      "step": 6646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887814216848537e-05,
+      "loss": 1.0944,
+      "step": 6647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887714875105762e-05,
+      "loss": 1.0843,
+      "step": 6648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8876155204841133e-05,
+      "loss": 1.0468,
+      "step": 6649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887516152985451e-05,
+      "loss": 1.0345,
+      "step": 6650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8874167726116364e-05,
+      "loss": 1.0675,
+      "step": 6651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8873173793645305e-05,
+      "loss": 1.1076,
+      "step": 6652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887217973245994e-05,
+      "loss": 1.0553,
+      "step": 6653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8871185542578893e-05,
+      "loss": 1.0183,
+      "step": 6654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.887019122402078e-05,
+      "loss": 1.0692,
+      "step": 6655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886919677680422e-05,
+      "loss": 1.0944,
+      "step": 6656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886820220094784e-05,
+      "loss": 1.0435,
+      "step": 6657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886720749647026e-05,
+      "loss": 1.1589,
+      "step": 6658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8866212663390104e-05,
+      "loss": 1.0372,
+      "step": 6659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8865217701726017e-05,
+      "loss": 1.0495,
+      "step": 6660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8864222611496624e-05,
+      "loss": 0.9665,
+      "step": 6661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8863227392720554e-05,
+      "loss": 1.1266,
+      "step": 6662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.886223204541645e-05,
+      "loss": 1.0771,
+      "step": 6663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8861236569602955e-05,
+      "loss": 1.07,
+      "step": 6664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8860240965298712e-05,
+      "loss": 1.0899,
+      "step": 6665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885924523252236e-05,
+      "loss": 1.046,
+      "step": 6666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885824937129254e-05,
+      "loss": 1.0515,
+      "step": 6667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.885725338162792e-05,
+      "loss": 1.0363,
+      "step": 6668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8856257263547136e-05,
+      "loss": 1.111,
+      "step": 6669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8855261017068848e-05,
+      "loss": 1.0811,
+      "step": 6670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8854264642211715e-05,
+      "loss": 1.0981,
+      "step": 6671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8853268138994397e-05,
+      "loss": 1.0434,
+      "step": 6672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8852271507435544e-05,
+      "loss": 1.0636,
+      "step": 6673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8851274747553833e-05,
+      "loss": 1.072,
+      "step": 6674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8850277859367927e-05,
+      "loss": 1.0397,
+      "step": 6675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8849280842896494e-05,
+      "loss": 1.0819,
+      "step": 6676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8848283698158203e-05,
+      "loss": 1.0457,
+      "step": 6677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884728642517173e-05,
+      "loss": 1.0623,
+      "step": 6678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884628902395575e-05,
+      "loss": 1.1369,
+      "step": 6679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8845291494528944e-05,
+      "loss": 1.0764,
+      "step": 6680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8844293836909985e-05,
+      "loss": 0.9692,
+      "step": 6681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8843296051117566e-05,
+      "loss": 1.1081,
+      "step": 6682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8842298137170367e-05,
+      "loss": 1.0463,
+      "step": 6683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884130009508708e-05,
+      "loss": 1.1071,
+      "step": 6684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.884030192488639e-05,
+      "loss": 1.038,
+      "step": 6685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883930362658699e-05,
+      "loss": 1.1352,
+      "step": 6686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8838305200207583e-05,
+      "loss": 1.1113,
+      "step": 6687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883730664576686e-05,
+      "loss": 1.0623,
+      "step": 6688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883630796328352e-05,
+      "loss": 1.0922,
+      "step": 6689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883530915277627e-05,
+      "loss": 1.0793,
+      "step": 6690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8834310214263812e-05,
+      "loss": 1.0486,
+      "step": 6691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8833311147764852e-05,
+      "loss": 1.0473,
+      "step": 6692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8832311953298102e-05,
+      "loss": 1.0941,
+      "step": 6693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8831312630882275e-05,
+      "loss": 1.0957,
+      "step": 6694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.883031318053608e-05,
+      "loss": 1.1204,
+      "step": 6695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882931360227824e-05,
+      "loss": 1.1367,
+      "step": 6696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882831389612747e-05,
+      "loss": 1.1382,
+      "step": 6697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.882731406210249e-05,
+      "loss": 1.03,
+      "step": 6698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8826314100222034e-05,
+      "loss": 1.0832,
+      "step": 6699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8825314010504817e-05,
+      "loss": 1.0738,
+      "step": 6700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8824313792969573e-05,
+      "loss": 1.0537,
+      "step": 6701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8823313447635034e-05,
+      "loss": 1.0413,
+      "step": 6702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8822312974519927e-05,
+      "loss": 1.1046,
+      "step": 6703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8821312373642992e-05,
+      "loss": 1.0538,
+      "step": 6704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8820311645022976e-05,
+      "loss": 1.0475,
+      "step": 6705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8819310788678607e-05,
+      "loss": 1.0428,
+      "step": 6706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881830980462863e-05,
+      "loss": 1.0894,
+      "step": 6707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8817308692891794e-05,
+      "loss": 1.0318,
+      "step": 6708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881630745348685e-05,
+      "loss": 1.1073,
+      "step": 6709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881530608643254e-05,
+      "loss": 1.0822,
+      "step": 6710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8814304591747623e-05,
+      "loss": 1.0775,
+      "step": 6711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881330296945085e-05,
+      "loss": 1.1185,
+      "step": 6712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.881230121956098e-05,
+      "loss": 1.114,
+      "step": 6713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8811299342096778e-05,
+      "loss": 1.0619,
+      "step": 6714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8810297337077e-05,
+      "loss": 1.0833,
+      "step": 6715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8809295204520407e-05,
+      "loss": 1.0894,
+      "step": 6716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8808292944445777e-05,
+      "loss": 1.0484,
+      "step": 6717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8807290556871872e-05,
+      "loss": 1.1231,
+      "step": 6718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8806288041817465e-05,
+      "loss": 1.0681,
+      "step": 6719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8805285399301326e-05,
+      "loss": 0.9727,
+      "step": 6720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880428262934224e-05,
+      "loss": 1.0662,
+      "step": 6721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880327973195898e-05,
+      "loss": 1.0351,
+      "step": 6722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.880227670717033e-05,
+      "loss": 1.0787,
+      "step": 6723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8801273554995074e-05,
+      "loss": 1.0572,
+      "step": 6724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8800270275451995e-05,
+      "loss": 1.0529,
+      "step": 6725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8799266868559888e-05,
+      "loss": 1.0791,
+      "step": 6726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8798263334337533e-05,
+      "loss": 1.1543,
+      "step": 6727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8797259672803733e-05,
+      "loss": 1.1428,
+      "step": 6728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879625588397728e-05,
+      "loss": 1.0902,
+      "step": 6729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879525196787697e-05,
+      "loss": 1.0437,
+      "step": 6730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8794247924521607e-05,
+      "loss": 1.0947,
+      "step": 6731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8793243753929995e-05,
+      "loss": 1.115,
+      "step": 6732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8792239456120933e-05,
+      "loss": 1.1336,
+      "step": 6733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.879123503111323e-05,
+      "loss": 1.0826,
+      "step": 6734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.87902304789257e-05,
+      "loss": 1.1199,
+      "step": 6735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8789225799577155e-05,
+      "loss": 1.099,
+      "step": 6736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8788220993086407e-05,
+      "loss": 1.1107,
+      "step": 6737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8787216059472273e-05,
+      "loss": 1.0662,
+      "step": 6738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8786210998753577e-05,
+      "loss": 1.0909,
+      "step": 6739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8785205810949135e-05,
+      "loss": 1.1117,
+      "step": 6740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.878420049607777e-05,
+      "loss": 1.0975,
+      "step": 6741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8783195054158314e-05,
+      "loss": 1.0137,
+      "step": 6742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8782189485209594e-05,
+      "loss": 1.1458,
+      "step": 6743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8781183789250446e-05,
+      "loss": 1.0422,
+      "step": 6744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8780177966299693e-05,
+      "loss": 1.099,
+      "step": 6745
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8779172016376177e-05,
+      "loss": 1.069,
+      "step": 6746
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877816593949874e-05,
+      "loss": 1.0684,
+      "step": 6747
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877715973568622e-05,
+      "loss": 1.0129,
+      "step": 6748
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.877615340495746e-05,
+      "loss": 1.1037,
+      "step": 6749
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8775146947331303e-05,
+      "loss": 1.0181,
+      "step": 6750
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8774140362826597e-05,
+      "loss": 1.0903,
+      "step": 6751
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8773133651462196e-05,
+      "loss": 1.0627,
+      "step": 6752
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8772126813256957e-05,
+      "loss": 1.0637,
+      "step": 6753
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8771119848229723e-05,
+      "loss": 1.1156,
+      "step": 6754
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8770112756399363e-05,
+      "loss": 1.0055,
+      "step": 6755
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8769105537784725e-05,
+      "loss": 1.1064,
+      "step": 6756
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8768098192404685e-05,
+      "loss": 1.1095,
+      "step": 6757
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87670907202781e-05,
+      "loss": 1.1541,
+      "step": 6758
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8766083121423833e-05,
+      "loss": 1.0672,
+      "step": 6759
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.876507539586076e-05,
+      "loss": 1.0179,
+      "step": 6760
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8764067543607756e-05,
+      "loss": 1.0387,
+      "step": 6761
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8763059564683684e-05,
+      "loss": 1.0344,
+      "step": 6762
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8762051459107426e-05,
+      "loss": 1.1138,
+      "step": 6763
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8761043226897866e-05,
+      "loss": 1.1453,
+      "step": 6764
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8760034868073875e-05,
+      "loss": 1.1368,
+      "step": 6765
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8759026382654346e-05,
+      "loss": 1.0516,
+      "step": 6766
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875801777065816e-05,
+      "loss": 1.0861,
+      "step": 6767
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8757009032104207e-05,
+      "loss": 1.1314,
+      "step": 6768
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8756000167011378e-05,
+      "loss": 1.1446,
+      "step": 6769
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8754991175398563e-05,
+      "loss": 1.0133,
+      "step": 6770
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8753982057284656e-05,
+      "loss": 1.1258,
+      "step": 6771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8752972812688565e-05,
+      "loss": 1.0668,
+      "step": 6772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875196344162918e-05,
+      "loss": 1.0425,
+      "step": 6773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.875095394412541e-05,
+      "loss": 1.1089,
+      "step": 6774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8749944320196152e-05,
+      "loss": 1.0439,
+      "step": 6775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8748934569860323e-05,
+      "loss": 1.0982,
+      "step": 6776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8747924693136828e-05,
+      "loss": 1.1323,
+      "step": 6777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8746914690044576e-05,
+      "loss": 1.1314,
+      "step": 6778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874590456060249e-05,
+      "loss": 1.134,
+      "step": 6779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8744894304829473e-05,
+      "loss": 1.0529,
+      "step": 6780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8743883922744457e-05,
+      "loss": 1.0705,
+      "step": 6781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.874287341436636e-05,
+      "loss": 1.095,
+      "step": 6782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87418627797141e-05,
+      "loss": 1.0967,
+      "step": 6783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8740852018806606e-05,
+      "loss": 1.0371,
+      "step": 6784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8739841131662813e-05,
+      "loss": 1.1235,
+      "step": 6785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8738830118301646e-05,
+      "loss": 1.0271,
+      "step": 6786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8737818978742042e-05,
+      "loss": 1.0503,
+      "step": 6787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8736807713002932e-05,
+      "loss": 1.0995,
+      "step": 6788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873579632110326e-05,
+      "loss": 1.0318,
+      "step": 6789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8734784803061954e-05,
+      "loss": 1.0523,
+      "step": 6790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8733773158897973e-05,
+      "loss": 1.0789,
+      "step": 6791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8732761388630254e-05,
+      "loss": 1.0803,
+      "step": 6792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873174949227774e-05,
+      "loss": 0.9919,
+      "step": 6793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.873073746985939e-05,
+      "loss": 1.1352,
+      "step": 6794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8729725321394153e-05,
+      "loss": 1.083,
+      "step": 6795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8728713046900978e-05,
+      "loss": 1.1196,
+      "step": 6796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8727700646398828e-05,
+      "loss": 1.0721,
+      "step": 6797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8726688119906664e-05,
+      "loss": 0.9931,
+      "step": 6798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872567546744344e-05,
+      "loss": 1.0205,
+      "step": 6799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8724662689028126e-05,
+      "loss": 1.1183,
+      "step": 6800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.872364978467969e-05,
+      "loss": 1.1313,
+      "step": 6801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8722636754417095e-05,
+      "loss": 1.0182,
+      "step": 6802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8721623598259313e-05,
+      "loss": 1.0227,
+      "step": 6803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8720610316225323e-05,
+      "loss": 1.0999,
+      "step": 6804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8719596908334092e-05,
+      "loss": 1.1377,
+      "step": 6805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8718583374604604e-05,
+      "loss": 1.1406,
+      "step": 6806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8717569715055843e-05,
+      "loss": 1.0499,
+      "step": 6807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8716555929706785e-05,
+      "loss": 1.0813,
+      "step": 6808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8715542018576414e-05,
+      "loss": 1.0519,
+      "step": 6809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8714527981683724e-05,
+      "loss": 1.0744,
+      "step": 6810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8713513819047702e-05,
+      "loss": 1.0761,
+      "step": 6811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.871249953068734e-05,
+      "loss": 1.1251,
+      "step": 6812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.871148511662163e-05,
+      "loss": 1.0813,
+      "step": 6813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8710470576869577e-05,
+      "loss": 1.1394,
+      "step": 6814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8709455911450173e-05,
+      "loss": 1.0435,
+      "step": 6815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8708441120382424e-05,
+      "loss": 1.0888,
+      "step": 6816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870742620368533e-05,
+      "loss": 1.0792,
+      "step": 6817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8706411161377897e-05,
+      "loss": 1.0017,
+      "step": 6818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870539599347914e-05,
+      "loss": 1.0562,
+      "step": 6819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.870438070000806e-05,
+      "loss": 1.0587,
+      "step": 6820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8703365280983687e-05,
+      "loss": 1.0871,
+      "step": 6821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8702349736425016e-05,
+      "loss": 1.0276,
+      "step": 6822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8701334066351082e-05,
+      "loss": 1.0618,
+      "step": 6823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.87003182707809e-05,
+      "loss": 1.0881,
+      "step": 6824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8699302349733488e-05,
+      "loss": 1.0542,
+      "step": 6825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869828630322788e-05,
+      "loss": 1.0803,
+      "step": 6826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8697270131283093e-05,
+      "loss": 0.9832,
+      "step": 6827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8696253833918167e-05,
+      "loss": 1.0059,
+      "step": 6828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869523741115213e-05,
+      "loss": 1.1028,
+      "step": 6829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8694220863004014e-05,
+      "loss": 1.022,
+      "step": 6830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8693204189492858e-05,
+      "loss": 1.1331,
+      "step": 6831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8692187390637702e-05,
+      "loss": 1.0268,
+      "step": 6832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869117046645759e-05,
+      "loss": 1.0743,
+      "step": 6833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.869015341697156e-05,
+      "loss": 1.0899,
+      "step": 6834
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8689136242198665e-05,
+      "loss": 1.0544,
+      "step": 6835
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8688118942157953e-05,
+      "loss": 1.0776,
+      "step": 6836
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8687101516868468e-05,
+      "loss": 1.0425,
+      "step": 6837
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8686083966349268e-05,
+      "loss": 1.1324,
+      "step": 6838
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868506629061941e-05,
+      "loss": 1.0784,
+      "step": 6839
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.868404848969795e-05,
+      "loss": 1.1069,
+      "step": 6840
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8683030563603952e-05,
+      "loss": 1.091,
+      "step": 6841
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8682012512356472e-05,
+      "loss": 1.1258,
+      "step": 6842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8680994335974578e-05,
+      "loss": 1.0967,
+      "step": 6843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8679976034477343e-05,
+      "loss": 1.0079,
+      "step": 6844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8678957607883827e-05,
+      "loss": 1.132,
+      "step": 6845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867793905621311e-05,
+      "loss": 1.123,
+      "step": 6846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8676920379484268e-05,
+      "loss": 1.0423,
+      "step": 6847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8675901577716367e-05,
+      "loss": 1.0841,
+      "step": 6848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8674882650928496e-05,
+      "loss": 1.0826,
+      "step": 6849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8673863599139732e-05,
+      "loss": 1.0243,
+      "step": 6850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8672844422369163e-05,
+      "loss": 1.1044,
+      "step": 6851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8671825120635867e-05,
+      "loss": 0.9913,
+      "step": 6852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.867080569395894e-05,
+      "loss": 1.0234,
+      "step": 6853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8669786142357474e-05,
+      "loss": 1.0373,
+      "step": 6854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8668766465850557e-05,
+      "loss": 1.0636,
+      "step": 6855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8667746664457286e-05,
+      "loss": 1.0738,
+      "step": 6856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.866672673819676e-05,
+      "loss": 1.0718,
+      "step": 6857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8665706687088078e-05,
+      "loss": 1.0925,
+      "step": 6858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8664686511150343e-05,
+      "loss": 1.1278,
+      "step": 6859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8663666210402658e-05,
+      "loss": 1.1055,
+      "step": 6860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8662645784864134e-05,
+      "loss": 1.1111,
+      "step": 6861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8661625234553878e-05,
+      "loss": 1.0688,
+      "step": 6862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8660604559491004e-05,
+      "loss": 1.0671,
+      "step": 6863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8659583759694625e-05,
+      "loss": 1.0974,
+      "step": 6864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8658562835183858e-05,
+      "loss": 1.071,
+      "step": 6865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8657541785977817e-05,
+      "loss": 1.0684,
+      "step": 6866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865652061209563e-05,
+      "loss": 1.1513,
+      "step": 6867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865549931355642e-05,
+      "loss": 1.1051,
+      "step": 6868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8654477890379306e-05,
+      "loss": 1.0616,
+      "step": 6869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8653456342583427e-05,
+      "loss": 1.1127,
+      "step": 6870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865243467018791e-05,
+      "loss": 1.0329,
+      "step": 6871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8651412873211878e-05,
+      "loss": 1.1369,
+      "step": 6872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.865039095167448e-05,
+      "loss": 1.1105,
+      "step": 6873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8649368905594844e-05,
+      "loss": 1.0732,
+      "step": 6874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8648346734992116e-05,
+      "loss": 1.1241,
+      "step": 6875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8647324439885435e-05,
+      "loss": 1.1798,
+      "step": 6876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8646302020293947e-05,
+      "loss": 1.1063,
+      "step": 6877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8645279476236795e-05,
+      "loss": 1.1293,
+      "step": 6878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8644256807733136e-05,
+      "loss": 1.089,
+      "step": 6879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8643234014802115e-05,
+      "loss": 1.0919,
+      "step": 6880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.864221109746289e-05,
+      "loss": 1.1029,
+      "step": 6881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8641188055734613e-05,
+      "loss": 1.0948,
+      "step": 6882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8640164889636444e-05,
+      "loss": 1.0283,
+      "step": 6883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8639141599187547e-05,
+      "loss": 1.0842,
+      "step": 6884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8638118184407082e-05,
+      "loss": 1.0406,
+      "step": 6885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8637094645314214e-05,
+      "loss": 1.119,
+      "step": 6886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863607098192811e-05,
+      "loss": 1.0137,
+      "step": 6887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.863504719426795e-05,
+      "loss": 1.0496,
+      "step": 6888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8634023282352898e-05,
+      "loss": 1.1101,
+      "step": 6889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8632999246202123e-05,
+      "loss": 1.089,
+      "step": 6890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8631975085834818e-05,
+      "loss": 1.1124,
+      "step": 6891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8630950801270147e-05,
+      "loss": 1.088,
+      "step": 6892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8629926392527303e-05,
+      "loss": 1.1362,
+      "step": 6893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862890185962546e-05,
+      "loss": 1.1022,
+      "step": 6894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8627877202583815e-05,
+      "loss": 1.0539,
+      "step": 6895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862685242142155e-05,
+      "loss": 1.0981,
+      "step": 6896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8625827516157854e-05,
+      "loss": 1.0991,
+      "step": 6897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.862480248681193e-05,
+      "loss": 1.0927,
+      "step": 6898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8623777333402966e-05,
+      "loss": 1.0888,
+      "step": 6899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8622752055950164e-05,
+      "loss": 1.0304,
+      "step": 6900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8621726654472722e-05,
+      "loss": 1.1432,
+      "step": 6901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8620701128989842e-05,
+      "loss": 1.0759,
+      "step": 6902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861967547952073e-05,
+      "loss": 1.0808,
+      "step": 6903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8618649706084597e-05,
+      "loss": 1.0595,
+      "step": 6904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861762380870065e-05,
+      "loss": 1.1218,
+      "step": 6905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8616597787388098e-05,
+      "loss": 1.0764,
+      "step": 6906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861557164216616e-05,
+      "loss": 1.1009,
+      "step": 6907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8614545373054048e-05,
+      "loss": 1.0504,
+      "step": 6908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8613518980070988e-05,
+      "loss": 1.0416,
+      "step": 6909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8612492463236193e-05,
+      "loss": 1.0286,
+      "step": 6910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861146582256889e-05,
+      "loss": 1.0836,
+      "step": 6911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.861043905808831e-05,
+      "loss": 1.1009,
+      "step": 6912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8609412169813675e-05,
+      "loss": 1.073,
+      "step": 6913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8608385157764222e-05,
+      "loss": 1.0709,
+      "step": 6914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8607358021959172e-05,
+      "loss": 1.0619,
+      "step": 6915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860633076241777e-05,
+      "loss": 1.1199,
+      "step": 6916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8605303379159253e-05,
+      "loss": 1.0467,
+      "step": 6917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8604275872202862e-05,
+      "loss": 1.143,
+      "step": 6918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8603248241567833e-05,
+      "loss": 1.0839,
+      "step": 6919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860222048727341e-05,
+      "loss": 1.0809,
+      "step": 6920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8601192609338845e-05,
+      "loss": 1.0301,
+      "step": 6921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.860016460778339e-05,
+      "loss": 1.016,
+      "step": 6922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859913648262629e-05,
+      "loss": 1.0831,
+      "step": 6923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85981082338868e-05,
+      "loss": 1.022,
+      "step": 6924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8597079861584178e-05,
+      "loss": 1.1048,
+      "step": 6925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8596051365737682e-05,
+      "loss": 1.0659,
+      "step": 6926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859502274636657e-05,
+      "loss": 1.153,
+      "step": 6927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8593994003490107e-05,
+      "loss": 1.1413,
+      "step": 6928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.859296513712756e-05,
+      "loss": 1.0998,
+      "step": 6929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8591936147298195e-05,
+      "loss": 1.1317,
+      "step": 6930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8590907034021283e-05,
+      "loss": 1.0947,
+      "step": 6931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.858987779731609e-05,
+      "loss": 1.0872,
+      "step": 6932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8588848437201902e-05,
+      "loss": 1.1006,
+      "step": 6933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8587818953697988e-05,
+      "loss": 1.0439,
+      "step": 6934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8586789346823627e-05,
+      "loss": 1.1157,
+      "step": 6935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8585759616598104e-05,
+      "loss": 1.0818,
+      "step": 6936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85847297630407e-05,
+      "loss": 1.1133,
+      "step": 6937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8583699786170703e-05,
+      "loss": 1.0595,
+      "step": 6938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8582669686007403e-05,
+      "loss": 1.1207,
+      "step": 6939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8581639462570085e-05,
+      "loss": 1.0941,
+      "step": 6940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8580609115878045e-05,
+      "loss": 1.0959,
+      "step": 6941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8579578645950582e-05,
+      "loss": 1.067,
+      "step": 6942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857854805280699e-05,
+      "loss": 1.1431,
+      "step": 6943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857751733646657e-05,
+      "loss": 1.0421,
+      "step": 6944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8576486496948622e-05,
+      "loss": 1.0689,
+      "step": 6945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.857545553427245e-05,
+      "loss": 1.0439,
+      "step": 6946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8574424448457367e-05,
+      "loss": 1.0722,
+      "step": 6947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8573393239522677e-05,
+      "loss": 1.048,
+      "step": 6948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8572361907487696e-05,
+      "loss": 1.1135,
+      "step": 6949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8571330452371733e-05,
+      "loss": 1.1142,
+      "step": 6950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8570298874194104e-05,
+      "loss": 1.0901,
+      "step": 6951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8569267172974135e-05,
+      "loss": 1.1075,
+      "step": 6952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8568235348731137e-05,
+      "loss": 1.0924,
+      "step": 6953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8567203401484434e-05,
+      "loss": 1.0934,
+      "step": 6954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8566171331253358e-05,
+      "loss": 1.1016,
+      "step": 6955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8565139138057232e-05,
+      "loss": 1.1103,
+      "step": 6956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8564106821915392e-05,
+      "loss": 1.047,
+      "step": 6957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856307438284716e-05,
+      "loss": 1.1811,
+      "step": 6958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8562041820871876e-05,
+      "loss": 1.079,
+      "step": 6959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.856100913600888e-05,
+      "loss": 1.048,
+      "step": 6960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85599763282775e-05,
+      "loss": 1.0769,
+      "step": 6961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8558943397697094e-05,
+      "loss": 1.1087,
+      "step": 6962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8557910344286995e-05,
+      "loss": 1.1135,
+      "step": 6963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8556877168066552e-05,
+      "loss": 1.1005,
+      "step": 6964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8555843869055106e-05,
+      "loss": 1.0517,
+      "step": 6965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855481044727202e-05,
+      "loss": 1.0139,
+      "step": 6966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.855377690273664e-05,
+      "loss": 1.0608,
+      "step": 6967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8552743235468322e-05,
+      "loss": 1.1131,
+      "step": 6968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8551709445486423e-05,
+      "loss": 1.0612,
+      "step": 6969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85506755328103e-05,
+      "loss": 1.0454,
+      "step": 6970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8549641497459324e-05,
+      "loss": 1.0569,
+      "step": 6971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8548607339452855e-05,
+      "loss": 1.0784,
+      "step": 6972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8547573058810258e-05,
+      "loss": 1.097,
+      "step": 6973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8546538655550897e-05,
+      "loss": 0.9982,
+      "step": 6974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8545504129694156e-05,
+      "loss": 1.0956,
+      "step": 6975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8544469481259398e-05,
+      "loss": 1.1214,
+      "step": 6976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8543434710266006e-05,
+      "loss": 1.0682,
+      "step": 6977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8542399816733353e-05,
+      "loss": 1.0267,
+      "step": 6978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8541364800680824e-05,
+      "loss": 1.1088,
+      "step": 6979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8540329662127798e-05,
+      "loss": 1.1558,
+      "step": 6980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.853929440109366e-05,
+      "loss": 1.1282,
+      "step": 6981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.85382590175978e-05,
+      "loss": 1.0413,
+      "step": 6982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853722351165961e-05,
+      "loss": 1.1568,
+      "step": 6983
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853618788329847e-05,
+      "loss": 1.0585,
+      "step": 6984
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.853515213253379e-05,
+      "loss": 1.0706,
+      "step": 6985
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8534116259384958e-05,
+      "loss": 1.0945,
+      "step": 6986
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8533080263871372e-05,
+      "loss": 1.0983,
+      "step": 6987
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8532044146012436e-05,
+      "loss": 1.1062,
+      "step": 6988
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8531007905827552e-05,
+      "loss": 1.1373,
+      "step": 6989
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8529971543336126e-05,
+      "loss": 1.1204,
+      "step": 6990
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852893505855757e-05,
+      "loss": 1.1016,
+      "step": 6991
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852789845151129e-05,
+      "loss": 1.1338,
+      "step": 6992
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8526861722216697e-05,
+      "loss": 1.0842,
+      "step": 6993
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852582487069321e-05,
+      "loss": 1.1122,
+      "step": 6994
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8524787896960244e-05,
+      "loss": 1.1196,
+      "step": 6995
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852375080103722e-05,
+      "loss": 0.9933,
+      "step": 6996
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8522713582943555e-05,
+      "loss": 1.0997,
+      "step": 6997
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852167624269868e-05,
+      "loss": 1.094,
+      "step": 6998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.852063878032202e-05,
+      "loss": 1.0913,
+      "step": 6999
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8519601195833e-05,
+      "loss": 1.0869,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "eval_text_loss": 0.9906487464904785,
+      "eval_text_runtime": 1.6281,
+      "eval_text_samples_per_second": 45.451,
+      "eval_text_steps_per_second": 1.228,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "eval_svit_loss": 1.1540175676345825,
+      "eval_svit_runtime": 3.8831,
+      "eval_svit_samples_per_second": 175.377,
+      "eval_svit_steps_per_second": 2.833,
+      "step": 7000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851856348925105e-05,
+      "loss": 1.0804,
+      "step": 7001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8517525660595607e-05,
+      "loss": 1.0042,
+      "step": 7002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8516487709886105e-05,
+      "loss": 1.0923,
+      "step": 7003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851544963714198e-05,
+      "loss": 1.1589,
+      "step": 7004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851441144238268e-05,
+      "loss": 1.1069,
+      "step": 7005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8513373125627638e-05,
+      "loss": 1.0838,
+      "step": 7006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8512334686896302e-05,
+      "loss": 1.0558,
+      "step": 7007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.851129612620812e-05,
+      "loss": 1.091,
+      "step": 7008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8510257443582535e-05,
+      "loss": 1.0419,
+      "step": 7009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850921863903901e-05,
+      "loss": 1.0983,
+      "step": 7010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.850817971259699e-05,
+      "loss": 1.0905,
+      "step": 7011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8507140664275933e-05,
+      "loss": 1.0834,
+      "step": 7012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8506101494095296e-05,
+      "loss": 1.071,
+      "step": 7013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8505062202074542e-05,
+      "loss": 1.0929,
+      "step": 7014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8504022788233134e-05,
+      "loss": 1.0648,
+      "step": 7015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8502983252590536e-05,
+      "loss": 1.063,
+      "step": 7016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8501943595166217e-05,
+      "loss": 1.0982,
+      "step": 7017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8500903815979645e-05,
+      "loss": 1.0727,
+      "step": 7018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849986391505029e-05,
+      "loss": 1.1023,
+      "step": 7019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8498823892397633e-05,
+      "loss": 1.1014,
+      "step": 7020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8497783748041143e-05,
+      "loss": 1.0362,
+      "step": 7021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84967434820003e-05,
+      "loss": 1.0762,
+      "step": 7022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849570309429459e-05,
+      "loss": 1.0834,
+      "step": 7023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8494662584943495e-05,
+      "loss": 1.1384,
+      "step": 7024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84936219539665e-05,
+      "loss": 1.065,
+      "step": 7025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.849258120138309e-05,
+      "loss": 1.0677,
+      "step": 7026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8491540327212755e-05,
+      "loss": 1.0691,
+      "step": 7027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8490499331474993e-05,
+      "loss": 1.0622,
+      "step": 7028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8489458214189294e-05,
+      "loss": 1.0722,
+      "step": 7029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848841697537516e-05,
+      "loss": 1.1172,
+      "step": 7030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8487375615052084e-05,
+      "loss": 1.002,
+      "step": 7031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8486334133239573e-05,
+      "loss": 1.0686,
+      "step": 7032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8485292529957126e-05,
+      "loss": 1.0522,
+      "step": 7033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8484250805224253e-05,
+      "loss": 1.0257,
+      "step": 7034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8483208959060466e-05,
+      "loss": 1.0229,
+      "step": 7035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8482166991485266e-05,
+      "loss": 1.0489,
+      "step": 7036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.848112490251817e-05,
+      "loss": 1.1655,
+      "step": 7037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8480082692178696e-05,
+      "loss": 1.0817,
+      "step": 7038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847904036048636e-05,
+      "loss": 1.1035,
+      "step": 7039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847799790746068e-05,
+      "loss": 1.0779,
+      "step": 7040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8476955333121178e-05,
+      "loss": 1.0659,
+      "step": 7041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8475912637487384e-05,
+      "loss": 1.0272,
+      "step": 7042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847486982057882e-05,
+      "loss": 1.142,
+      "step": 7043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8473826882415015e-05,
+      "loss": 1.0222,
+      "step": 7044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84727838230155e-05,
+      "loss": 1.0561,
+      "step": 7045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847174064239981e-05,
+      "loss": 1.0515,
+      "step": 7046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.847069734058748e-05,
+      "loss": 1.1046,
+      "step": 7047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8469653917598048e-05,
+      "loss": 1.0379,
+      "step": 7048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.846861037345105e-05,
+      "loss": 1.0962,
+      "step": 7049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.846756670816604e-05,
+      "loss": 1.0917,
+      "step": 7050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8466522921762545e-05,
+      "loss": 1.0655,
+      "step": 7051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8465479014260132e-05,
+      "loss": 1.0933,
+      "step": 7052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8464434985678334e-05,
+      "loss": 1.1342,
+      "step": 7053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8463390836036714e-05,
+      "loss": 1.0609,
+      "step": 7054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8462346565354818e-05,
+      "loss": 1.0787,
+      "step": 7055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8461302173652208e-05,
+      "loss": 1.094,
+      "step": 7056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8460257660948434e-05,
+      "loss": 1.0846,
+      "step": 7057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8459213027263063e-05,
+      "loss": 1.0036,
+      "step": 7058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8458168272615665e-05,
+      "loss": 1.0549,
+      "step": 7059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.845712339702579e-05,
+      "loss": 1.0431,
+      "step": 7060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.845607840051301e-05,
+      "loss": 1.1092,
+      "step": 7061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8455033283096906e-05,
+      "loss": 1.1869,
+      "step": 7062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8453988044797035e-05,
+      "loss": 1.0912,
+      "step": 7063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8452942685632983e-05,
+      "loss": 1.0833,
+      "step": 7064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8451897205624316e-05,
+      "loss": 1.0779,
+      "step": 7065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.845085160479062e-05,
+      "loss": 1.0864,
+      "step": 7066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8449805883151473e-05,
+      "loss": 1.152,
+      "step": 7067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844876004072646e-05,
+      "loss": 1.089,
+      "step": 7068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844771407753517e-05,
+      "loss": 0.9813,
+      "step": 7069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8446667993597183e-05,
+      "loss": 1.1009,
+      "step": 7070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844562178893209e-05,
+      "loss": 1.068,
+      "step": 7071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8444575463559486e-05,
+      "loss": 1.0532,
+      "step": 7072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.844352901749897e-05,
+      "loss": 1.1089,
+      "step": 7073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8442482450770134e-05,
+      "loss": 1.0538,
+      "step": 7074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8441435763392572e-05,
+      "loss": 1.0755,
+      "step": 7075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8440388955385894e-05,
+      "loss": 1.0542,
+      "step": 7076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8439342026769703e-05,
+      "loss": 1.0984,
+      "step": 7077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84382949775636e-05,
+      "loss": 1.0575,
+      "step": 7078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8437247807787196e-05,
+      "loss": 1.036,
+      "step": 7079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.84362005174601e-05,
+      "loss": 1.1249,
+      "step": 7080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8435153106601926e-05,
+      "loss": 1.0828,
+      "step": 7081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.843410557523229e-05,
+      "loss": 1.092,
+      "step": 7082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8433057923370803e-05,
+      "loss": 1.0897,
+      "step": 7083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8432010151037094e-05,
+      "loss": 1.0811,
+      "step": 7084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.843096225825078e-05,
+      "loss": 1.0957,
+      "step": 7085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8429914245031483e-05,
+      "loss": 1.0625,
+      "step": 7086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842886611139883e-05,
+      "loss": 1.0438,
+      "step": 7087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8427817857372455e-05,
+      "loss": 1.0448,
+      "step": 7088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842676948297198e-05,
+      "loss": 1.0799,
+      "step": 7089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8425720988217043e-05,
+      "loss": 1.0747,
+      "step": 7090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842467237312728e-05,
+      "loss": 1.1234,
+      "step": 7091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842362363772232e-05,
+      "loss": 1.0376,
+      "step": 7092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.842257478202182e-05,
+      "loss": 1.0625,
+      "step": 7093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8421525806045404e-05,
+      "loss": 1.1062,
+      "step": 7094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8420476709812727e-05,
+      "loss": 1.0731,
+      "step": 7095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841942749334343e-05,
+      "loss": 1.0744,
+      "step": 7096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8418378156657164e-05,
+      "loss": 1.0246,
+      "step": 7097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8417328699773583e-05,
+      "loss": 1.056,
+      "step": 7098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8416279122712334e-05,
+      "loss": 1.1091,
+      "step": 7099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8415229425493078e-05,
+      "loss": 1.0048,
+      "step": 7100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841417960813547e-05,
+      "loss": 1.03,
+      "step": 7101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8413129670659167e-05,
+      "loss": 0.9882,
+      "step": 7102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.841207961308384e-05,
+      "loss": 1.0512,
+      "step": 7103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8411029435429145e-05,
+      "loss": 1.0927,
+      "step": 7104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840997913771475e-05,
+      "loss": 1.1488,
+      "step": 7105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840892871996033e-05,
+      "loss": 1.0413,
+      "step": 7106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840787818218555e-05,
+      "loss": 1.0674,
+      "step": 7107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8406827524410083e-05,
+      "loss": 1.0897,
+      "step": 7108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840577674665361e-05,
+      "loss": 1.125,
+      "step": 7109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840472584893581e-05,
+      "loss": 1.1251,
+      "step": 7110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8403674831276352e-05,
+      "loss": 1.1126,
+      "step": 7111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.840262369369493e-05,
+      "loss": 1.0852,
+      "step": 7112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8401572436211224e-05,
+      "loss": 1.0493,
+      "step": 7113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8400521058844922e-05,
+      "loss": 1.0853,
+      "step": 7114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8399469561615713e-05,
+      "loss": 1.0948,
+      "step": 7115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839841794454329e-05,
+      "loss": 1.0714,
+      "step": 7116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8397366207647343e-05,
+      "loss": 1.0721,
+      "step": 7117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839631435094757e-05,
+      "loss": 1.0903,
+      "step": 7118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8395262374463667e-05,
+      "loss": 1.066,
+      "step": 7119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839421027821534e-05,
+      "loss": 1.1313,
+      "step": 7120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8393158062222282e-05,
+      "loss": 1.089,
+      "step": 7121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.839210572650421e-05,
+      "loss": 1.0806,
+      "step": 7122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8391053271080825e-05,
+      "loss": 1.1125,
+      "step": 7123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8390000695971833e-05,
+      "loss": 1.0503,
+      "step": 7124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838894800119695e-05,
+      "loss": 1.0785,
+      "step": 7125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838789518677589e-05,
+      "loss": 1.1131,
+      "step": 7126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838684225272837e-05,
+      "loss": 1.1573,
+      "step": 7127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8385789199074104e-05,
+      "loss": 1.0611,
+      "step": 7128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8384736025832816e-05,
+      "loss": 1.0588,
+      "step": 7129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.838368273302423e-05,
+      "loss": 1.0602,
+      "step": 7130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8382629320668065e-05,
+      "loss": 1.0499,
+      "step": 7131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8381575788784055e-05,
+      "loss": 1.0729,
+      "step": 7132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8380522137391927e-05,
+      "loss": 1.0686,
+      "step": 7133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837946836651141e-05,
+      "loss": 1.036,
+      "step": 7134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8378414476162243e-05,
+      "loss": 1.1029,
+      "step": 7135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837736046636416e-05,
+      "loss": 1.1411,
+      "step": 7136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83763063371369e-05,
+      "loss": 0.996,
+      "step": 7137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8375252088500203e-05,
+      "loss": 1.0791,
+      "step": 7138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837419772047381e-05,
+      "loss": 0.9531,
+      "step": 7139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837314323307747e-05,
+      "loss": 1.0046,
+      "step": 7140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8372088626330927e-05,
+      "loss": 1.0689,
+      "step": 7141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.837103390025394e-05,
+      "loss": 1.098,
+      "step": 7142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8369979054866242e-05,
+      "loss": 1.0573,
+      "step": 7143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8368924090187605e-05,
+      "loss": 1.178,
+      "step": 7144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836786900623778e-05,
+      "loss": 1.1052,
+      "step": 7145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836681380303652e-05,
+      "loss": 1.1315,
+      "step": 7146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8365758480603595e-05,
+      "loss": 1.1189,
+      "step": 7147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836470303895876e-05,
+      "loss": 1.1438,
+      "step": 7148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8363647478121784e-05,
+      "loss": 0.9684,
+      "step": 7149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.836259179811244e-05,
+      "loss": 1.1127,
+      "step": 7150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8361535998950484e-05,
+      "loss": 1.1081,
+      "step": 7151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83604800806557e-05,
+      "loss": 1.0793,
+      "step": 7152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835942404324786e-05,
+      "loss": 1.0824,
+      "step": 7153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8358367886746738e-05,
+      "loss": 1.0879,
+      "step": 7154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835731161117211e-05,
+      "loss": 1.0731,
+      "step": 7155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8356255216543766e-05,
+      "loss": 1.0935,
+      "step": 7156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835519870288148e-05,
+      "loss": 1.0161,
+      "step": 7157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.835414207020504e-05,
+      "loss": 1.034,
+      "step": 7158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8353085318534236e-05,
+      "loss": 0.9902,
+      "step": 7159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8352028447888857e-05,
+      "loss": 1.0166,
+      "step": 7160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8350971458288696e-05,
+      "loss": 1.1491,
+      "step": 7161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834991434975354e-05,
+      "loss": 1.0674,
+      "step": 7162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8348857122303198e-05,
+      "loss": 1.0614,
+      "step": 7163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834779977595746e-05,
+      "loss": 1.1039,
+      "step": 7164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834674231073613e-05,
+      "loss": 1.1106,
+      "step": 7165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834568472665901e-05,
+      "loss": 1.0818,
+      "step": 7166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8344627023745902e-05,
+      "loss": 1.0681,
+      "step": 7167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8343569202016623e-05,
+      "loss": 1.1483,
+      "step": 7168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8342511261490973e-05,
+      "loss": 1.0317,
+      "step": 7169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834145320218877e-05,
+      "loss": 1.093,
+      "step": 7170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.834039502412983e-05,
+      "loss": 1.0564,
+      "step": 7171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8339336727333962e-05,
+      "loss": 0.9939,
+      "step": 7172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833827831182099e-05,
+      "loss": 1.0932,
+      "step": 7173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8337219777610733e-05,
+      "loss": 1.1096,
+      "step": 7174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8336161124723015e-05,
+      "loss": 1.1066,
+      "step": 7175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8335102353177667e-05,
+      "loss": 1.097,
+      "step": 7176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.833404346299451e-05,
+      "loss": 1.0514,
+      "step": 7177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8332984454193373e-05,
+      "loss": 1.0895,
+      "step": 7178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8331925326794088e-05,
+      "loss": 1.0983,
+      "step": 7179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8330866080816492e-05,
+      "loss": 1.0833,
+      "step": 7180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8329806716280427e-05,
+      "loss": 1.0927,
+      "step": 7181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8328747233205722e-05,
+      "loss": 1.0963,
+      "step": 7182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8327687631612223e-05,
+      "loss": 1.0721,
+      "step": 7183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8326627911519773e-05,
+      "loss": 1.0554,
+      "step": 7184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8325568072948214e-05,
+      "loss": 1.0869,
+      "step": 7185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8324508115917403e-05,
+      "loss": 1.0928,
+      "step": 7186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832344804044718e-05,
+      "loss": 1.1275,
+      "step": 7187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83223878465574e-05,
+      "loss": 1.1225,
+      "step": 7188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8321327534267913e-05,
+      "loss": 1.1198,
+      "step": 7189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.832026710359859e-05,
+      "loss": 1.0513,
+      "step": 7190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831920655456927e-05,
+      "loss": 1.0943,
+      "step": 7191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8318145887199832e-05,
+      "loss": 1.0958,
+      "step": 7192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8317085101510126e-05,
+      "loss": 1.0656,
+      "step": 7193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8316024197520024e-05,
+      "loss": 1.1104,
+      "step": 7194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8314963175249394e-05,
+      "loss": 1.1057,
+      "step": 7195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.83139020347181e-05,
+      "loss": 1.0463,
+      "step": 7196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.831284077594602e-05,
+      "loss": 1.0206,
+      "step": 7197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8311779398953025e-05,
+      "loss": 1.0941,
+      "step": 7198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8310717903758995e-05,
+      "loss": 1.037,
+      "step": 7199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8309656290383804e-05,
+      "loss": 1.0894,
+      "step": 7200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8308594558847337e-05,
+      "loss": 1.0824,
+      "step": 7201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8307532709169476e-05,
+      "loss": 1.0618,
+      "step": 7202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8306470741370106e-05,
+      "loss": 1.0517,
+      "step": 7203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8305408655469112e-05,
+      "loss": 1.0374,
+      "step": 7204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830434645148639e-05,
+      "loss": 1.0217,
+      "step": 7205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830328412944182e-05,
+      "loss": 1.0966,
+      "step": 7206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830222168935531e-05,
+      "loss": 1.0289,
+      "step": 7207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.830115913124675e-05,
+      "loss": 1.0718,
+      "step": 7208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8300096455136038e-05,
+      "loss": 1.1048,
+      "step": 7209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8299033661043078e-05,
+      "loss": 1.1202,
+      "step": 7210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8297970748987767e-05,
+      "loss": 1.0906,
+      "step": 7211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8296907718990016e-05,
+      "loss": 1.1107,
+      "step": 7212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829584457106973e-05,
+      "loss": 1.0296,
+      "step": 7213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8294781305246823e-05,
+      "loss": 1.0777,
+      "step": 7214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8293717921541198e-05,
+      "loss": 1.1139,
+      "step": 7215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8292654419972775e-05,
+      "loss": 1.077,
+      "step": 7216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.829159080056147e-05,
+      "loss": 1.129,
+      "step": 7217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.82905270633272e-05,
+      "loss": 1.0787,
+      "step": 7218
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8289463208289885e-05,
+      "loss": 1.0665,
+      "step": 7219
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8288399235469454e-05,
+      "loss": 1.0651,
+      "step": 7220
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8287335144885822e-05,
+      "loss": 0.9874,
+      "step": 7221
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8286270936558925e-05,
+      "loss": 1.1569,
+      "step": 7222
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8285206610508686e-05,
+      "loss": 1.1594,
+      "step": 7223
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8284142166755042e-05,
+      "loss": 1.0398,
+      "step": 7224
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8283077605317925e-05,
+      "loss": 1.0333,
+      "step": 7225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8282012926217272e-05,
+      "loss": 1.0243,
+      "step": 7226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8280948129473016e-05,
+      "loss": 1.082,
+      "step": 7227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8279883215105102e-05,
+      "loss": 1.1033,
+      "step": 7228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8278818183133478e-05,
+      "loss": 1.0599,
+      "step": 7229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8277753033578078e-05,
+      "loss": 1.0724,
+      "step": 7230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8276687766458855e-05,
+      "loss": 1.1074,
+      "step": 7231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8275622381795756e-05,
+      "loss": 1.0157,
+      "step": 7232
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8274556879608736e-05,
+      "loss": 1.0699,
+      "step": 7233
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8273491259917748e-05,
+      "loss": 1.0638,
+      "step": 7234
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8272425522742748e-05,
+      "loss": 1.0462,
+      "step": 7235
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.827135966810369e-05,
+      "loss": 1.0411,
+      "step": 7236
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8270293696020537e-05,
+      "loss": 1.0786,
+      "step": 7237
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826922760651325e-05,
+      "loss": 1.0389,
+      "step": 7238
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82681613996018e-05,
+      "loss": 1.0691,
+      "step": 7239
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8267095075306145e-05,
+      "loss": 1.1311,
+      "step": 7240
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8266028633646263e-05,
+      "loss": 1.0909,
+      "step": 7241
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8264962074642117e-05,
+      "loss": 1.1061,
+      "step": 7242
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826389539831369e-05,
+      "loss": 1.0411,
+      "step": 7243
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8262828604680945e-05,
+      "loss": 1.124,
+      "step": 7244
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826176169376387e-05,
+      "loss": 1.0506,
+      "step": 7245
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.826069466558244e-05,
+      "loss": 1.1151,
+      "step": 7246
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8259627520156642e-05,
+      "loss": 1.1169,
+      "step": 7247
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8258560257506455e-05,
+      "loss": 1.0692,
+      "step": 7248
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8257492877651874e-05,
+      "loss": 1.0489,
+      "step": 7249
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8256425380612878e-05,
+      "loss": 1.0632,
+      "step": 7250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8255357766409462e-05,
+      "loss": 1.1493,
+      "step": 7251
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.825429003506162e-05,
+      "loss": 1.1322,
+      "step": 7252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8253222186589348e-05,
+      "loss": 1.0174,
+      "step": 7253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8252154221012643e-05,
+      "loss": 1.0574,
+      "step": 7254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8251086138351506e-05,
+      "loss": 1.0443,
+      "step": 7255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8250017938625933e-05,
+      "loss": 1.0891,
+      "step": 7256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8248949621855935e-05,
+      "loss": 1.0408,
+      "step": 7257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824788118806152e-05,
+      "loss": 1.1125,
+      "step": 7258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824681263726269e-05,
+      "loss": 1.1081,
+      "step": 7259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824574396947946e-05,
+      "loss": 1.0358,
+      "step": 7260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8244675184731838e-05,
+      "loss": 1.0877,
+      "step": 7261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8243606283039843e-05,
+      "loss": 1.1215,
+      "step": 7262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.82425372644235e-05,
+      "loss": 1.0751,
+      "step": 7263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8241468128902813e-05,
+      "loss": 1.0662,
+      "step": 7264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.824039887649782e-05,
+      "loss": 1.0979,
+      "step": 7265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8239329507228526e-05,
+      "loss": 1.0892,
+      "step": 7266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.823826002111497e-05,
+      "loss": 1.067,
+      "step": 7267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.823719041817719e-05,
+      "loss": 1.0297,
+      "step": 7268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8236120698435194e-05,
+      "loss": 1.1278,
+      "step": 7269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8235050861909025e-05,
+      "loss": 1.1486,
+      "step": 7270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.823398090861872e-05,
+      "loss": 1.0514,
+      "step": 7271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8232910838584312e-05,
+      "loss": 1.0703,
+      "step": 7272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8231840651825846e-05,
+      "loss": 1.0489,
+      "step": 7273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8230770348363363e-05,
+      "loss": 1.0501,
+      "step": 7274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8229699928216898e-05,
+      "loss": 1.0721,
+      "step": 7275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8228629391406505e-05,
+      "loss": 1.1452,
+      "step": 7276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8227558737952227e-05,
+      "loss": 0.9975,
+      "step": 7277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8226487967874117e-05,
+      "loss": 1.0917,
+      "step": 7278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8225417081192228e-05,
+      "loss": 1.0038,
+      "step": 7279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8224346077926615e-05,
+      "loss": 1.0832,
+      "step": 7280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822327495809733e-05,
+      "loss": 0.9907,
+      "step": 7281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822220372172444e-05,
+      "loss": 1.1056,
+      "step": 7282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8221132368827996e-05,
+      "loss": 1.051,
+      "step": 7283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.822006089942807e-05,
+      "loss": 1.0973,
+      "step": 7284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8218989313544722e-05,
+      "loss": 1.1142,
+      "step": 7285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8217917611198023e-05,
+      "loss": 1.1209,
+      "step": 7286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821684579240804e-05,
+      "loss": 1.0655,
+      "step": 7287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8215773857194848e-05,
+      "loss": 1.0286,
+      "step": 7288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821470180557852e-05,
+      "loss": 1.1334,
+      "step": 7289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8213629637579132e-05,
+      "loss": 0.9768,
+      "step": 7290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8212557353216762e-05,
+      "loss": 1.0587,
+      "step": 7291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.821148495251149e-05,
+      "loss": 1.0356,
+      "step": 7292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8210412435483403e-05,
+      "loss": 1.1053,
+      "step": 7293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820933980215258e-05,
+      "loss": 1.0789,
+      "step": 7294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8208267052539115e-05,
+      "loss": 1.1237,
+      "step": 7295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820719418666309e-05,
+      "loss": 1.0576,
+      "step": 7296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820612120454461e-05,
+      "loss": 1.0472,
+      "step": 7297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820504810620375e-05,
+      "loss": 1.1558,
+      "step": 7298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820397489166062e-05,
+      "loss": 1.096,
+      "step": 7299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820290156093531e-05,
+      "loss": 1.0988,
+      "step": 7300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8201828114047927e-05,
+      "loss": 1.1243,
+      "step": 7301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.820075455101857e-05,
+      "loss": 1.0943,
+      "step": 7302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819968087186734e-05,
+      "loss": 1.1014,
+      "step": 7303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819860707661435e-05,
+      "loss": 1.0695,
+      "step": 7304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819753316527971e-05,
+      "loss": 1.0231,
+      "step": 7305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8196459137883525e-05,
+      "loss": 0.9865,
+      "step": 7306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8195384994445918e-05,
+      "loss": 1.0808,
+      "step": 7307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819431073498699e-05,
+      "loss": 1.0573,
+      "step": 7308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819323635952687e-05,
+      "loss": 1.0663,
+      "step": 7309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.819216186808567e-05,
+      "loss": 1.1061,
+      "step": 7310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8191087260683524e-05,
+      "loss": 1.0984,
+      "step": 7311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8190012537340547e-05,
+      "loss": 1.0272,
+      "step": 7312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8188937698076866e-05,
+      "loss": 1.0669,
+      "step": 7313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818786274291261e-05,
+      "loss": 1.0645,
+      "step": 7314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8186787671867914e-05,
+      "loss": 1.0852,
+      "step": 7315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8185712484962907e-05,
+      "loss": 1.1318,
+      "step": 7316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8184637182217722e-05,
+      "loss": 1.1004,
+      "step": 7317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.81835617636525e-05,
+      "loss": 1.1082,
+      "step": 7318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.818248622928738e-05,
+      "loss": 1.0811,
+      "step": 7319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.81814105791425e-05,
+      "loss": 1.0559,
+      "step": 7320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8180334813238012e-05,
+      "loss": 1.067,
+      "step": 7321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8179258931594055e-05,
+      "loss": 1.093,
+      "step": 7322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8178182934230777e-05,
+      "loss": 0.9849,
+      "step": 7323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8177106821168328e-05,
+      "loss": 1.125,
+      "step": 7324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8176030592426863e-05,
+      "loss": 1.0163,
+      "step": 7325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817495424802654e-05,
+      "loss": 1.0986,
+      "step": 7326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817387778798751e-05,
+      "loss": 1.0415,
+      "step": 7327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8172801212329936e-05,
+      "loss": 1.1404,
+      "step": 7328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.817172452107397e-05,
+      "loss": 1.1073,
+      "step": 7329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8170647714239788e-05,
+      "loss": 1.0387,
+      "step": 7330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816957079184755e-05,
+      "loss": 1.1208,
+      "step": 7331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816849375391742e-05,
+      "loss": 1.0242,
+      "step": 7332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816741660046957e-05,
+      "loss": 1.0246,
+      "step": 7333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8166339331524175e-05,
+      "loss": 1.0389,
+      "step": 7334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8165261947101406e-05,
+      "loss": 1.064,
+      "step": 7335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816418444722144e-05,
+      "loss": 1.0459,
+      "step": 7336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8163106831904456e-05,
+      "loss": 1.111,
+      "step": 7337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.816202910117063e-05,
+      "loss": 1.1476,
+      "step": 7338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8160951255040154e-05,
+      "loss": 1.0324,
+      "step": 7339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8159873293533205e-05,
+      "loss": 1.0128,
+      "step": 7340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8158795216669974e-05,
+      "loss": 1.0501,
+      "step": 7341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8157717024470647e-05,
+      "loss": 1.0127,
+      "step": 7342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8156638716955418e-05,
+      "loss": 1.0586,
+      "step": 7343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815556029414448e-05,
+      "loss": 1.0853,
+      "step": 7344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.815448175605803e-05,
+      "loss": 1.1266,
+      "step": 7345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8153403102716263e-05,
+      "loss": 1.0477,
+      "step": 7346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8152324334139378e-05,
+      "loss": 1.0847,
+      "step": 7347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8151245450347584e-05,
+      "loss": 1.0593,
+      "step": 7348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8150166451361077e-05,
+      "loss": 1.0068,
+      "step": 7349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8149087337200066e-05,
+      "loss": 1.0773,
+      "step": 7350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8148008107884765e-05,
+      "loss": 1.0962,
+      "step": 7351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8146928763435377e-05,
+      "loss": 1.1038,
+      "step": 7352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814584930387212e-05,
+      "loss": 1.0949,
+      "step": 7353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8144769729215206e-05,
+      "loss": 1.0978,
+      "step": 7354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8143690039484857e-05,
+      "loss": 1.0688,
+      "step": 7355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814261023470129e-05,
+      "loss": 1.0745,
+      "step": 7356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8141530314884723e-05,
+      "loss": 1.1198,
+      "step": 7357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.814045028005538e-05,
+      "loss": 1.0874,
+      "step": 7358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8139370130233493e-05,
+      "loss": 1.0883,
+      "step": 7359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8138289865439287e-05,
+      "loss": 1.0155,
+      "step": 7360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8137209485692993e-05,
+      "loss": 1.1181,
+      "step": 7361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813612899101484e-05,
+      "loss": 1.0735,
+      "step": 7362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8135048381425065e-05,
+      "loss": 1.0828,
+      "step": 7363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8133967656943903e-05,
+      "loss": 1.0793,
+      "step": 7364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8132886817591595e-05,
+      "loss": 1.0693,
+      "step": 7365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.813180586338838e-05,
+      "loss": 1.1392,
+      "step": 7366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8130724794354503e-05,
+      "loss": 1.0491,
+      "step": 7367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8129643610510206e-05,
+      "loss": 1.125,
+      "step": 7368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812856231187574e-05,
+      "loss": 1.0845,
+      "step": 7369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812748089847135e-05,
+      "loss": 0.9897,
+      "step": 7370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8126399370317296e-05,
+      "loss": 1.0048,
+      "step": 7371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8125317727433824e-05,
+      "loss": 0.9539,
+      "step": 7372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8124235969841192e-05,
+      "loss": 1.073,
+      "step": 7373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.812315409755966e-05,
+      "loss": 1.0956,
+      "step": 7374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8122072110609486e-05,
+      "loss": 1.0713,
+      "step": 7375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8120990009010934e-05,
+      "loss": 1.0551,
+      "step": 7376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811990779278427e-05,
+      "loss": 1.0729,
+      "step": 7377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8118825461949755e-05,
+      "loss": 1.0713,
+      "step": 7378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8117743016527665e-05,
+      "loss": 1.1269,
+      "step": 7379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8116660456538266e-05,
+      "loss": 1.069,
+      "step": 7380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811557778200183e-05,
+      "loss": 1.1141,
+      "step": 7381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811449499293864e-05,
+      "loss": 1.0386,
+      "step": 7382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8113412089368967e-05,
+      "loss": 1.1268,
+      "step": 7383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8112329071313088e-05,
+      "loss": 1.0894,
+      "step": 7384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8111245938791295e-05,
+      "loss": 1.0602,
+      "step": 7385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.811016269182386e-05,
+      "loss": 1.0221,
+      "step": 7386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.810907933043108e-05,
+      "loss": 1.0042,
+      "step": 7387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8107995854633238e-05,
+      "loss": 1.1133,
+      "step": 7388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8106912264450622e-05,
+      "loss": 1.0539,
+      "step": 7389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8105828559903525e-05,
+      "loss": 1.0831,
+      "step": 7390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8104744741012246e-05,
+      "loss": 1.1056,
+      "step": 7391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8103660807797078e-05,
+      "loss": 1.1187,
+      "step": 7392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.810257676027832e-05,
+      "loss": 1.1082,
+      "step": 7393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8101492598476276e-05,
+      "loss": 1.1277,
+      "step": 7394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8100408322411245e-05,
+      "loss": 1.0517,
+      "step": 7395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8099323932103537e-05,
+      "loss": 1.1143,
+      "step": 7396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8098239427573455e-05,
+      "loss": 1.045,
+      "step": 7397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809715480884131e-05,
+      "loss": 0.9959,
+      "step": 7398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8096070075927416e-05,
+      "loss": 1.0435,
+      "step": 7399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8094985228852084e-05,
+      "loss": 1.0239,
+      "step": 7400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.809390026763563e-05,
+      "loss": 1.0145,
+      "step": 7401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8092815192298375e-05,
+      "loss": 1.0758,
+      "step": 7402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8091730002860634e-05,
+      "loss": 1.0468,
+      "step": 7403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8090644699342737e-05,
+      "loss": 1.1084,
+      "step": 7404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8089559281765e-05,
+      "loss": 1.0834,
+      "step": 7405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8088473750147758e-05,
+      "loss": 1.0936,
+      "step": 7406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8087388104511332e-05,
+      "loss": 1.1172,
+      "step": 7407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8086302344876057e-05,
+      "loss": 1.062,
+      "step": 7408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.808521647126227e-05,
+      "loss": 1.0793,
+      "step": 7409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8084130483690298e-05,
+      "loss": 1.0791,
+      "step": 7410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8083044382180483e-05,
+      "loss": 1.1296,
+      "step": 7411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8081958166753166e-05,
+      "loss": 1.0648,
+      "step": 7412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8080871837428684e-05,
+      "loss": 1.0549,
+      "step": 7413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8079785394227384e-05,
+      "loss": 1.0801,
+      "step": 7414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8078698837169613e-05,
+      "loss": 1.1071,
+      "step": 7415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8077612166275716e-05,
+      "loss": 1.0395,
+      "step": 7416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8076525381566043e-05,
+      "loss": 1.1261,
+      "step": 7417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8075438483060945e-05,
+      "loss": 1.0812,
+      "step": 7418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8074351470780787e-05,
+      "loss": 1.1887,
+      "step": 7419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807326434474591e-05,
+      "loss": 1.0833,
+      "step": 7420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8072177104976683e-05,
+      "loss": 1.0867,
+      "step": 7421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8071089751493462e-05,
+      "loss": 1.1049,
+      "step": 7422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.807000228431661e-05,
+      "loss": 1.0616,
+      "step": 7423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80689147034665e-05,
+      "loss": 1.0214,
+      "step": 7424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8067827008963493e-05,
+      "loss": 1.099,
+      "step": 7425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8066739200827952e-05,
+      "loss": 1.1245,
+      "step": 7426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.806565127908026e-05,
+      "loss": 1.1069,
+      "step": 7427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8064563243740783e-05,
+      "loss": 1.003,
+      "step": 7428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80634750948299e-05,
+      "loss": 1.1605,
+      "step": 7429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.806238683236799e-05,
+      "loss": 1.055,
+      "step": 7430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.806129845637543e-05,
+      "loss": 1.0578,
+      "step": 7431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.80602099668726e-05,
+      "loss": 1.0982,
+      "step": 7432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805912136387989e-05,
+      "loss": 1.1272,
+      "step": 7433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8058032647417683e-05,
+      "loss": 0.9781,
+      "step": 7434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805694381750637e-05,
+      "loss": 1.1375,
+      "step": 7435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8055854874166335e-05,
+      "loss": 1.213,
+      "step": 7436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8054765817417978e-05,
+      "loss": 0.9775,
+      "step": 7437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805367664728169e-05,
+      "loss": 1.0032,
+      "step": 7438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805258736377787e-05,
+      "loss": 1.015,
+      "step": 7439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8051497966926917e-05,
+      "loss": 1.0801,
+      "step": 7440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.805040845674923e-05,
+      "loss": 1.0829,
+      "step": 7441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8049318833265212e-05,
+      "loss": 1.0614,
+      "step": 7442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804822909649527e-05,
+      "loss": 1.0625,
+      "step": 7443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8047139246459817e-05,
+      "loss": 1.1411,
+      "step": 7444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8046049283179252e-05,
+      "loss": 1.0113,
+      "step": 7445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804495920667399e-05,
+      "loss": 1.1031,
+      "step": 7446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804386901696445e-05,
+      "loss": 1.1049,
+      "step": 7447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8042778714071047e-05,
+      "loss": 1.0522,
+      "step": 7448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8041688298014198e-05,
+      "loss": 1.1013,
+      "step": 7449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.804059776881432e-05,
+      "loss": 1.0723,
+      "step": 7450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8039507126491836e-05,
+      "loss": 1.135,
+      "step": 7451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8038416371067174e-05,
+      "loss": 1.1234,
+      "step": 7452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.803732550256076e-05,
+      "loss": 1.0769,
+      "step": 7453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8036234520993023e-05,
+      "loss": 1.1685,
+      "step": 7454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8035143426384392e-05,
+      "loss": 1.0765,
+      "step": 7455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8034052218755298e-05,
+      "loss": 1.0893,
+      "step": 7456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.803296089812618e-05,
+      "loss": 1.0904,
+      "step": 7457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8031869464517474e-05,
+      "loss": 1.0145,
+      "step": 7458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8030777917949623e-05,
+      "loss": 1.0123,
+      "step": 7459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802968625844306e-05,
+      "loss": 1.103,
+      "step": 7460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8028594486018234e-05,
+      "loss": 1.0862,
+      "step": 7461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8027502600695592e-05,
+      "loss": 1.1046,
+      "step": 7462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8026410602495577e-05,
+      "loss": 1.0743,
+      "step": 7463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8025318491438644e-05,
+      "loss": 1.0901,
+      "step": 7464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8024226267545243e-05,
+      "loss": 1.0775,
+      "step": 7465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8023133930835826e-05,
+      "loss": 1.16,
+      "step": 7466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802204148133085e-05,
+      "loss": 1.125,
+      "step": 7467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.802094891905078e-05,
+      "loss": 1.1459,
+      "step": 7468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8019856244016067e-05,
+      "loss": 1.0672,
+      "step": 7469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801876345624718e-05,
+      "loss": 1.0605,
+      "step": 7470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8017670555764575e-05,
+      "loss": 1.0522,
+      "step": 7471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8016577542588733e-05,
+      "loss": 1.1303,
+      "step": 7472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8015484416740112e-05,
+      "loss": 1.0923,
+      "step": 7473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801439117823918e-05,
+      "loss": 1.0663,
+      "step": 7474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8013297827106424e-05,
+      "loss": 1.1754,
+      "step": 7475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801220436336231e-05,
+      "loss": 1.0602,
+      "step": 7476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8011110787027316e-05,
+      "loss": 1.0116,
+      "step": 7477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.801001709812192e-05,
+      "loss": 1.0918,
+      "step": 7478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8008923296666608e-05,
+      "loss": 1.1013,
+      "step": 7479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.800782938268186e-05,
+      "loss": 1.0824,
+      "step": 7480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8006735356188163e-05,
+      "loss": 1.1146,
+      "step": 7481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.800564121720601e-05,
+      "loss": 1.0882,
+      "step": 7482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8004546965755883e-05,
+      "loss": 1.0385,
+      "step": 7483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8003452601858275e-05,
+      "loss": 1.0962,
+      "step": 7484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8002358125533685e-05,
+      "loss": 1.1143,
+      "step": 7485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8001263536802602e-05,
+      "loss": 1.0626,
+      "step": 7486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8000168835685538e-05,
+      "loss": 1.0767,
+      "step": 7487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7999074022202977e-05,
+      "loss": 1.0464,
+      "step": 7488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7997979096375435e-05,
+      "loss": 1.1619,
+      "step": 7489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7996884058223405e-05,
+      "loss": 1.0792,
+      "step": 7490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7995788907767404e-05,
+      "loss": 0.977,
+      "step": 7491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799469364502794e-05,
+      "loss": 1.1005,
+      "step": 7492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7993598270025512e-05,
+      "loss": 1.142,
+      "step": 7493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799250278278065e-05,
+      "loss": 1.0958,
+      "step": 7494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.799140718331386e-05,
+      "loss": 1.129,
+      "step": 7495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7990311471645654e-05,
+      "loss": 1.0227,
+      "step": 7496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7989215647796566e-05,
+      "loss": 1.0367,
+      "step": 7497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7988119711787105e-05,
+      "loss": 1.058,
+      "step": 7498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7987023663637804e-05,
+      "loss": 1.0678,
+      "step": 7499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798592750336918e-05,
+      "loss": 1.0194,
+      "step": 7500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7984831231001767e-05,
+      "loss": 1.1103,
+      "step": 7501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7983734846556093e-05,
+      "loss": 1.0977,
+      "step": 7502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7982638350052686e-05,
+      "loss": 1.0653,
+      "step": 7503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7981541741512086e-05,
+      "loss": 1.0586,
+      "step": 7504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.798044502095483e-05,
+      "loss": 1.0575,
+      "step": 7505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7979348188401453e-05,
+      "loss": 1.0412,
+      "step": 7506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7978251243872494e-05,
+      "loss": 1.0701,
+      "step": 7507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79771541873885e-05,
+      "loss": 1.1123,
+      "step": 7508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7976057018970012e-05,
+      "loss": 1.1363,
+      "step": 7509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7974959738637576e-05,
+      "loss": 1.1254,
+      "step": 7510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7973862346411746e-05,
+      "loss": 1.0051,
+      "step": 7511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797276484231307e-05,
+      "loss": 1.0415,
+      "step": 7512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79716672263621e-05,
+      "loss": 1.0579,
+      "step": 7513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.797056949857939e-05,
+      "loss": 1.1038,
+      "step": 7514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79694716589855e-05,
+      "loss": 0.991,
+      "step": 7515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7968373707600992e-05,
+      "loss": 1.0849,
+      "step": 7516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796727564444642e-05,
+      "loss": 1.1808,
+      "step": 7517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7966177469542348e-05,
+      "loss": 1.0915,
+      "step": 7518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7965079182909353e-05,
+      "loss": 1.1102,
+      "step": 7519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796398078456799e-05,
+      "loss": 1.0682,
+      "step": 7520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7962882274538836e-05,
+      "loss": 0.9779,
+      "step": 7521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7961783652842453e-05,
+      "loss": 0.9916,
+      "step": 7522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.796068491949943e-05,
+      "loss": 1.1811,
+      "step": 7523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7959586074530332e-05,
+      "loss": 1.1528,
+      "step": 7524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.795848711795574e-05,
+      "loss": 1.0815,
+      "step": 7525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7957388049796236e-05,
+      "loss": 1.0328,
+      "step": 7526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79562888700724e-05,
+      "loss": 1.0581,
+      "step": 7527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7955189578804817e-05,
+      "loss": 1.1434,
+      "step": 7528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7954090176014074e-05,
+      "loss": 1.0777,
+      "step": 7529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7952990661720758e-05,
+      "loss": 1.0308,
+      "step": 7530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7951891035945465e-05,
+      "loss": 1.0874,
+      "step": 7531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.795079129870878e-05,
+      "loss": 1.0587,
+      "step": 7532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7949691450031302e-05,
+      "loss": 1.0523,
+      "step": 7533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794859148993363e-05,
+      "loss": 1.1141,
+      "step": 7534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794749141843636e-05,
+      "loss": 0.9789,
+      "step": 7535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794639123556009e-05,
+      "loss": 1.0635,
+      "step": 7536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794529094132543e-05,
+      "loss": 1.0503,
+      "step": 7537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794419053575298e-05,
+      "loss": 1.0198,
+      "step": 7538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794309001886335e-05,
+      "loss": 1.1257,
+      "step": 7539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794198939067715e-05,
+      "loss": 1.0132,
+      "step": 7540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.794088865121499e-05,
+      "loss": 1.1286,
+      "step": 7541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793978780049748e-05,
+      "loss": 1.026,
+      "step": 7542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7938686838545243e-05,
+      "loss": 1.0501,
+      "step": 7543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7937585765378895e-05,
+      "loss": 1.0148,
+      "step": 7544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7936484581019055e-05,
+      "loss": 1.0793,
+      "step": 7545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.793538328548634e-05,
+      "loss": 1.0885,
+      "step": 7546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7934281878801382e-05,
+      "loss": 1.0705,
+      "step": 7547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.79331803609848e-05,
+      "loss": 1.0729,
+      "step": 7548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7932078732057227e-05,
+      "loss": 1.0843,
+      "step": 7549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7930976992039295e-05,
+      "loss": 1.0657,
+      "step": 7550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7929875140951626e-05,
+      "loss": 1.0735,
+      "step": 7551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792877317881487e-05,
+      "loss": 1.0737,
+      "step": 7552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792767110564965e-05,
+      "loss": 1.0749,
+      "step": 7553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7926568921476607e-05,
+      "loss": 1.1043,
+      "step": 7554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792546662631639e-05,
+      "loss": 1.1197,
+      "step": 7555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7924364220189632e-05,
+      "loss": 1.1187,
+      "step": 7556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7923261703116986e-05,
+      "loss": 1.0574,
+      "step": 7557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.792215907511909e-05,
+      "loss": 1.0906,
+      "step": 7558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7921056336216603e-05,
+      "loss": 1.1327,
+      "step": 7559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7919953486430167e-05,
+      "loss": 1.1226,
+      "step": 7560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791885052578044e-05,
+      "loss": 1.0964,
+      "step": 7561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7917747454288078e-05,
+      "loss": 1.0312,
+      "step": 7562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7916644271973732e-05,
+      "loss": 1.149,
+      "step": 7563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7915540978858068e-05,
+      "loss": 1.0801,
+      "step": 7564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791443757496175e-05,
+      "loss": 1.1007,
+      "step": 7565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791333406030543e-05,
+      "loss": 1.1014,
+      "step": 7566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.791223043490978e-05,
+      "loss": 1.207,
+      "step": 7567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7911126698795473e-05,
+      "loss": 1.0866,
+      "step": 7568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7910022851983174e-05,
+      "loss": 1.0768,
+      "step": 7569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7908918894493548e-05,
+      "loss": 1.103,
+      "step": 7570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790781482634728e-05,
+      "loss": 1.1128,
+      "step": 7571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790671064756504e-05,
+      "loss": 1.0803,
+      "step": 7572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790560635816751e-05,
+      "loss": 1.0631,
+      "step": 7573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7904501958175363e-05,
+      "loss": 1.1083,
+      "step": 7574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.790339744760929e-05,
+      "loss": 1.0319,
+      "step": 7575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7902292826489967e-05,
+      "loss": 1.0573,
+      "step": 7576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7901188094838086e-05,
+      "loss": 1.0588,
+      "step": 7577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7900083252674335e-05,
+      "loss": 1.0503,
+      "step": 7578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78989783000194e-05,
+      "loss": 1.1339,
+      "step": 7579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7897873236893976e-05,
+      "loss": 1.0797,
+      "step": 7580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.789676806331876e-05,
+      "loss": 1.1255,
+      "step": 7581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7895662779314443e-05,
+      "loss": 1.123,
+      "step": 7582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7894557384901734e-05,
+      "loss": 1.0907,
+      "step": 7583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7893451880101324e-05,
+      "loss": 1.0297,
+      "step": 7584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7892346264933916e-05,
+      "loss": 1.0509,
+      "step": 7585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7891240539420222e-05,
+      "loss": 1.1558,
+      "step": 7586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.789013470358094e-05,
+      "loss": 1.0222,
+      "step": 7587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7889028757436788e-05,
+      "loss": 1.0694,
+      "step": 7588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7887922701008474e-05,
+      "loss": 0.9899,
+      "step": 7589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7886816534316708e-05,
+      "loss": 1.0599,
+      "step": 7590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7885710257382207e-05,
+      "loss": 0.9908,
+      "step": 7591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7884603870225688e-05,
+      "loss": 1.0359,
+      "step": 7592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7883497372867875e-05,
+      "loss": 1.1177,
+      "step": 7593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7882390765329483e-05,
+      "loss": 1.1106,
+      "step": 7594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7881284047631238e-05,
+      "loss": 1.1673,
+      "step": 7595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7880177219793865e-05,
+      "loss": 1.0242,
+      "step": 7596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787907028183809e-05,
+      "loss": 1.0229,
+      "step": 7597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787796323378465e-05,
+      "loss": 1.0867,
+      "step": 7598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787685607565427e-05,
+      "loss": 1.0042,
+      "step": 7599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7875748807467686e-05,
+      "loss": 1.0813,
+      "step": 7600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7874641429245634e-05,
+      "loss": 1.1122,
+      "step": 7601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.787353394100885e-05,
+      "loss": 1.0077,
+      "step": 7602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7872426342778075e-05,
+      "loss": 1.0444,
+      "step": 7603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7871318634574054e-05,
+      "loss": 1.0597,
+      "step": 7604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7870210816417522e-05,
+      "loss": 1.0331,
+      "step": 7605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7869102888329236e-05,
+      "loss": 1.1085,
+      "step": 7606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786799485032994e-05,
+      "loss": 1.1369,
+      "step": 7607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7866886702440386e-05,
+      "loss": 1.1593,
+      "step": 7608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7865778444681324e-05,
+      "loss": 1.0847,
+      "step": 7609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7864670077073503e-05,
+      "loss": 1.0459,
+      "step": 7610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786356159963769e-05,
+      "loss": 1.0709,
+      "step": 7611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.786245301239464e-05,
+      "loss": 1.1154,
+      "step": 7612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7861344315365113e-05,
+      "loss": 1.0061,
+      "step": 7613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7860235508569865e-05,
+      "loss": 1.0025,
+      "step": 7614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785912659202967e-05,
+      "loss": 1.0731,
+      "step": 7615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7858017565765293e-05,
+      "loss": 1.0618,
+      "step": 7616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7856908429797504e-05,
+      "loss": 1.076,
+      "step": 7617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7855799184147067e-05,
+      "loss": 1.0643,
+      "step": 7618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.785468982883476e-05,
+      "loss": 0.9919,
+      "step": 7619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7853580363881356e-05,
+      "loss": 1.0559,
+      "step": 7620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7852470789307635e-05,
+      "loss": 1.055,
+      "step": 7621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7851361105134373e-05,
+      "loss": 1.1358,
+      "step": 7622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7850251311382352e-05,
+      "loss": 1.0363,
+      "step": 7623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7849141408072353e-05,
+      "loss": 1.0272,
+      "step": 7624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7848031395225167e-05,
+      "loss": 1.0541,
+      "step": 7625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784692127286158e-05,
+      "loss": 1.0727,
+      "step": 7626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7845811041002373e-05,
+      "loss": 1.1178,
+      "step": 7627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784470069966835e-05,
+      "loss": 1.0324,
+      "step": 7628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.784359024888029e-05,
+      "loss": 1.0665,
+      "step": 7629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7842479688659003e-05,
+      "loss": 1.0729,
+      "step": 7630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7841369019025282e-05,
+      "loss": 1.0039,
+      "step": 7631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7840258239999916e-05,
+      "loss": 1.1317,
+      "step": 7632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783914735160372e-05,
+      "loss": 1.0483,
+      "step": 7633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7838036353857493e-05,
+      "loss": 1.0184,
+      "step": 7634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7836925246782043e-05,
+      "loss": 1.1842,
+      "step": 7635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783581403039817e-05,
+      "loss": 1.0434,
+      "step": 7636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7834702704726695e-05,
+      "loss": 1.0861,
+      "step": 7637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783359126978842e-05,
+      "loss": 1.0779,
+      "step": 7638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7832479725604162e-05,
+      "loss": 1.0678,
+      "step": 7639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.783136807219474e-05,
+      "loss": 1.1171,
+      "step": 7640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7830256309580968e-05,
+      "loss": 1.1338,
+      "step": 7641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7829144437783672e-05,
+      "loss": 1.0815,
+      "step": 7642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7828032456823666e-05,
+      "loss": 1.0416,
+      "step": 7643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.782692036672178e-05,
+      "loss": 1.0474,
+      "step": 7644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7825808167498837e-05,
+      "loss": 1.1005,
+      "step": 7645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7824695859175665e-05,
+      "loss": 1.0574,
+      "step": 7646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.78235834417731e-05,
+      "loss": 1.0892,
+      "step": 7647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7822470915311967e-05,
+      "loss": 1.0584,
+      "step": 7648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7821358279813106e-05,
+      "loss": 1.1076,
+      "step": 7649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7820245535297347e-05,
+      "loss": 1.0888,
+      "step": 7650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7819132681785535e-05,
+      "loss": 1.0332,
+      "step": 7651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7818019719298505e-05,
+      "loss": 1.0462,
+      "step": 7652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7816906647857106e-05,
+      "loss": 1.0197,
+      "step": 7653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7815793467482177e-05,
+      "loss": 1.1412,
+      "step": 7654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7814680178194566e-05,
+      "loss": 1.0037,
+      "step": 7655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.781356678001512e-05,
+      "loss": 1.0584,
+      "step": 7656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7812453272964693e-05,
+      "loss": 1.0925,
+      "step": 7657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7811339657064135e-05,
+      "loss": 1.1068,
+      "step": 7658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7810225932334304e-05,
+      "loss": 1.0904,
+      "step": 7659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7809112098796058e-05,
+      "loss": 1.0798,
+      "step": 7660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7807998156470246e-05,
+      "loss": 1.0674,
+      "step": 7661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7806884105377743e-05,
+      "loss": 1.0572,
+      "step": 7662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7805769945539396e-05,
+      "loss": 0.9934,
+      "step": 7663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7804655676976087e-05,
+      "loss": 1.0925,
+      "step": 7664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7803541299708665e-05,
+      "loss": 1.0908,
+      "step": 7665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7802426813758015e-05,
+      "loss": 1.1117,
+      "step": 7666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7801312219145002e-05,
+      "loss": 1.0251,
+      "step": 7667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7800197515890497e-05,
+      "loss": 1.0745,
+      "step": 7668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7799082704015372e-05,
+      "loss": 1.1111,
+      "step": 7669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779796778354051e-05,
+      "loss": 1.0435,
+      "step": 7670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7796852754486796e-05,
+      "loss": 0.9958,
+      "step": 7671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7795737616875096e-05,
+      "loss": 1.1132,
+      "step": 7672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7794622370726305e-05,
+      "loss": 1.0481,
+      "step": 7673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7793507016061307e-05,
+      "loss": 1.0617,
+      "step": 7674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.779239155290098e-05,
+      "loss": 1.1018,
+      "step": 7675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7791275981266228e-05,
+      "loss": 1.0945,
+      "step": 7676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7790160301177928e-05,
+      "loss": 1.1008,
+      "step": 7677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7789044512656983e-05,
+      "loss": 1.0606,
+      "step": 7678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7787928615724285e-05,
+      "loss": 1.0825,
+      "step": 7679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778681261040073e-05,
+      "loss": 1.0317,
+      "step": 7680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7785696496707226e-05,
+      "loss": 1.1107,
+      "step": 7681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778458027466466e-05,
+      "loss": 1.0444,
+      "step": 7682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7783463944293948e-05,
+      "loss": 1.1097,
+      "step": 7683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7782347505615992e-05,
+      "loss": 1.0059,
+      "step": 7684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7781230958651697e-05,
+      "loss": 1.0631,
+      "step": 7685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.778011430342197e-05,
+      "loss": 1.074,
+      "step": 7686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7778997539947734e-05,
+      "loss": 1.0415,
+      "step": 7687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.777788066824989e-05,
+      "loss": 1.1326,
+      "step": 7688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7776763688349362e-05,
+      "loss": 1.0829,
+      "step": 7689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7775646600267065e-05,
+      "loss": 1.0155,
+      "step": 7690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7774529404023918e-05,
+      "loss": 1.066,
+      "step": 7691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.7773412099640847e-05,
+      "loss": 1.055,
+      "step": 7692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7772294687138768e-05,
+      "loss": 1.0415,
+      "step": 7693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.777117716653861e-05,
+      "loss": 1.1462,
+      "step": 7694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.777005953786131e-05,
+      "loss": 1.0777,
+      "step": 7695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7768941801127784e-05,
+      "loss": 1.0457,
+      "step": 7696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7767823956358974e-05,
+      "loss": 1.1139,
+      "step": 7697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7766706003575807e-05,
+      "loss": 1.077,
+      "step": 7698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7765587942799223e-05,
+      "loss": 1.0436,
+      "step": 7699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7764469774050157e-05,
+      "loss": 1.1345,
+      "step": 7700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7763351497349553e-05,
+      "loss": 1.1077,
+      "step": 7701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7762233112718348e-05,
+      "loss": 1.0816,
+      "step": 7702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7761114620177493e-05,
+      "loss": 1.0845,
+      "step": 7703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7759996019747928e-05,
+      "loss": 1.0573,
+      "step": 7704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7758877311450607e-05,
+      "loss": 0.9635,
+      "step": 7705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775775849530647e-05,
+      "loss": 1.0334,
+      "step": 7706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7756639571336478e-05,
+      "loss": 1.0585,
+      "step": 7707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7755520539561584e-05,
+      "loss": 1.0369,
+      "step": 7708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775440140000274e-05,
+      "loss": 1.0656,
+      "step": 7709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775328215268091e-05,
+      "loss": 1.109,
+      "step": 7710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775216279761705e-05,
+      "loss": 0.994,
+      "step": 7711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.775104333483212e-05,
+      "loss": 1.0931,
+      "step": 7712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774992376434709e-05,
+      "loss": 1.0469,
+      "step": 7713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7748804086182925e-05,
+      "loss": 1.0948,
+      "step": 7714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7747684300360593e-05,
+      "loss": 1.1225,
+      "step": 7715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774656440690106e-05,
+      "loss": 1.1432,
+      "step": 7716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7745444405825304e-05,
+      "loss": 1.096,
+      "step": 7717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7744324297154294e-05,
+      "loss": 1.0897,
+      "step": 7718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7743204080909015e-05,
+      "loss": 1.0579,
+      "step": 7719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7742083757110433e-05,
+      "loss": 1.0481,
+      "step": 7720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.774096332577954e-05,
+      "loss": 1.0909,
+      "step": 7721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7739842786937312e-05,
+      "loss": 1.0601,
+      "step": 7722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7738722140604732e-05,
+      "loss": 1.0759,
+      "step": 7723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7737601386802797e-05,
+      "loss": 1.1289,
+      "step": 7724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7736480525552482e-05,
+      "loss": 1.0878,
+      "step": 7725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7735359556874785e-05,
+      "loss": 1.079,
+      "step": 7726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7734238480790694e-05,
+      "loss": 1.0778,
+      "step": 7727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7733117297321207e-05,
+      "loss": 1.0188,
+      "step": 7728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.773199600648732e-05,
+      "loss": 1.0506,
+      "step": 7729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7730874608310032e-05,
+      "loss": 1.1203,
+      "step": 7730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772975310281034e-05,
+      "loss": 1.0275,
+      "step": 7731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7728631490009248e-05,
+      "loss": 1.0315,
+      "step": 7732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7727509769927762e-05,
+      "loss": 1.0792,
+      "step": 7733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.772638794258689e-05,
+      "loss": 1.1012,
+      "step": 7734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7725266008007638e-05,
+      "loss": 1.0551,
+      "step": 7735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7724143966211018e-05,
+      "loss": 1.0142,
+      "step": 7736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7723021817218042e-05,
+      "loss": 1.0898,
+      "step": 7737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7721899561049723e-05,
+      "loss": 1.0643,
+      "step": 7738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7720777197727077e-05,
+      "loss": 1.0918,
+      "step": 7739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7719654727271125e-05,
+      "loss": 1.1097,
+      "step": 7740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7718532149702886e-05,
+      "loss": 1.0996,
+      "step": 7741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7717409465043384e-05,
+      "loss": 1.1138,
+      "step": 7742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7716286673313642e-05,
+      "loss": 1.0188,
+      "step": 7743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.771516377453469e-05,
+      "loss": 1.0609,
+      "step": 7744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7714040768727553e-05,
+      "loss": 1.0077,
+      "step": 7745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7712917655913262e-05,
+      "loss": 1.1339,
+      "step": 7746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7711794436112852e-05,
+      "loss": 1.0552,
+      "step": 7747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7710671109347353e-05,
+      "loss": 1.0599,
+      "step": 7748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7709547675637808e-05,
+      "loss": 1.0364,
+      "step": 7749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7708424135005248e-05,
+      "loss": 1.0664,
+      "step": 7750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.770730048747072e-05,
+      "loss": 1.0547,
+      "step": 7751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7706176733055263e-05,
+      "loss": 1.1414,
+      "step": 7752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7705052871779922e-05,
+      "loss": 1.057,
+      "step": 7753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7703928903665748e-05,
+      "loss": 1.0618,
+      "step": 7754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7702804828733783e-05,
+      "loss": 1.1014,
+      "step": 7755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7701680647005082e-05,
+      "loss": 0.9764,
+      "step": 7756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7700556358500695e-05,
+      "loss": 1.1089,
+      "step": 7757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769943196324168e-05,
+      "loss": 1.0785,
+      "step": 7758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7698307461249093e-05,
+      "loss": 0.9957,
+      "step": 7759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769718285254399e-05,
+      "loss": 1.1047,
+      "step": 7760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769605813714743e-05,
+      "loss": 1.0807,
+      "step": 7761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.769493331508048e-05,
+      "loss": 1.0431,
+      "step": 7762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.76938083863642e-05,
+      "loss": 1.0383,
+      "step": 7763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7692683351019667e-05,
+      "loss": 1.0291,
+      "step": 7764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7691558209067935e-05,
+      "loss": 1.0476,
+      "step": 7765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7690432960530087e-05,
+      "loss": 1.1047,
+      "step": 7766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7689307605427186e-05,
+      "loss": 1.0836,
+      "step": 7767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768818214378031e-05,
+      "loss": 1.1008,
+      "step": 7768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768705657561054e-05,
+      "loss": 1.0461,
+      "step": 7769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7685930900938952e-05,
+      "loss": 0.9797,
+      "step": 7770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7684805119786624e-05,
+      "loss": 1.0906,
+      "step": 7771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.768367923217464e-05,
+      "loss": 1.1114,
+      "step": 7772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7682553238124083e-05,
+      "loss": 1.0013,
+      "step": 7773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7681427137656044e-05,
+      "loss": 1.0206,
+      "step": 7774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7680300930791608e-05,
+      "loss": 1.0154,
+      "step": 7775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7679174617551868e-05,
+      "loss": 1.1026,
+      "step": 7776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7678048197957916e-05,
+      "loss": 1.0664,
+      "step": 7777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7676921672030842e-05,
+      "loss": 1.1452,
+      "step": 7778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767579503979175e-05,
+      "loss": 1.0645,
+      "step": 7779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767466830126173e-05,
+      "loss": 1.039,
+      "step": 7780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7673541456461886e-05,
+      "loss": 1.0501,
+      "step": 7781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7672414505413327e-05,
+      "loss": 1.0337,
+      "step": 7782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767128744813715e-05,
+      "loss": 1.0824,
+      "step": 7783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.767016028465446e-05,
+      "loss": 1.1362,
+      "step": 7784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7669033014986374e-05,
+      "loss": 1.0459,
+      "step": 7785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766790563915399e-05,
+      "loss": 1.124,
+      "step": 7786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7666778157178437e-05,
+      "loss": 1.0951,
+      "step": 7787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7665650569080813e-05,
+      "loss": 1.0678,
+      "step": 7788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7664522874882246e-05,
+      "loss": 1.0191,
+      "step": 7789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7663395074603847e-05,
+      "loss": 1.0844,
+      "step": 7790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7662267168266742e-05,
+      "loss": 1.035,
+      "step": 7791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.766113915589205e-05,
+      "loss": 1.1126,
+      "step": 7792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7660011037500893e-05,
+      "loss": 1.1276,
+      "step": 7793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7658882813114403e-05,
+      "loss": 1.0578,
+      "step": 7794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7657754482753706e-05,
+      "loss": 1.0896,
+      "step": 7795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765662604643993e-05,
+      "loss": 1.0317,
+      "step": 7796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765549750419421e-05,
+      "loss": 1.0797,
+      "step": 7797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.765436885603768e-05,
+      "loss": 1.1181,
+      "step": 7798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7653240101991476e-05,
+      "loss": 1.0469,
+      "step": 7799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7652111242076738e-05,
+      "loss": 1.0734,
+      "step": 7800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.76509822763146e-05,
+      "loss": 1.0152,
+      "step": 7801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7649853204726213e-05,
+      "loss": 1.1099,
+      "step": 7802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.764872402733271e-05,
+      "loss": 0.985,
+      "step": 7803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.764759474415525e-05,
+      "loss": 1.0924,
+      "step": 7804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7646465355214975e-05,
+      "loss": 1.0831,
+      "step": 7805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.764533586053303e-05,
+      "loss": 1.041,
+      "step": 7806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7644206260130577e-05,
+      "loss": 1.0491,
+      "step": 7807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7643076554028765e-05,
+      "loss": 1.0524,
+      "step": 7808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7641946742248748e-05,
+      "loss": 1.0744,
+      "step": 7809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.764081682481169e-05,
+      "loss": 1.0686,
+      "step": 7810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7639686801738743e-05,
+      "loss": 1.0889,
+      "step": 7811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7638556673051076e-05,
+      "loss": 1.0253,
+      "step": 7812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763742643876985e-05,
+      "loss": 1.0291,
+      "step": 7813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763629609891623e-05,
+      "loss": 1.1077,
+      "step": 7814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763516565351139e-05,
+      "loss": 1.075,
+      "step": 7815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763403510257649e-05,
+      "loss": 1.0822,
+      "step": 7816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7632904446132707e-05,
+      "loss": 1.0375,
+      "step": 7817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.763177368420122e-05,
+      "loss": 1.0612,
+      "step": 7818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7630642816803197e-05,
+      "loss": 1.1126,
+      "step": 7819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7629511843959815e-05,
+      "loss": 1.0228,
+      "step": 7820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7628380765692262e-05,
+      "loss": 1.0305,
+      "step": 7821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7627249582021715e-05,
+      "loss": 1.0722,
+      "step": 7822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7626118292969357e-05,
+      "loss": 1.0996,
+      "step": 7823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7624986898556376e-05,
+      "loss": 1.0208,
+      "step": 7824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7623855398803955e-05,
+      "loss": 1.0534,
+      "step": 7825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762272379373329e-05,
+      "loss": 1.0904,
+      "step": 7826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.762159208336557e-05,
+      "loss": 1.1102,
+      "step": 7827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7620460267721984e-05,
+      "loss": 1.107,
+      "step": 7828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7619328346823735e-05,
+      "loss": 1.0417,
+      "step": 7829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7618196320692016e-05,
+      "loss": 1.1612,
+      "step": 7830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7617064189348028e-05,
+      "loss": 1.0907,
+      "step": 7831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7615931952812973e-05,
+      "loss": 1.0765,
+      "step": 7832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7614799611108058e-05,
+      "loss": 1.0084,
+      "step": 7833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7613667164254478e-05,
+      "loss": 1.1103,
+      "step": 7834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7612534612273447e-05,
+      "loss": 1.0552,
+      "step": 7835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7611401955186175e-05,
+      "loss": 1.1023,
+      "step": 7836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7610269193013877e-05,
+      "loss": 1.1388,
+      "step": 7837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7609136325777754e-05,
+      "loss": 1.0476,
+      "step": 7838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7608003353499032e-05,
+      "loss": 1.0475,
+      "step": 7839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760687027619893e-05,
+      "loss": 1.104,
+      "step": 7840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7605737093898655e-05,
+      "loss": 1.0787,
+      "step": 7841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760460380661944e-05,
+      "loss": 1.058,
+      "step": 7842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.76034704143825e-05,
+      "loss": 1.0846,
+      "step": 7843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760233691720907e-05,
+      "loss": 1.0965,
+      "step": 7844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760120331512036e-05,
+      "loss": 1.001,
+      "step": 7845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.760006960813762e-05,
+      "loss": 1.0717,
+      "step": 7846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7598935796282068e-05,
+      "loss": 1.0653,
+      "step": 7847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759780187957494e-05,
+      "loss": 1.0523,
+      "step": 7848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7596667858037467e-05,
+      "loss": 1.0226,
+      "step": 7849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7595533731690893e-05,
+      "loss": 1.0734,
+      "step": 7850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7594399500556454e-05,
+      "loss": 1.0612,
+      "step": 7851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7593265164655386e-05,
+      "loss": 1.0313,
+      "step": 7852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.759213072400894e-05,
+      "loss": 1.0542,
+      "step": 7853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7590996178638355e-05,
+      "loss": 1.2466,
+      "step": 7854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758986152856488e-05,
+      "loss": 1.1196,
+      "step": 7855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7588726773809762e-05,
+      "loss": 1.0525,
+      "step": 7856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758759191439425e-05,
+      "loss": 1.0215,
+      "step": 7857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7586456950339608e-05,
+      "loss": 1.0646,
+      "step": 7858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7585321881667074e-05,
+      "loss": 1.0798,
+      "step": 7859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7584186708397915e-05,
+      "loss": 1.0268,
+      "step": 7860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7583051430553387e-05,
+      "loss": 1.116,
+      "step": 7861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.758191604815475e-05,
+      "loss": 1.0035,
+      "step": 7862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7580780561223266e-05,
+      "loss": 1.095,
+      "step": 7863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75796449697802e-05,
+      "loss": 1.0999,
+      "step": 7864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7578509273846818e-05,
+      "loss": 1.0363,
+      "step": 7865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757737347344439e-05,
+      "loss": 1.0246,
+      "step": 7866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757623756859418e-05,
+      "loss": 1.0745,
+      "step": 7867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7575101559317468e-05,
+      "loss": 1.0854,
+      "step": 7868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7573965445635525e-05,
+      "loss": 1.0316,
+      "step": 7869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7572829227569626e-05,
+      "loss": 1.1314,
+      "step": 7870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7571692905141052e-05,
+      "loss": 1.0199,
+      "step": 7871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.757055647837108e-05,
+      "loss": 1.042,
+      "step": 7872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756941994728099e-05,
+      "loss": 1.1043,
+      "step": 7873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756828331189207e-05,
+      "loss": 1.0695,
+      "step": 7874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7567146572225605e-05,
+      "loss": 1.034,
+      "step": 7875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756600972830288e-05,
+      "loss": 1.1236,
+      "step": 7876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756487278014519e-05,
+      "loss": 1.0604,
+      "step": 7877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7563735727773824e-05,
+      "loss": 1.0487,
+      "step": 7878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7562598571210076e-05,
+      "loss": 1.0673,
+      "step": 7879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.756146131047524e-05,
+      "loss": 1.0891,
+      "step": 7880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7560323945590613e-05,
+      "loss": 1.1037,
+      "step": 7881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7559186476577495e-05,
+      "loss": 1.0743,
+      "step": 7882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755804890345719e-05,
+      "loss": 0.9897,
+      "step": 7883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7556911226251e-05,
+      "loss": 1.0286,
+      "step": 7884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7555773444980233e-05,
+      "loss": 1.1259,
+      "step": 7885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7554635559666187e-05,
+      "loss": 1.0712,
+      "step": 7886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7553497570330183e-05,
+      "loss": 1.0698,
+      "step": 7887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7552359476993525e-05,
+      "loss": 1.0933,
+      "step": 7888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7551221279677526e-05,
+      "loss": 0.9566,
+      "step": 7889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.755008297840351e-05,
+      "loss": 1.0902,
+      "step": 7890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.754894457319278e-05,
+      "loss": 1.0392,
+      "step": 7891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7547806064066666e-05,
+      "loss": 1.0832,
+      "step": 7892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.754666745104648e-05,
+      "loss": 0.9643,
+      "step": 7893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7545528734153555e-05,
+      "loss": 1.0829,
+      "step": 7894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7544389913409208e-05,
+      "loss": 1.062,
+      "step": 7895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7543250988834772e-05,
+      "loss": 1.0695,
+      "step": 7896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.754211196045157e-05,
+      "loss": 1.0232,
+      "step": 7897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7540972828280936e-05,
+      "loss": 1.0713,
+      "step": 7898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.75398335923442e-05,
+      "loss": 1.0711,
+      "step": 7899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7538694252662697e-05,
+      "loss": 1.0212,
+      "step": 7900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7537554809257765e-05,
+      "loss": 1.054,
+      "step": 7901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7536415262150742e-05,
+      "loss": 1.0311,
+      "step": 7902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753527561136297e-05,
+      "loss": 1.0745,
+      "step": 7903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7534135856915787e-05,
+      "loss": 1.1032,
+      "step": 7904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753299599883054e-05,
+      "loss": 1.0726,
+      "step": 7905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7531856037128578e-05,
+      "loss": 1.0299,
+      "step": 7906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.753071597183124e-05,
+      "loss": 1.0155,
+      "step": 7907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.752957580295989e-05,
+      "loss": 1.0407,
+      "step": 7908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7528435530535863e-05,
+      "loss": 1.0247,
+      "step": 7909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7527295154580532e-05,
+      "loss": 1.1199,
+      "step": 7910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.752615467511524e-05,
+      "loss": 1.0531,
+      "step": 7911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7525014092161343e-05,
+      "loss": 1.0493,
+      "step": 7912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7523873405740206e-05,
+      "loss": 1.0774,
+      "step": 7913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7522732615873195e-05,
+      "loss": 1.1156,
+      "step": 7914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7521591722581666e-05,
+      "loss": 1.0473,
+      "step": 7915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.752045072588699e-05,
+      "loss": 1.1017,
+      "step": 7916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7519309625810528e-05,
+      "loss": 1.1509,
+      "step": 7917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7518168422373654e-05,
+      "loss": 1.0792,
+      "step": 7918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7517027115597743e-05,
+      "loss": 1.0471,
+      "step": 7919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751588570550416e-05,
+      "loss": 1.0631,
+      "step": 7920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7514744192114286e-05,
+      "loss": 1.0787,
+      "step": 7921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7513602575449495e-05,
+      "loss": 1.0888,
+      "step": 7922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7512460855531167e-05,
+      "loss": 1.0712,
+      "step": 7923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.751131903238069e-05,
+      "loss": 1.0708,
+      "step": 7924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7510177106019432e-05,
+      "loss": 1.0658,
+      "step": 7925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7509035076468793e-05,
+      "loss": 1.0913,
+      "step": 7926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7507892943750148e-05,
+      "loss": 1.1259,
+      "step": 7927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.7506750707884895e-05,
+      "loss": 1.1317,
+      "step": 7928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7505608368894418e-05,
+      "loss": 1.0982,
+      "step": 7929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7504465926800116e-05,
+      "loss": 1.1158,
+      "step": 7930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.750332338162338e-05,
+      "loss": 1.0204,
+      "step": 7931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7502180733385603e-05,
+      "loss": 1.0349,
+      "step": 7932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.750103798210819e-05,
+      "loss": 1.0044,
+      "step": 7933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749989512781254e-05,
+      "loss": 0.9806,
+      "step": 7934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7498752170520053e-05,
+      "loss": 1.0318,
+      "step": 7935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749760911025213e-05,
+      "loss": 1.1259,
+      "step": 7936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7496465947030186e-05,
+      "loss": 0.9801,
+      "step": 7937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7495322680875622e-05,
+      "loss": 1.0837,
+      "step": 7938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7494179311809852e-05,
+      "loss": 1.0713,
+      "step": 7939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7493035839854285e-05,
+      "loss": 1.0377,
+      "step": 7940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.749189226503033e-05,
+      "loss": 1.1111,
+      "step": 7941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7490748587359415e-05,
+      "loss": 1.0704,
+      "step": 7942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7489604806862952e-05,
+      "loss": 1.06,
+      "step": 7943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7488460923562358e-05,
+      "loss": 1.0396,
+      "step": 7944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.748731693747906e-05,
+      "loss": 1.06,
+      "step": 7945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7486172848634475e-05,
+      "loss": 1.069,
+      "step": 7946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.748502865705003e-05,
+      "loss": 1.1188,
+      "step": 7947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.748388436274716e-05,
+      "loss": 1.0596,
+      "step": 7948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7482739965747282e-05,
+      "loss": 1.0676,
+      "step": 7949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7481595466071838e-05,
+      "loss": 1.0349,
+      "step": 7950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7480450863742254e-05,
+      "loss": 0.9732,
+      "step": 7951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.747930615877997e-05,
+      "loss": 1.1441,
+      "step": 7952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7478161351206418e-05,
+      "loss": 1.1039,
+      "step": 7953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7477016441043038e-05,
+      "loss": 1.0663,
+      "step": 7954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7475871428311277e-05,
+      "loss": 1.045,
+      "step": 7955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7474726313032567e-05,
+      "loss": 1.045,
+      "step": 7956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7473581095228363e-05,
+      "loss": 1.1036,
+      "step": 7957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7472435774920107e-05,
+      "loss": 0.9622,
+      "step": 7958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7471290352129244e-05,
+      "loss": 1.0429,
+      "step": 7959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.747014482687723e-05,
+      "loss": 1.1371,
+      "step": 7960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7468999199185516e-05,
+      "loss": 1.0427,
+      "step": 7961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7467853469075554e-05,
+      "loss": 0.9944,
+      "step": 7962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7466707636568803e-05,
+      "loss": 1.105,
+      "step": 7963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7465561701686723e-05,
+      "loss": 1.0901,
+      "step": 7964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7464415664450763e-05,
+      "loss": 1.0759,
+      "step": 7965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74632695248824e-05,
+      "loss": 1.0908,
+      "step": 7966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7462123283003086e-05,
+      "loss": 1.0527,
+      "step": 7967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7460976938834292e-05,
+      "loss": 1.1281,
+      "step": 7968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7459830492397485e-05,
+      "loss": 1.0574,
+      "step": 7969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745868394371413e-05,
+      "loss": 1.0621,
+      "step": 7970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745753729280571e-05,
+      "loss": 1.0968,
+      "step": 7971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7456390539693687e-05,
+      "loss": 1.0849,
+      "step": 7972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745524368439954e-05,
+      "loss": 1.104,
+      "step": 7973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745409672694475e-05,
+      "loss": 1.1375,
+      "step": 7974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7452949667350792e-05,
+      "loss": 1.0919,
+      "step": 7975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.745180250563915e-05,
+      "loss": 1.0211,
+      "step": 7976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74506552418313e-05,
+      "loss": 1.1239,
+      "step": 7977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7449507875948736e-05,
+      "loss": 1.0789,
+      "step": 7978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.744836040801294e-05,
+      "loss": 1.0492,
+      "step": 7979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74472128380454e-05,
+      "loss": 1.073,
+      "step": 7980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7446065166067606e-05,
+      "loss": 1.1046,
+      "step": 7981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7444917392101058e-05,
+      "loss": 1.0675,
+      "step": 7982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7443769516167237e-05,
+      "loss": 1.021,
+      "step": 7983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7442621538287655e-05,
+      "loss": 1.0822,
+      "step": 7984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7441473458483798e-05,
+      "loss": 1.0218,
+      "step": 7985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.744032527677717e-05,
+      "loss": 1.0503,
+      "step": 7986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7439176993189273e-05,
+      "loss": 1.0596,
+      "step": 7987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7438028607741614e-05,
+      "loss": 1.0786,
+      "step": 7988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7436880120455696e-05,
+      "loss": 1.0042,
+      "step": 7989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.743573153135303e-05,
+      "loss": 1.0922,
+      "step": 7990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7434582840455115e-05,
+      "loss": 1.0106,
+      "step": 7991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7433434047783478e-05,
+      "loss": 1.0534,
+      "step": 7992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7432285153359623e-05,
+      "loss": 0.9921,
+      "step": 7993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7431136157205068e-05,
+      "loss": 1.0657,
+      "step": 7994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7429987059341324e-05,
+      "loss": 1.0493,
+      "step": 7995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7428837859789924e-05,
+      "loss": 1.1003,
+      "step": 7996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742768855857238e-05,
+      "loss": 1.08,
+      "step": 7997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7426539155710212e-05,
+      "loss": 1.0263,
+      "step": 7998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.742538965122495e-05,
+      "loss": 1.0694,
+      "step": 7999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7424240045138123e-05,
+      "loss": 1.0649,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "eval_text_loss": 0.9789003729820251,
+      "eval_text_runtime": 1.5938,
+      "eval_text_samples_per_second": 46.43,
+      "eval_text_steps_per_second": 1.255,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "eval_svit_loss": 1.1422746181488037,
+      "eval_svit_runtime": 3.8277,
+      "eval_svit_samples_per_second": 177.915,
+      "eval_svit_steps_per_second": 2.874,
+      "step": 8000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7423090337471254e-05,
+      "loss": 1.1753,
+      "step": 8001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7421940528245876e-05,
+      "loss": 1.0578,
+      "step": 8002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7420790617483522e-05,
+      "loss": 1.025,
+      "step": 8003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.741964060520573e-05,
+      "loss": 1.0554,
+      "step": 8004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7418490491434024e-05,
+      "loss": 1.0989,
+      "step": 8005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7417340276189956e-05,
+      "loss": 1.0701,
+      "step": 8006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7416189959495064e-05,
+      "loss": 1.0377,
+      "step": 8007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7415039541370885e-05,
+      "loss": 1.0874,
+      "step": 8008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7413889021838962e-05,
+      "loss": 1.0167,
+      "step": 8009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7412738400920844e-05,
+      "loss": 1.0456,
+      "step": 8010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.741158767863808e-05,
+      "loss": 1.115,
+      "step": 8011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.741043685501222e-05,
+      "loss": 1.0938,
+      "step": 8012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7409285930064817e-05,
+      "loss": 1.0425,
+      "step": 8013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7408134903817413e-05,
+      "loss": 1.0907,
+      "step": 8014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7406983776291576e-05,
+      "loss": 1.0771,
+      "step": 8015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740583254750886e-05,
+      "loss": 1.0365,
+      "step": 8016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7404681217490818e-05,
+      "loss": 1.0705,
+      "step": 8017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7403529786259017e-05,
+      "loss": 1.0743,
+      "step": 8018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740237825383502e-05,
+      "loss": 1.1114,
+      "step": 8019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.740122662024039e-05,
+      "loss": 1.1016,
+      "step": 8020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.74000748854967e-05,
+      "loss": 1.1141,
+      "step": 8021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7398923049625506e-05,
+      "loss": 1.0764,
+      "step": 8022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7397771112648384e-05,
+      "loss": 1.0985,
+      "step": 8023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7396619074586913e-05,
+      "loss": 1.0109,
+      "step": 8024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7395466935462655e-05,
+      "loss": 1.0404,
+      "step": 8025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7394314695297202e-05,
+      "loss": 1.1052,
+      "step": 8026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739316235411212e-05,
+      "loss": 1.0536,
+      "step": 8027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.739200991192899e-05,
+      "loss": 1.0766,
+      "step": 8028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73908573687694e-05,
+      "loss": 1.0228,
+      "step": 8029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7389704724654926e-05,
+      "loss": 1.1399,
+      "step": 8030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.738855197960716e-05,
+      "loss": 1.061,
+      "step": 8031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7387399133647687e-05,
+      "loss": 1.1466,
+      "step": 8032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7386246186798092e-05,
+      "loss": 1.0716,
+      "step": 8033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7385093139079976e-05,
+      "loss": 1.0079,
+      "step": 8034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7383939990514922e-05,
+      "loss": 1.0742,
+      "step": 8035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7382786741124536e-05,
+      "loss": 1.0324,
+      "step": 8036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7381633390930405e-05,
+      "loss": 1.0856,
+      "step": 8037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7380479939954132e-05,
+      "loss": 1.0665,
+      "step": 8038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7379326388217317e-05,
+      "loss": 0.9812,
+      "step": 8039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7378172735741565e-05,
+      "loss": 1.0092,
+      "step": 8040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7377018982548477e-05,
+      "loss": 1.1089,
+      "step": 8041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737586512865966e-05,
+      "loss": 1.1227,
+      "step": 8042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7374711174096728e-05,
+      "loss": 1.0829,
+      "step": 8043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737355711888128e-05,
+      "loss": 1.0325,
+      "step": 8044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7372402963034938e-05,
+      "loss": 1.0232,
+      "step": 8045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7371248706579315e-05,
+      "loss": 1.1203,
+      "step": 8046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.737009434953602e-05,
+      "loss": 1.064,
+      "step": 8047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7368939891926682e-05,
+      "loss": 1.0393,
+      "step": 8048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7367785333772905e-05,
+      "loss": 1.0339,
+      "step": 8049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7366630675096324e-05,
+      "loss": 1.0741,
+      "step": 8050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7365475915918556e-05,
+      "loss": 1.088,
+      "step": 8051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.736432105626123e-05,
+      "loss": 1.0811,
+      "step": 8052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7363166096145965e-05,
+      "loss": 1.0433,
+      "step": 8053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73620110355944e-05,
+      "loss": 1.037,
+      "step": 8054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7360855874628158e-05,
+      "loss": 1.0539,
+      "step": 8055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7359700613268882e-05,
+      "loss": 1.1404,
+      "step": 8056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7358545251538194e-05,
+      "loss": 1.0905,
+      "step": 8057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735738978945774e-05,
+      "loss": 1.0906,
+      "step": 8058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7356234227049155e-05,
+      "loss": 1.1581,
+      "step": 8059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735507856433408e-05,
+      "loss": 1.1088,
+      "step": 8060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.735392280133415e-05,
+      "loss": 1.0517,
+      "step": 8061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7352766938071023e-05,
+      "loss": 1.0693,
+      "step": 8062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7351610974566333e-05,
+      "loss": 1.0602,
+      "step": 8063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7350454910841735e-05,
+      "loss": 1.049,
+      "step": 8064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7349298746918875e-05,
+      "loss": 1.0151,
+      "step": 8065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7348142482819404e-05,
+      "loss": 1.0784,
+      "step": 8066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734698611856498e-05,
+      "loss": 1.058,
+      "step": 8067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.734582965417725e-05,
+      "loss": 1.1018,
+      "step": 8068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7344673089677878e-05,
+      "loss": 1.033,
+      "step": 8069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7343516425088527e-05,
+      "loss": 1.0725,
+      "step": 8070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7342359660430847e-05,
+      "loss": 1.1243,
+      "step": 8071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7341202795726506e-05,
+      "loss": 1.056,
+      "step": 8072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7340045830997167e-05,
+      "loss": 1.07,
+      "step": 8073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.73388887662645e-05,
+      "loss": 1.0408,
+      "step": 8074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7337731601550175e-05,
+      "loss": 1.0964,
+      "step": 8075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733657433687586e-05,
+      "loss": 1.0349,
+      "step": 8076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7335416972263218e-05,
+      "loss": 1.0046,
+      "step": 8077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.733425950773394e-05,
+      "loss": 1.0328,
+      "step": 8078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7333101943309685e-05,
+      "loss": 0.9991,
+      "step": 8079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7331944279012144e-05,
+      "loss": 1.0061,
+      "step": 8080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7330786514862992e-05,
+      "loss": 1.1463,
+      "step": 8081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7329628650883906e-05,
+      "loss": 1.0472,
+      "step": 8082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7328470687096576e-05,
+      "loss": 1.0531,
+      "step": 8083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7327312623522683e-05,
+      "loss": 1.0568,
+      "step": 8084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7326154460183918e-05,
+      "loss": 1.042,
+      "step": 8085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7324996197101965e-05,
+      "loss": 1.0335,
+      "step": 8086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7323837834298518e-05,
+      "loss": 1.097,
+      "step": 8087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.732267937179527e-05,
+      "loss": 1.0533,
+      "step": 8088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7321520809613916e-05,
+      "loss": 1.0584,
+      "step": 8089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7320362147776153e-05,
+      "loss": 1.0495,
+      "step": 8090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7319203386303674e-05,
+      "loss": 0.9899,
+      "step": 8091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.731804452521818e-05,
+      "loss": 1.1238,
+      "step": 8092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7316885564541384e-05,
+      "loss": 1.0044,
+      "step": 8093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.731572650429498e-05,
+      "loss": 1.1029,
+      "step": 8094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.731456734450067e-05,
+      "loss": 0.9944,
+      "step": 8095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7313408085180172e-05,
+      "loss": 1.0395,
+      "step": 8096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.731224872635519e-05,
+      "loss": 1.0755,
+      "step": 8097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7311089268047436e-05,
+      "loss": 0.9863,
+      "step": 8098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7309929710278622e-05,
+      "loss": 1.0449,
+      "step": 8099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7308770053070464e-05,
+      "loss": 1.0762,
+      "step": 8100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730761029644468e-05,
+      "loss": 1.1172,
+      "step": 8101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730645044042299e-05,
+      "loss": 1.0607,
+      "step": 8102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7305290485027113e-05,
+      "loss": 1.0406,
+      "step": 8103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7304130430278773e-05,
+      "loss": 1.122,
+      "step": 8104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7302970276199692e-05,
+      "loss": 1.1052,
+      "step": 8105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.730181002281159e-05,
+      "loss": 1.0454,
+      "step": 8106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7300649670136214e-05,
+      "loss": 1.0973,
+      "step": 8107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7299489218195273e-05,
+      "loss": 1.0707,
+      "step": 8108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7298328667010514e-05,
+      "loss": 1.0814,
+      "step": 8109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7297168016603662e-05,
+      "loss": 1.0858,
+      "step": 8110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7296007266996455e-05,
+      "loss": 0.9545,
+      "step": 8111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7294846418210634e-05,
+      "loss": 1.0462,
+      "step": 8112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7293685470267934e-05,
+      "loss": 1.0659,
+      "step": 8113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7292524423190095e-05,
+      "loss": 1.0429,
+      "step": 8114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7291363276998863e-05,
+      "loss": 1.0429,
+      "step": 8115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.729020203171598e-05,
+      "loss": 1.1491,
+      "step": 8116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7289040687363197e-05,
+      "loss": 0.9761,
+      "step": 8117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7287879243962264e-05,
+      "loss": 1.069,
+      "step": 8118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7286717701534925e-05,
+      "loss": 0.988,
+      "step": 8119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7285556060102933e-05,
+      "loss": 1.0973,
+      "step": 8120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7284394319688046e-05,
+      "loss": 0.9917,
+      "step": 8121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.728323248031202e-05,
+      "loss": 1.032,
+      "step": 8122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7282070541996606e-05,
+      "loss": 1.0187,
+      "step": 8123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7280908504763573e-05,
+      "loss": 0.9981,
+      "step": 8124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7279746368634674e-05,
+      "loss": 1.0255,
+      "step": 8125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7278584133631682e-05,
+      "loss": 1.0472,
+      "step": 8126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727742179977635e-05,
+      "loss": 1.0933,
+      "step": 8127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7276259367090463e-05,
+      "loss": 1.0991,
+      "step": 8128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727509683559577e-05,
+      "loss": 1.116,
+      "step": 8129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727393420531405e-05,
+      "loss": 1.0747,
+      "step": 8130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727277147626708e-05,
+      "loss": 1.1631,
+      "step": 8131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7271608648476628e-05,
+      "loss": 1.0221,
+      "step": 8132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.727044572196448e-05,
+      "loss": 1.0546,
+      "step": 8133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72692826967524e-05,
+      "loss": 1.0958,
+      "step": 8134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7268119572862176e-05,
+      "loss": 1.0747,
+      "step": 8135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7266956350315587e-05,
+      "loss": 1.0338,
+      "step": 8136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7265793029134424e-05,
+      "loss": 1.0861,
+      "step": 8137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726462960934047e-05,
+      "loss": 1.0221,
+      "step": 8138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7263466090955504e-05,
+      "loss": 1.0875,
+      "step": 8139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7262302474001322e-05,
+      "loss": 1.0532,
+      "step": 8140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.726113875849972e-05,
+      "loss": 1.0228,
+      "step": 8141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7259974944472476e-05,
+      "loss": 1.0015,
+      "step": 8142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.72588110319414e-05,
+      "loss": 1.0601,
+      "step": 8143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725764702092828e-05,
+      "loss": 1.1158,
+      "step": 8144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725648291145492e-05,
+      "loss": 1.0634,
+      "step": 8145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7255318703543116e-05,
+      "loss": 1.0163,
+      "step": 8146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725415439721467e-05,
+      "loss": 0.992,
+      "step": 8147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7252989992491393e-05,
+      "loss": 1.0833,
+      "step": 8148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7251825489395083e-05,
+      "loss": 1.0736,
+      "step": 8149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.725066088794755e-05,
+      "loss": 1.081,
+      "step": 8150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7249496188170606e-05,
+      "loss": 1.0743,
+      "step": 8151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724833139008606e-05,
+      "loss": 1.1272,
+      "step": 8152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7247166493715725e-05,
+      "loss": 1.0289,
+      "step": 8153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7246001499081414e-05,
+      "loss": 1.0298,
+      "step": 8154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724483640620495e-05,
+      "loss": 1.1051,
+      "step": 8155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7243671215108143e-05,
+      "loss": 1.0757,
+      "step": 8156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7242505925812825e-05,
+      "loss": 1.0808,
+      "step": 8157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.724134053834081e-05,
+      "loss": 1.1105,
+      "step": 8158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7240175052713926e-05,
+      "loss": 1.0983,
+      "step": 8159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7239009468953998e-05,
+      "loss": 1.1257,
+      "step": 8160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723784378708285e-05,
+      "loss": 1.094,
+      "step": 8161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7236678007122318e-05,
+      "loss": 1.161,
+      "step": 8162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7235512129094232e-05,
+      "loss": 1.0261,
+      "step": 8163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.7234346153020423e-05,
+      "loss": 1.1482,
+      "step": 8164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.723318007892273e-05,
+      "loss": 1.0403,
+      "step": 8165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7232013906822986e-05,
+      "loss": 1.0567,
+      "step": 8166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7230847636743035e-05,
+      "loss": 1.0914,
+      "step": 8167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722968126870471e-05,
+      "loss": 1.0961,
+      "step": 8168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7228514802729862e-05,
+      "loss": 1.0921,
+      "step": 8169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722734823884033e-05,
+      "loss": 1.0921,
+      "step": 8170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7226181577057962e-05,
+      "loss": 1.0407,
+      "step": 8171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.722501481740461e-05,
+      "loss": 1.0368,
+      "step": 8172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7223847959902115e-05,
+      "loss": 1.0696,
+      "step": 8173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7222681004572336e-05,
+      "loss": 1.089,
+      "step": 8174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7221513951437127e-05,
+      "loss": 1.0284,
+      "step": 8175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7220346800518337e-05,
+      "loss": 1.0512,
+      "step": 8176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7219179551837834e-05,
+      "loss": 1.0569,
+      "step": 8177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7218012205417466e-05,
+      "loss": 1.1284,
+      "step": 8178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.72168447612791e-05,
+      "loss": 1.0351,
+      "step": 8179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7215677219444596e-05,
+      "loss": 1.0822,
+      "step": 8180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7214509579935823e-05,
+      "loss": 0.965,
+      "step": 8181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.721334184277464e-05,
+      "loss": 1.0809,
+      "step": 8182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7212174007982926e-05,
+      "loss": 1.0699,
+      "step": 8183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7211006075582537e-05,
+      "loss": 1.0848,
+      "step": 8184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7209838045595356e-05,
+      "loss": 1.0899,
+      "step": 8185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720866991804326e-05,
+      "loss": 1.0151,
+      "step": 8186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7207501692948107e-05,
+      "loss": 1.0789,
+      "step": 8187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7206333370331795e-05,
+      "loss": 1.1514,
+      "step": 8188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720516495021619e-05,
+      "loss": 1.145,
+      "step": 8189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7203996432623178e-05,
+      "loss": 1.1087,
+      "step": 8190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.720282781757464e-05,
+      "loss": 1.0931,
+      "step": 8191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7201659105092458e-05,
+      "loss": 1.0282,
+      "step": 8192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7200490295198526e-05,
+      "loss": 1.1116,
+      "step": 8193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7199321387914727e-05,
+      "loss": 1.0943,
+      "step": 8194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7198152383262953e-05,
+      "loss": 1.0633,
+      "step": 8195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7196983281265098e-05,
+      "loss": 1.1227,
+      "step": 8196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719581408194305e-05,
+      "loss": 1.113,
+      "step": 8197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719464478531871e-05,
+      "loss": 1.0775,
+      "step": 8198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.719347539141397e-05,
+      "loss": 1.0169,
+      "step": 8199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7192305900250738e-05,
+      "loss": 1.0644,
+      "step": 8200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7191136311850908e-05,
+      "loss": 1.1153,
+      "step": 8201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7189966626236384e-05,
+      "loss": 1.0545,
+      "step": 8202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7188796843429076e-05,
+      "loss": 1.0752,
+      "step": 8203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.718762696345088e-05,
+      "loss": 1.1298,
+      "step": 8204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7186456986323712e-05,
+      "loss": 1.0038,
+      "step": 8205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7185286912069483e-05,
+      "loss": 1.0815,
+      "step": 8206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7184116740710103e-05,
+      "loss": 1.0227,
+      "step": 8207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.718294647226749e-05,
+      "loss": 0.9965,
+      "step": 8208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7181776106763546e-05,
+      "loss": 1.1077,
+      "step": 8209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7180605644220205e-05,
+      "loss": 1.0695,
+      "step": 8210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7179435084659377e-05,
+      "loss": 1.083,
+      "step": 8211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7178264428102984e-05,
+      "loss": 1.0646,
+      "step": 8212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7177093674572953e-05,
+      "loss": 1.0797,
+      "step": 8213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71759228240912e-05,
+      "loss": 1.0577,
+      "step": 8214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717475187667967e-05,
+      "loss": 1.1455,
+      "step": 8215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.717358083236027e-05,
+      "loss": 1.0185,
+      "step": 8216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7172409691154944e-05,
+      "loss": 1.044,
+      "step": 8217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7171238453085614e-05,
+      "loss": 1.0405,
+      "step": 8218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7170067118174225e-05,
+      "loss": 1.1064,
+      "step": 8219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71688956864427e-05,
+      "loss": 1.0706,
+      "step": 8220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.716772415791299e-05,
+      "loss": 1.0536,
+      "step": 8221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7166552532607024e-05,
+      "loss": 1.0358,
+      "step": 8222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7165380810546748e-05,
+      "loss": 1.0995,
+      "step": 8223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7164208991754102e-05,
+      "loss": 1.1131,
+      "step": 8224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7163037076251034e-05,
+      "loss": 1.0821,
+      "step": 8225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7161865064059492e-05,
+      "loss": 1.0738,
+      "step": 8226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7160692955201417e-05,
+      "loss": 1.0547,
+      "step": 8227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7159520749698764e-05,
+      "loss": 1.093,
+      "step": 8228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7158348447573483e-05,
+      "loss": 1.0116,
+      "step": 8229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715717604884753e-05,
+      "loss": 1.0831,
+      "step": 8230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7156003553542863e-05,
+      "loss": 1.1099,
+      "step": 8231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715483096168143e-05,
+      "loss": 1.0722,
+      "step": 8232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71536582732852e-05,
+      "loss": 1.0353,
+      "step": 8233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.715248548837613e-05,
+      "loss": 1.0571,
+      "step": 8234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7151312606976182e-05,
+      "loss": 1.0374,
+      "step": 8235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7150139629107325e-05,
+      "loss": 1.0595,
+      "step": 8236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.714896655479152e-05,
+      "loss": 1.0783,
+      "step": 8237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7147793384050735e-05,
+      "loss": 1.1231,
+      "step": 8238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7146620116906943e-05,
+      "loss": 1.0674,
+      "step": 8239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7145446753382118e-05,
+      "loss": 1.054,
+      "step": 8240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7144273293498228e-05,
+      "loss": 1.0056,
+      "step": 8241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7143099737277253e-05,
+      "loss": 0.9892,
+      "step": 8242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7141926084741164e-05,
+      "loss": 1.0609,
+      "step": 8243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.714075233591195e-05,
+      "loss": 1.0912,
+      "step": 8244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713957849081159e-05,
+      "loss": 1.1018,
+      "step": 8245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7138404549462053e-05,
+      "loss": 0.9758,
+      "step": 8246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713723051188534e-05,
+      "loss": 1.0243,
+      "step": 8247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713605637810343e-05,
+      "loss": 1.0574,
+      "step": 8248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713488214813831e-05,
+      "loss": 1.0493,
+      "step": 8249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713370782201197e-05,
+      "loss": 1.0429,
+      "step": 8250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7132533399746408e-05,
+      "loss": 1.0397,
+      "step": 8251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.713135888136361e-05,
+      "loss": 1.0626,
+      "step": 8252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7130184266885577e-05,
+      "loss": 1.0416,
+      "step": 8253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71290095563343e-05,
+      "loss": 1.0501,
+      "step": 8254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7127834749731784e-05,
+      "loss": 1.0402,
+      "step": 8255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7126659847100027e-05,
+      "loss": 1.0714,
+      "step": 8256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7125484848461028e-05,
+      "loss": 1.102,
+      "step": 8257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7124309753836795e-05,
+      "loss": 0.9749,
+      "step": 8258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7123134563249336e-05,
+      "loss": 1.1355,
+      "step": 8259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7121959276720654e-05,
+      "loss": 1.052,
+      "step": 8260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.712078389427276e-05,
+      "loss": 0.9999,
+      "step": 8261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.711960841592767e-05,
+      "loss": 1.0634,
+      "step": 8262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7118432841707393e-05,
+      "loss": 1.0499,
+      "step": 8263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7117257171633946e-05,
+      "loss": 1.0527,
+      "step": 8264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7116081405729342e-05,
+      "loss": 1.1475,
+      "step": 8265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7114905544015602e-05,
+      "loss": 1.0501,
+      "step": 8266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.711372958651475e-05,
+      "loss": 1.0648,
+      "step": 8267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.71125535332488e-05,
+      "loss": 1.1257,
+      "step": 8268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7111377384239787e-05,
+      "loss": 1.0729,
+      "step": 8269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7110201139509725e-05,
+      "loss": 1.038,
+      "step": 8270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7109024799080648e-05,
+      "loss": 0.988,
+      "step": 8271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7107848362974587e-05,
+      "loss": 1.0853,
+      "step": 8272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.710667183121357e-05,
+      "loss": 1.0512,
+      "step": 8273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7105495203819634e-05,
+      "loss": 1.0533,
+      "step": 8274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7104318480814808e-05,
+      "loss": 1.095,
+      "step": 8275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7103141662221127e-05,
+      "loss": 1.0512,
+      "step": 8276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7101964748060642e-05,
+      "loss": 1.0831,
+      "step": 8277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7100787738355376e-05,
+      "loss": 1.0288,
+      "step": 8278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7099610633127386e-05,
+      "loss": 1.031,
+      "step": 8279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7098433432398708e-05,
+      "loss": 0.99,
+      "step": 8280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7097256136191388e-05,
+      "loss": 1.0504,
+      "step": 8281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7096078744527473e-05,
+      "loss": 1.0357,
+      "step": 8282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7094901257429017e-05,
+      "loss": 0.9992,
+      "step": 8283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7093723674918063e-05,
+      "loss": 1.1219,
+      "step": 8284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.709254599701667e-05,
+      "loss": 1.0483,
+      "step": 8285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7091368223746887e-05,
+      "loss": 1.0846,
+      "step": 8286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7090190355130778e-05,
+      "loss": 1.0586,
+      "step": 8287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708901239119039e-05,
+      "loss": 1.162,
+      "step": 8288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7087834331947796e-05,
+      "loss": 1.0857,
+      "step": 8289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708665617742505e-05,
+      "loss": 1.0718,
+      "step": 8290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7085477927644214e-05,
+      "loss": 0.9939,
+      "step": 8291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7084299582627356e-05,
+      "loss": 1.1451,
+      "step": 8292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708312114239654e-05,
+      "loss": 1.0452,
+      "step": 8293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.708194260697384e-05,
+      "loss": 1.081,
+      "step": 8294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7080763976381322e-05,
+      "loss": 1.0555,
+      "step": 8295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707958525064106e-05,
+      "loss": 1.0573,
+      "step": 8296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7078406429775124e-05,
+      "loss": 1.0319,
+      "step": 8297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70772275138056e-05,
+      "loss": 1.0493,
+      "step": 8298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7076048502754558e-05,
+      "loss": 1.1069,
+      "step": 8299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7074869396644076e-05,
+      "loss": 1.0447,
+      "step": 8300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707369019549624e-05,
+      "loss": 1.1101,
+      "step": 8301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707251089933313e-05,
+      "loss": 1.08,
+      "step": 8302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707133150817683e-05,
+      "loss": 1.0941,
+      "step": 8303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.707015202204943e-05,
+      "loss": 1.0512,
+      "step": 8304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7068972440973015e-05,
+      "loss": 1.0285,
+      "step": 8305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7067792764969676e-05,
+      "loss": 1.0916,
+      "step": 8306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7066612994061505e-05,
+      "loss": 1.1049,
+      "step": 8307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7065433128270598e-05,
+      "loss": 1.0113,
+      "step": 8308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706425316761905e-05,
+      "loss": 1.1029,
+      "step": 8309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7063073112128953e-05,
+      "loss": 1.0758,
+      "step": 8310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706189296182241e-05,
+      "loss": 1.0186,
+      "step": 8311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.706071271672152e-05,
+      "loss": 1.0167,
+      "step": 8312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.705953237684839e-05,
+      "loss": 1.1547,
+      "step": 8313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7058351942225117e-05,
+      "loss": 1.092,
+      "step": 8314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7057171412873817e-05,
+      "loss": 1.0409,
+      "step": 8315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.705599078881659e-05,
+      "loss": 1.0162,
+      "step": 8316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7054810070075543e-05,
+      "loss": 1.1031,
+      "step": 8317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7053629256672794e-05,
+      "loss": 1.028,
+      "step": 8318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7052448348630458e-05,
+      "loss": 1.0747,
+      "step": 8319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.705126734597064e-05,
+      "loss": 1.0598,
+      "step": 8320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7050086248715466e-05,
+      "loss": 1.0803,
+      "step": 8321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7048905056887054e-05,
+      "loss": 1.0625,
+      "step": 8322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7047723770507517e-05,
+      "loss": 1.123,
+      "step": 8323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7046542389598982e-05,
+      "loss": 1.0846,
+      "step": 8324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7045360914183573e-05,
+      "loss": 1.0231,
+      "step": 8325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7044179344283414e-05,
+      "loss": 1.0526,
+      "step": 8326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7042997679920636e-05,
+      "loss": 1.0878,
+      "step": 8327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7041815921117364e-05,
+      "loss": 1.0482,
+      "step": 8328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.704063406789573e-05,
+      "loss": 1.0919,
+      "step": 8329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7039452120277866e-05,
+      "loss": 1.1456,
+      "step": 8330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7038270078285902e-05,
+      "loss": 1.0452,
+      "step": 8331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7037087941941987e-05,
+      "loss": 1.0964,
+      "step": 8332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7035905711268248e-05,
+      "loss": 1.056,
+      "step": 8333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.703472338628683e-05,
+      "loss": 1.0405,
+      "step": 8334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.703354096701987e-05,
+      "loss": 1.0964,
+      "step": 8335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7032358453489512e-05,
+      "loss": 1.1259,
+      "step": 8336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7031175845717907e-05,
+      "loss": 0.995,
+      "step": 8337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7029993143727195e-05,
+      "loss": 1.0966,
+      "step": 8338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7028810347539525e-05,
+      "loss": 0.9209,
+      "step": 8339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702762745717705e-05,
+      "loss": 1.1042,
+      "step": 8340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7026444472661922e-05,
+      "loss": 1.1698,
+      "step": 8341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70252613940163e-05,
+      "loss": 1.0558,
+      "step": 8342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7024078221262324e-05,
+      "loss": 1.0951,
+      "step": 8343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7022894954422166e-05,
+      "loss": 1.0784,
+      "step": 8344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.702171159351798e-05,
+      "loss": 1.0933,
+      "step": 8345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7020528138571924e-05,
+      "loss": 1.0668,
+      "step": 8346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7019344589606165e-05,
+      "loss": 1.0567,
+      "step": 8347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701816094664287e-05,
+      "loss": 1.0659,
+      "step": 8348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7016977209704196e-05,
+      "loss": 1.0725,
+      "step": 8349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7015793378812316e-05,
+      "loss": 1.0132,
+      "step": 8350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.70146094539894e-05,
+      "loss": 1.0218,
+      "step": 8351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701342543525762e-05,
+      "loss": 1.149,
+      "step": 8352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7012241322639147e-05,
+      "loss": 1.0433,
+      "step": 8353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.701105711615616e-05,
+      "loss": 1.0059,
+      "step": 8354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.700987281583083e-05,
+      "loss": 1.1227,
+      "step": 8355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.700868842168534e-05,
+      "loss": 1.0993,
+      "step": 8356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.700750393374187e-05,
+      "loss": 1.0492,
+      "step": 8357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7006319352022595e-05,
+      "loss": 1.0484,
+      "step": 8358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7005134676549712e-05,
+      "loss": 1.1073,
+      "step": 8359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7003949907345392e-05,
+      "loss": 1.0308,
+      "step": 8360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7002765044431832e-05,
+      "loss": 1.013,
+      "step": 8361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7001580087831217e-05,
+      "loss": 1.0161,
+      "step": 8362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.7000395037565742e-05,
+      "loss": 1.033,
+      "step": 8363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.699920989365759e-05,
+      "loss": 1.0428,
+      "step": 8364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6998024656128968e-05,
+      "loss": 1.0119,
+      "step": 8365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6996839325002062e-05,
+      "loss": 0.9818,
+      "step": 8366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6995653900299074e-05,
+      "loss": 1.0647,
+      "step": 8367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.69944683820422e-05,
+      "loss": 1.0589,
+      "step": 8368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6993282770253645e-05,
+      "loss": 0.9573,
+      "step": 8369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6992097064955614e-05,
+      "loss": 1.0838,
+      "step": 8370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.699091126617031e-05,
+      "loss": 1.106,
+      "step": 8371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6989725373919933e-05,
+      "loss": 1.0365,
+      "step": 8372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6988539388226698e-05,
+      "loss": 1.1034,
+      "step": 8373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6987353309112816e-05,
+      "loss": 1.0594,
+      "step": 8374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6986167136600495e-05,
+      "loss": 1.0419,
+      "step": 8375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698498087071195e-05,
+      "loss": 1.0315,
+      "step": 8376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6983794511469395e-05,
+      "loss": 1.0401,
+      "step": 8377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698260805889505e-05,
+      "loss": 1.0513,
+      "step": 8378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.698142151301113e-05,
+      "loss": 1.023,
+      "step": 8379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6980234873839856e-05,
+      "loss": 1.0361,
+      "step": 8380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6979048141403456e-05,
+      "loss": 1.0579,
+      "step": 8381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6977861315724147e-05,
+      "loss": 1.0684,
+      "step": 8382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6976674396824157e-05,
+      "loss": 1.0843,
+      "step": 8383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6975487384725717e-05,
+      "loss": 1.0815,
+      "step": 8384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697430027945105e-05,
+      "loss": 1.036,
+      "step": 8385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6973113081022395e-05,
+      "loss": 1.0862,
+      "step": 8386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6971925789461976e-05,
+      "loss": 1.0214,
+      "step": 8387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.697073840479203e-05,
+      "loss": 1.1022,
+      "step": 8388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6969550927034797e-05,
+      "loss": 1.0626,
+      "step": 8389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6968363356212513e-05,
+      "loss": 1.1381,
+      "step": 8390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6967175692347415e-05,
+      "loss": 1.0834,
+      "step": 8391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6965987935461748e-05,
+      "loss": 1.07,
+      "step": 8392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6964800085577755e-05,
+      "loss": 1.0192,
+      "step": 8393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696361214271768e-05,
+      "loss": 1.0831,
+      "step": 8394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.696242410690377e-05,
+      "loss": 0.981,
+      "step": 8395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6961235978158267e-05,
+      "loss": 1.0346,
+      "step": 8396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6960047756503433e-05,
+      "loss": 1.0856,
+      "step": 8397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.695885944196151e-05,
+      "loss": 1.0431,
+      "step": 8398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6957671034554756e-05,
+      "loss": 1.056,
+      "step": 8399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.695648253430543e-05,
+      "loss": 1.0983,
+      "step": 8400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6955293941235782e-05,
+      "loss": 1.0064,
+      "step": 8401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.6954105255368075e-05,
+      "loss": 1.061,
+      "step": 8402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6952916476724568e-05,
+      "loss": 1.0624,
+      "step": 8403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6951727605327525e-05,
+      "loss": 1.0339,
+      "step": 8404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6950538641199208e-05,
+      "loss": 0.9897,
+      "step": 8405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6949349584361882e-05,
+      "loss": 1.0047,
+      "step": 8406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6948160434837816e-05,
+      "loss": 1.0707,
+      "step": 8407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.694697119264928e-05,
+      "loss": 1.0465,
+      "step": 8408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6945781857818547e-05,
+      "loss": 1.0976,
+      "step": 8409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6944592430367884e-05,
+      "loss": 1.0317,
+      "step": 8410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6943402910319573e-05,
+      "loss": 1.0753,
+      "step": 8411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6942213297695886e-05,
+      "loss": 1.0945,
+      "step": 8412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6941023592519095e-05,
+      "loss": 1.046,
+      "step": 8413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693983379481149e-05,
+      "loss": 1.1637,
+      "step": 8414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6938643904595346e-05,
+      "loss": 1.0031,
+      "step": 8415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693745392189295e-05,
+      "loss": 1.044,
+      "step": 8416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693626384672658e-05,
+      "loss": 1.0342,
+      "step": 8417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693507367911853e-05,
+      "loss": 1.0337,
+      "step": 8418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6933883419091086e-05,
+      "loss": 1.0851,
+      "step": 8419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6932693066666544e-05,
+      "loss": 1.0871,
+      "step": 8420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.693150262186718e-05,
+      "loss": 1.1043,
+      "step": 8421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6930312084715305e-05,
+      "loss": 1.0543,
+      "step": 8422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6929121455233203e-05,
+      "loss": 1.0922,
+      "step": 8423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6927930733443172e-05,
+      "loss": 1.0603,
+      "step": 8424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6926739919367516e-05,
+      "loss": 1.0818,
+      "step": 8425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6925549013028532e-05,
+      "loss": 1.0684,
+      "step": 8426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6924358014448524e-05,
+      "loss": 1.0631,
+      "step": 8427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6923166923649792e-05,
+      "loss": 1.1338,
+      "step": 8428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.692197574065465e-05,
+      "loss": 1.0853,
+      "step": 8429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6920784465485393e-05,
+      "loss": 1.0899,
+      "step": 8430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691959309816434e-05,
+      "loss": 1.019,
+      "step": 8431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.69184016387138e-05,
+      "loss": 1.0048,
+      "step": 8432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691721008715608e-05,
+      "loss": 1.2013,
+      "step": 8433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6916018443513503e-05,
+      "loss": 1.0634,
+      "step": 8434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6914826707808375e-05,
+      "loss": 1.1166,
+      "step": 8435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691363488006302e-05,
+      "loss": 1.0504,
+      "step": 8436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691244296029976e-05,
+      "loss": 1.072,
+      "step": 8437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.691125094854091e-05,
+      "loss": 1.0086,
+      "step": 8438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6910058844808796e-05,
+      "loss": 1.0182,
+      "step": 8439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6908866649125746e-05,
+      "loss": 1.0156,
+      "step": 8440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6907674361514078e-05,
+      "loss": 1.0409,
+      "step": 8441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6906481981996124e-05,
+      "loss": 1.0707,
+      "step": 8442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6905289510594214e-05,
+      "loss": 1.0369,
+      "step": 8443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6904096947330683e-05,
+      "loss": 1.0351,
+      "step": 8444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6902904292227856e-05,
+      "loss": 1.0917,
+      "step": 8445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6901711545308076e-05,
+      "loss": 1.0247,
+      "step": 8446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6900518706593673e-05,
+      "loss": 1.0152,
+      "step": 8447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6899325776106993e-05,
+      "loss": 1.0492,
+      "step": 8448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689813275387037e-05,
+      "loss": 1.1055,
+      "step": 8449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689693963990615e-05,
+      "loss": 1.0681,
+      "step": 8450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689574643423667e-05,
+      "loss": 1.1074,
+      "step": 8451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689455313688428e-05,
+      "loss": 1.0294,
+      "step": 8452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6893359747871326e-05,
+      "loss": 1.1518,
+      "step": 8453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.689216626722016e-05,
+      "loss": 1.0799,
+      "step": 8454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6890972694953126e-05,
+      "loss": 1.0535,
+      "step": 8455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688977903109258e-05,
+      "loss": 1.0031,
+      "step": 8456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6888585275660874e-05,
+      "loss": 1.116,
+      "step": 8457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688739142868037e-05,
+      "loss": 1.0069,
+      "step": 8458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688619749017342e-05,
+      "loss": 1.0458,
+      "step": 8459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6885003460162378e-05,
+      "loss": 1.0125,
+      "step": 8460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6883809338669607e-05,
+      "loss": 1.0091,
+      "step": 8461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6882615125717478e-05,
+      "loss": 1.0607,
+      "step": 8462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688142082132835e-05,
+      "loss": 1.0854,
+      "step": 8463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.688022642552458e-05,
+      "loss": 0.9949,
+      "step": 8464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687903193832855e-05,
+      "loss": 1.123,
+      "step": 8465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.687783735976262e-05,
+      "loss": 1.0452,
+      "step": 8466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6876642689849167e-05,
+      "loss": 1.07,
+      "step": 8467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6875447928610555e-05,
+      "loss": 0.9845,
+      "step": 8468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6874253076069166e-05,
+      "loss": 1.0723,
+      "step": 8469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6873058132247377e-05,
+      "loss": 0.9641,
+      "step": 8470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6871863097167556e-05,
+      "loss": 1.011,
+      "step": 8471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6870667970852092e-05,
+      "loss": 1.0485,
+      "step": 8472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6869472753323363e-05,
+      "loss": 1.066,
+      "step": 8473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686827744460375e-05,
+      "loss": 0.9952,
+      "step": 8474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6867082044715646e-05,
+      "loss": 1.0564,
+      "step": 8475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6865886553681425e-05,
+      "loss": 1.0501,
+      "step": 8476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686469097152348e-05,
+      "loss": 1.0893,
+      "step": 8477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686349529826421e-05,
+      "loss": 1.0544,
+      "step": 8478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6862299533925992e-05,
+      "loss": 1.0391,
+      "step": 8479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.686110367853122e-05,
+      "loss": 1.087,
+      "step": 8480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6859907732102303e-05,
+      "loss": 1.0467,
+      "step": 8481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6858711694661622e-05,
+      "loss": 1.0799,
+      "step": 8482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6857515566231588e-05,
+      "loss": 1.0834,
+      "step": 8483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685631934683459e-05,
+      "loss": 1.1487,
+      "step": 8484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6855123036493036e-05,
+      "loss": 1.0986,
+      "step": 8485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685392663522933e-05,
+      "loss": 1.1147,
+      "step": 8486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685273014306587e-05,
+      "loss": 1.0422,
+      "step": 8487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6851533560025076e-05,
+      "loss": 0.9965,
+      "step": 8488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.685033688612934e-05,
+      "loss": 1.0718,
+      "step": 8489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6849140121401086e-05,
+      "loss": 1.0574,
+      "step": 8490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6847943265862713e-05,
+      "loss": 0.9768,
+      "step": 8491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.684674631953665e-05,
+      "loss": 1.0732,
+      "step": 8492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6845549282445302e-05,
+      "loss": 1.0764,
+      "step": 8493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6844352154611085e-05,
+      "loss": 1.0427,
+      "step": 8494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6843154936056424e-05,
+      "loss": 1.0544,
+      "step": 8495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6841957626803735e-05,
+      "loss": 1.016,
+      "step": 8496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6840760226875444e-05,
+      "loss": 1.0866,
+      "step": 8497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6839562736293967e-05,
+      "loss": 1.0942,
+      "step": 8498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6838365155081736e-05,
+      "loss": 0.9986,
+      "step": 8499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683716748326118e-05,
+      "loss": 1.0052,
+      "step": 8500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6835969720854716e-05,
+      "loss": 1.0505,
+      "step": 8501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6834771867884792e-05,
+      "loss": 0.9735,
+      "step": 8502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683357392437383e-05,
+      "loss": 1.1213,
+      "step": 8503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6832375890344263e-05,
+      "loss": 1.0559,
+      "step": 8504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.683117776581853e-05,
+      "loss": 1.0301,
+      "step": 8505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6829979550819067e-05,
+      "loss": 1.0728,
+      "step": 8506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6828781245368313e-05,
+      "loss": 1.0367,
+      "step": 8507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6827582849488708e-05,
+      "loss": 1.1145,
+      "step": 8508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6826384363202698e-05,
+      "loss": 1.0581,
+      "step": 8509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6825185786532723e-05,
+      "loss": 1.0575,
+      "step": 8510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682398711950123e-05,
+      "loss": 1.0434,
+      "step": 8511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682278836213067e-05,
+      "loss": 1.0455,
+      "step": 8512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6821589514443483e-05,
+      "loss": 1.0202,
+      "step": 8513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.682039057646213e-05,
+      "loss": 1.1086,
+      "step": 8514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.681919154820906e-05,
+      "loss": 1.0105,
+      "step": 8515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6817992429706728e-05,
+      "loss": 1.0436,
+      "step": 8516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.681679322097759e-05,
+      "loss": 1.0529,
+      "step": 8517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6815593922044098e-05,
+      "loss": 1.0115,
+      "step": 8518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6814394532928712e-05,
+      "loss": 1.0379,
+      "step": 8519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6813195053653903e-05,
+      "loss": 1.0438,
+      "step": 8520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6811995484242126e-05,
+      "loss": 1.1259,
+      "step": 8521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.681079582471585e-05,
+      "loss": 1.0368,
+      "step": 8522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6809596075097536e-05,
+      "loss": 1.0438,
+      "step": 8523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6808396235409652e-05,
+      "loss": 1.022,
+      "step": 8524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680719630567467e-05,
+      "loss": 1.0993,
+      "step": 8525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.680599628591506e-05,
+      "loss": 1.0028,
+      "step": 8526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.68047961761533e-05,
+      "loss": 1.0755,
+      "step": 8527
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6803595976411856e-05,
+      "loss": 1.0666,
+      "step": 8528
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6802395686713205e-05,
+      "loss": 1.0718,
+      "step": 8529
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6801195307079832e-05,
+      "loss": 1.0852,
+      "step": 8530
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679999483753421e-05,
+      "loss": 1.1569,
+      "step": 8531
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6798794278098826e-05,
+      "loss": 1.0025,
+      "step": 8532
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679759362879616e-05,
+      "loss": 0.9999,
+      "step": 8533
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679639288964869e-05,
+      "loss": 1.0723,
+      "step": 8534
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6795192060678912e-05,
+      "loss": 1.0959,
+      "step": 8535
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6793991141909312e-05,
+      "loss": 1.021,
+      "step": 8536
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6792790133362375e-05,
+      "loss": 1.0939,
+      "step": 8537
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6791589035060598e-05,
+      "loss": 1.0193,
+      "step": 8538
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.679038784702647e-05,
+      "loss": 1.119,
+      "step": 8539
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678918656928249e-05,
+      "loss": 1.027,
+      "step": 8540
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678798520185115e-05,
+      "loss": 1.0766,
+      "step": 8541
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6786783744754952e-05,
+      "loss": 1.056,
+      "step": 8542
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6785582198016392e-05,
+      "loss": 1.0944,
+      "step": 8543
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6784380561657973e-05,
+      "loss": 1.0262,
+      "step": 8544
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6783178835702196e-05,
+      "loss": 1.0756,
+      "step": 8545
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.678197702017157e-05,
+      "loss": 1.0601,
+      "step": 8546
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6780775115088598e-05,
+      "loss": 1.0768,
+      "step": 8547
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6779573120475794e-05,
+      "loss": 1.0351,
+      "step": 8548
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6778371036355656e-05,
+      "loss": 1.0761,
+      "step": 8549
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6777168862750706e-05,
+      "loss": 1.0633,
+      "step": 8550
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6775966599683457e-05,
+      "loss": 1.0287,
+      "step": 8551
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.677476424717642e-05,
+      "loss": 1.0452,
+      "step": 8552
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6773561805252107e-05,
+      "loss": 1.0747,
+      "step": 8553
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6772359273933047e-05,
+      "loss": 1.0824,
+      "step": 8554
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6771156653241752e-05,
+      "loss": 1.0236,
+      "step": 8555
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6769953943200747e-05,
+      "loss": 1.0274,
+      "step": 8556
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6768751143832554e-05,
+      "loss": 1.066,
+      "step": 8557
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6767548255159696e-05,
+      "loss": 0.9801,
+      "step": 8558
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6766345277204704e-05,
+      "loss": 1.0725,
+      "step": 8559
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67651422099901e-05,
+      "loss": 1.0112,
+      "step": 8560
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6763939053538422e-05,
+      "loss": 1.0146,
+      "step": 8561
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67627358078722e-05,
+      "loss": 1.0889,
+      "step": 8562
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.676153247301396e-05,
+      "loss": 1.0352,
+      "step": 8563
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6760329048986242e-05,
+      "loss": 1.053,
+      "step": 8564
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6759125535811584e-05,
+      "loss": 1.0206,
+      "step": 8565
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675792193351252e-05,
+      "loss": 0.9985,
+      "step": 8566
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6756718242111592e-05,
+      "loss": 1.0366,
+      "step": 8567
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675551446163134e-05,
+      "loss": 0.9725,
+      "step": 8568
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6754310592094313e-05,
+      "loss": 1.0401,
+      "step": 8569
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.675310663352305e-05,
+      "loss": 1.0876,
+      "step": 8570
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6751902585940103e-05,
+      "loss": 1.0821,
+      "step": 8571
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6750698449368014e-05,
+      "loss": 1.0654,
+      "step": 8572
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6749494223829337e-05,
+      "loss": 1.0632,
+      "step": 8573
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.674828990934662e-05,
+      "loss": 1.0211,
+      "step": 8574
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6747085505942422e-05,
+      "loss": 1.1153,
+      "step": 8575
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6745881013639294e-05,
+      "loss": 1.0566,
+      "step": 8576
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.674467643245979e-05,
+      "loss": 1.0478,
+      "step": 8577
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6743471762426472e-05,
+      "loss": 1.0228,
+      "step": 8578
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6742267003561906e-05,
+      "loss": 1.1257,
+      "step": 8579
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6741062155888642e-05,
+      "loss": 1.0695,
+      "step": 8580
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673985721942925e-05,
+      "loss": 1.0248,
+      "step": 8581
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673865219420629e-05,
+      "loss": 1.1035,
+      "step": 8582
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673744708024233e-05,
+      "loss": 1.0719,
+      "step": 8583
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6736241877559947e-05,
+      "loss": 1.0341,
+      "step": 8584
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.67350365861817e-05,
+      "loss": 1.0695,
+      "step": 8585
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6733831206130164e-05,
+      "loss": 1.0176,
+      "step": 8586
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6732625737427912e-05,
+      "loss": 1.0308,
+      "step": 8587
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.673142018009752e-05,
+      "loss": 1.0764,
+      "step": 8588
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6730214534161565e-05,
+      "loss": 1.0389,
+      "step": 8589
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6729008799642627e-05,
+      "loss": 1.0099,
+      "step": 8590
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.672780297656328e-05,
+      "loss": 1.0765,
+      "step": 8591
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6726597064946107e-05,
+      "loss": 1.0489,
+      "step": 8592
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6725391064813693e-05,
+      "loss": 1.0522,
+      "step": 8593
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6724184976188624e-05,
+      "loss": 1.1054,
+      "step": 8594
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6722978799093487e-05,
+      "loss": 1.0917,
+      "step": 8595
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6721772533550864e-05,
+      "loss": 1.0828,
+      "step": 8596
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6720566179583352e-05,
+      "loss": 1.0652,
+      "step": 8597
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6719359737213542e-05,
+      "loss": 1.0582,
+      "step": 8598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.671815320646402e-05,
+      "loss": 1.0615,
+      "step": 8599
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6716946587357387e-05,
+      "loss": 1.0126,
+      "step": 8600
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6715739879916237e-05,
+      "loss": 1.0633,
+      "step": 8601
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.671453308416317e-05,
+      "loss": 0.9934,
+      "step": 8602
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6713326200120783e-05,
+      "loss": 1.0481,
+      "step": 8603
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.671211922781168e-05,
+      "loss": 0.993,
+      "step": 8604
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6710912167258465e-05,
+      "loss": 1.0551,
+      "step": 8605
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6709705018483737e-05,
+      "loss": 1.0155,
+      "step": 8606
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6708497781510107e-05,
+      "loss": 1.0031,
+      "step": 8607
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6707290456360187e-05,
+      "loss": 1.0609,
+      "step": 8608
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6706083043056578e-05,
+      "loss": 1.0905,
+      "step": 8609
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6704875541621894e-05,
+      "loss": 1.0865,
+      "step": 8610
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6703667952078748e-05,
+      "loss": 1.0069,
+      "step": 8611
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6702460274449756e-05,
+      "loss": 1.0752,
+      "step": 8612
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6701252508757537e-05,
+      "loss": 1.1155,
+      "step": 8613
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6700044655024703e-05,
+      "loss": 1.0571,
+      "step": 8614
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6698836713273877e-05,
+      "loss": 1.0292,
+      "step": 8615
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6697628683527678e-05,
+      "loss": 1.1086,
+      "step": 8616
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6696420565808735e-05,
+      "loss": 1.1099,
+      "step": 8617
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.669521236013966e-05,
+      "loss": 1.0413,
+      "step": 8618
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6694004066543095e-05,
+      "loss": 1.0607,
+      "step": 8619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6692795685041654e-05,
+      "loss": 1.0459,
+      "step": 8620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6691587215657975e-05,
+      "loss": 1.0667,
+      "step": 8621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6690378658414683e-05,
+      "loss": 1.072,
+      "step": 8622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6689170013334417e-05,
+      "loss": 1.0275,
+      "step": 8623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6687961280439805e-05,
+      "loss": 1.0009,
+      "step": 8624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6686752459753488e-05,
+      "loss": 1.1593,
+      "step": 8625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6685543551298105e-05,
+      "loss": 1.007,
+      "step": 8626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6684334555096288e-05,
+      "loss": 0.9939,
+      "step": 8627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6683125471170684e-05,
+      "loss": 1.0499,
+      "step": 8628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6681916299543936e-05,
+      "loss": 1.0099,
+      "step": 8629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.668070704023868e-05,
+      "loss": 1.1011,
+      "step": 8630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6679497693277572e-05,
+      "loss": 0.9753,
+      "step": 8631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6678288258683252e-05,
+      "loss": 1.0525,
+      "step": 8632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6677078736478376e-05,
+      "loss": 1.0359,
+      "step": 8633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667586912668559e-05,
+      "loss": 1.0477,
+      "step": 8634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6674659429327546e-05,
+      "loss": 1.0686,
+      "step": 8635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.6673449644426904e-05,
+      "loss": 1.0971,
+      "step": 8636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667223977200631e-05,
+      "loss": 1.0323,
+      "step": 8637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.667102981208843e-05,
+      "loss": 1.0142,
+      "step": 8638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.666981976469592e-05,
+      "loss": 1.0809,
+      "step": 8639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666860962985144e-05,
+      "loss": 1.0411,
+      "step": 8640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666739940757765e-05,
+      "loss": 1.0225,
+      "step": 8641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666618909789722e-05,
+      "loss": 1.0129,
+      "step": 8642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666497870083281e-05,
+      "loss": 1.068,
+      "step": 8643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666376821640709e-05,
+      "loss": 1.0383,
+      "step": 8644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.666255764464273e-05,
+      "loss": 1.0421,
+      "step": 8645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6661346985562396e-05,
+      "loss": 1.0969,
+      "step": 8646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6660136239188764e-05,
+      "loss": 1.05,
+      "step": 8647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6658925405544504e-05,
+      "loss": 0.9871,
+      "step": 8648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6657714484652296e-05,
+      "loss": 1.0442,
+      "step": 8649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.665650347653481e-05,
+      "loss": 1.0674,
+      "step": 8650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6655292381214735e-05,
+      "loss": 1.0755,
+      "step": 8651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6654081198714744e-05,
+      "loss": 1.1284,
+      "step": 8652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.665286992905752e-05,
+      "loss": 1.0175,
+      "step": 8653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6651658572265746e-05,
+      "loss": 1.092,
+      "step": 8654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6650447128362108e-05,
+      "loss": 1.0152,
+      "step": 8655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6649235597369297e-05,
+      "loss": 1.1171,
+      "step": 8656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664802397930999e-05,
+      "loss": 1.0068,
+      "step": 8657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664681227420689e-05,
+      "loss": 1.0106,
+      "step": 8658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664560048208268e-05,
+      "loss": 1.0151,
+      "step": 8659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6644388602960054e-05,
+      "loss": 1.0473,
+      "step": 8660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6643176636861713e-05,
+      "loss": 1.0382,
+      "step": 8661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.664196458381035e-05,
+      "loss": 1.1033,
+      "step": 8662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6640752443828663e-05,
+      "loss": 1.0056,
+      "step": 8663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6639540216939353e-05,
+      "loss": 1.0583,
+      "step": 8664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6638327903165116e-05,
+      "loss": 1.0276,
+      "step": 8665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663711550252866e-05,
+      "loss": 1.0944,
+      "step": 8666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663590301505269e-05,
+      "loss": 1.0717,
+      "step": 8667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663469044075991e-05,
+      "loss": 0.9869,
+      "step": 8668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663347777967303e-05,
+      "loss": 1.0838,
+      "step": 8669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.663226503181476e-05,
+      "loss": 1.0543,
+      "step": 8670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6631052197207806e-05,
+      "loss": 1.0376,
+      "step": 8671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662983927587489e-05,
+      "loss": 1.0468,
+      "step": 8672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6628626267838716e-05,
+      "loss": 1.0662,
+      "step": 8673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662741317312201e-05,
+      "loss": 1.0825,
+      "step": 8674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6626199991747484e-05,
+      "loss": 1.1003,
+      "step": 8675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6624986723737855e-05,
+      "loss": 1.0306,
+      "step": 8676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.662377336911585e-05,
+      "loss": 1.0789,
+      "step": 8677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6622559927904187e-05,
+      "loss": 1.1138,
+      "step": 8678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6621346400125593e-05,
+      "loss": 1.0766,
+      "step": 8679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6620132785802793e-05,
+      "loss": 1.0923,
+      "step": 8680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6618919084958512e-05,
+      "loss": 1.0864,
+      "step": 8681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6617705297615487e-05,
+      "loss": 1.1134,
+      "step": 8682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6616491423796436e-05,
+      "loss": 1.0494,
+      "step": 8683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6615277463524103e-05,
+      "loss": 1.0887,
+      "step": 8684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6614063416821213e-05,
+      "loss": 1.0441,
+      "step": 8685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.661284928371051e-05,
+      "loss": 1.0756,
+      "step": 8686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6611635064214724e-05,
+      "loss": 1.0063,
+      "step": 8687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6610420758356598e-05,
+      "loss": 1.0684,
+      "step": 8688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6609206366158865e-05,
+      "loss": 1.0435,
+      "step": 8689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6607991887644277e-05,
+      "loss": 1.0044,
+      "step": 8690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6606777322835575e-05,
+      "loss": 1.0739,
+      "step": 8691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6605562671755503e-05,
+      "loss": 1.0255,
+      "step": 8692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6604347934426806e-05,
+      "loss": 1.086,
+      "step": 8693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6603133110872236e-05,
+      "loss": 1.062,
+      "step": 8694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6601918201114536e-05,
+      "loss": 1.033,
+      "step": 8695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6600703205176462e-05,
+      "loss": 1.0879,
+      "step": 8696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6599488123080775e-05,
+      "loss": 0.9931,
+      "step": 8697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6598272954850216e-05,
+      "loss": 1.0533,
+      "step": 8698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6597057700507553e-05,
+      "loss": 1.0625,
+      "step": 8699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6595842360075535e-05,
+      "loss": 1.1054,
+      "step": 8700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6594626933576936e-05,
+      "loss": 1.0403,
+      "step": 8701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65934114210345e-05,
+      "loss": 1.0362,
+      "step": 8702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6592195822470997e-05,
+      "loss": 1.0051,
+      "step": 8703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6590980137909194e-05,
+      "loss": 0.9993,
+      "step": 8704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6589764367371857e-05,
+      "loss": 1.0022,
+      "step": 8705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658854851088175e-05,
+      "loss": 1.0546,
+      "step": 8706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6587332568461643e-05,
+      "loss": 1.0589,
+      "step": 8707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.658611654013431e-05,
+      "loss": 1.1314,
+      "step": 8708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6584900425922523e-05,
+      "loss": 1.0251,
+      "step": 8709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6583684225849058e-05,
+      "loss": 1.0171,
+      "step": 8710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6582467939936685e-05,
+      "loss": 1.05,
+      "step": 8711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6581251568208186e-05,
+      "loss": 1.0315,
+      "step": 8712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6580035110686338e-05,
+      "loss": 1.1254,
+      "step": 8713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657881856739392e-05,
+      "loss": 1.1209,
+      "step": 8714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6577601938353722e-05,
+      "loss": 1.099,
+      "step": 8715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657638522358852e-05,
+      "loss": 1.0602,
+      "step": 8716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65751684231211e-05,
+      "loss": 1.0899,
+      "step": 8717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6573951536974253e-05,
+      "loss": 1.0786,
+      "step": 8718
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657273456517077e-05,
+      "loss": 1.0143,
+      "step": 8719
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6571517507733435e-05,
+      "loss": 1.0248,
+      "step": 8720
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.657030036468504e-05,
+      "loss": 0.9949,
+      "step": 8721
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6569083136048383e-05,
+      "loss": 1.1251,
+      "step": 8722
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6567865821846252e-05,
+      "loss": 1.0453,
+      "step": 8723
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656664842210145e-05,
+      "loss": 1.0519,
+      "step": 8724
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6565430936836774e-05,
+      "loss": 1.0255,
+      "step": 8725
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6564213366075028e-05,
+      "loss": 1.1036,
+      "step": 8726
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6562995709839004e-05,
+      "loss": 1.0458,
+      "step": 8727
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6561777968151515e-05,
+      "loss": 1.0445,
+      "step": 8728
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.656056014103536e-05,
+      "loss": 1.0321,
+      "step": 8729
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6559342228513342e-05,
+      "loss": 1.0963,
+      "step": 8730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6558124230608276e-05,
+      "loss": 1.1263,
+      "step": 8731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6556906147342967e-05,
+      "loss": 1.0212,
+      "step": 8732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.655568797874023e-05,
+      "loss": 1.0393,
+      "step": 8733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6554469724822875e-05,
+      "loss": 1.0887,
+      "step": 8734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6553251385613717e-05,
+      "loss": 1.0921,
+      "step": 8735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6552032961135574e-05,
+      "loss": 1.0601,
+      "step": 8736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6550814451411257e-05,
+      "loss": 1.0818,
+      "step": 8737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6549595856463593e-05,
+      "loss": 1.0772,
+      "step": 8738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.65483771763154e-05,
+      "loss": 1.0362,
+      "step": 8739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6547158410989495e-05,
+      "loss": 1.0623,
+      "step": 8740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6545939560508708e-05,
+      "loss": 1.1039,
+      "step": 8741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6544720624895862e-05,
+      "loss": 1.0957,
+      "step": 8742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6543501604173786e-05,
+      "loss": 1.1292,
+      "step": 8743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.654228249836531e-05,
+      "loss": 1.0472,
+      "step": 8744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6541063307493257e-05,
+      "loss": 1.0638,
+      "step": 8745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6539844031580462e-05,
+      "loss": 1.1192,
+      "step": 8746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6538624670649768e-05,
+      "loss": 1.0673,
+      "step": 8747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6537405224723996e-05,
+      "loss": 1.1421,
+      "step": 8748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653618569382599e-05,
+      "loss": 1.1102,
+      "step": 8749
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653496607797859e-05,
+      "loss": 1.0796,
+      "step": 8750
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.653374637720463e-05,
+      "loss": 1.0588,
+      "step": 8751
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6532526591526952e-05,
+      "loss": 1.0043,
+      "step": 8752
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6531306720968402e-05,
+      "loss": 1.0871,
+      "step": 8753
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6530086765551825e-05,
+      "loss": 1.025,
+      "step": 8754
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6528866725300066e-05,
+      "loss": 1.0981,
+      "step": 8755
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6527646600235973e-05,
+      "loss": 1.0571,
+      "step": 8756
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6526426390382397e-05,
+      "loss": 1.0269,
+      "step": 8757
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6525206095762183e-05,
+      "loss": 1.0317,
+      "step": 8758
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.652398571639819e-05,
+      "loss": 1.0843,
+      "step": 8759
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6522765252313268e-05,
+      "loss": 1.1157,
+      "step": 8760
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6521544703530273e-05,
+      "loss": 1.0111,
+      "step": 8761
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6520324070072065e-05,
+      "loss": 1.074,
+      "step": 8762
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6519103351961503e-05,
+      "loss": 1.1141,
+      "step": 8763
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6517882549221442e-05,
+      "loss": 1.1494,
+      "step": 8764
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651666166187475e-05,
+      "loss": 1.0716,
+      "step": 8765
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651544068994429e-05,
+      "loss": 1.0949,
+      "step": 8766
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6514219633452926e-05,
+      "loss": 1.0153,
+      "step": 8767
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6512998492423523e-05,
+      "loss": 0.976,
+      "step": 8768
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651177726687895e-05,
+      "loss": 1.0708,
+      "step": 8769
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.651055595684208e-05,
+      "loss": 1.0242,
+      "step": 8770
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6509334562335783e-05,
+      "loss": 1.0229,
+      "step": 8771
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6508113083382927e-05,
+      "loss": 1.125,
+      "step": 8772
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6506891520006396e-05,
+      "loss": 1.0482,
+      "step": 8773
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650566987222906e-05,
+      "loss": 1.0519,
+      "step": 8774
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6504448140073797e-05,
+      "loss": 1.1005,
+      "step": 8775
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6503226323563492e-05,
+      "loss": 1.0867,
+      "step": 8776
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6502004422721015e-05,
+      "loss": 1.1049,
+      "step": 8777
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.650078243756926e-05,
+      "loss": 1.0866,
+      "step": 8778
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6499560368131108e-05,
+      "loss": 1.1123,
+      "step": 8779
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6498338214429438e-05,
+      "loss": 1.0768,
+      "step": 8780
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6497115976487147e-05,
+      "loss": 1.0797,
+      "step": 8781
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6495893654327117e-05,
+      "loss": 1.0525,
+      "step": 8782
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6494671247972243e-05,
+      "loss": 1.0676,
+      "step": 8783
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6493448757445418e-05,
+      "loss": 1.0233,
+      "step": 8784
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6492226182769526e-05,
+      "loss": 1.113,
+      "step": 8785
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6491003523967476e-05,
+      "loss": 1.0838,
+      "step": 8786
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6489780781062153e-05,
+      "loss": 1.0192,
+      "step": 8787
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648855795407646e-05,
+      "loss": 1.0576,
+      "step": 8788
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6487335043033302e-05,
+      "loss": 1.0871,
+      "step": 8789
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648611204795557e-05,
+      "loss": 1.1113,
+      "step": 8790
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6484888968866178e-05,
+      "loss": 1.0422,
+      "step": 8791
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6483665805788023e-05,
+      "loss": 1.0301,
+      "step": 8792
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648244255874401e-05,
+      "loss": 1.085,
+      "step": 8793
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.648121922775706e-05,
+      "loss": 1.0858,
+      "step": 8794
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6479995812850066e-05,
+      "loss": 1.1129,
+      "step": 8795
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6478772314045948e-05,
+      "loss": 1.1016,
+      "step": 8796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6477548731367614e-05,
+      "loss": 0.985,
+      "step": 8797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6476325064837985e-05,
+      "loss": 1.0567,
+      "step": 8798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647510131447997e-05,
+      "loss": 1.0424,
+      "step": 8799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.647387748031649e-05,
+      "loss": 1.0611,
+      "step": 8800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6472653562370456e-05,
+      "loss": 1.0261,
+      "step": 8801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6471429560664798e-05,
+      "loss": 1.0181,
+      "step": 8802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6470205475222435e-05,
+      "loss": 1.1226,
+      "step": 8803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6468981306066295e-05,
+      "loss": 1.0378,
+      "step": 8804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646775705321929e-05,
+      "loss": 1.06,
+      "step": 8805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646653271670436e-05,
+      "loss": 1.057,
+      "step": 8806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6465308296544424e-05,
+      "loss": 1.0625,
+      "step": 8807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646408379276242e-05,
+      "loss": 1.0312,
+      "step": 8808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6462859205381274e-05,
+      "loss": 1.1023,
+      "step": 8809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.646163453442392e-05,
+      "loss": 1.0728,
+      "step": 8810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6460409779913288e-05,
+      "loss": 1.0515,
+      "step": 8811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6459184941872326e-05,
+      "loss": 1.0593,
+      "step": 8812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.645796002032396e-05,
+      "loss": 1.0266,
+      "step": 8813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6456735015291137e-05,
+      "loss": 1.0561,
+      "step": 8814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.645550992679679e-05,
+      "loss": 1.0816,
+      "step": 8815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6454284754863866e-05,
+      "loss": 1.1034,
+      "step": 8816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6453059499515307e-05,
+      "loss": 1.0709,
+      "step": 8817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6451834160774067e-05,
+      "loss": 1.082,
+      "step": 8818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6450608738663078e-05,
+      "loss": 1.035,
+      "step": 8819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6449383233205296e-05,
+      "loss": 1.095,
+      "step": 8820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6448157644423674e-05,
+      "loss": 1.0426,
+      "step": 8821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6446931972341163e-05,
+      "loss": 1.0053,
+      "step": 8822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6445706216980714e-05,
+      "loss": 1.065,
+      "step": 8823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644448037836528e-05,
+      "loss": 1.0351,
+      "step": 8824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644325445651782e-05,
+      "loss": 1.0602,
+      "step": 8825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644202845146129e-05,
+      "loss": 1.0631,
+      "step": 8826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.644080236321865e-05,
+      "loss": 1.0624,
+      "step": 8827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6439576191812866e-05,
+      "loss": 1.0713,
+      "step": 8828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643834993726689e-05,
+      "loss": 1.0295,
+      "step": 8829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64371235996037e-05,
+      "loss": 1.0952,
+      "step": 8830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643589717884625e-05,
+      "loss": 1.0828,
+      "step": 8831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643467067501751e-05,
+      "loss": 1.0204,
+      "step": 8832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6433444088140454e-05,
+      "loss": 1.1624,
+      "step": 8833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6432217418238047e-05,
+      "loss": 1.0893,
+      "step": 8834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.643099066533326e-05,
+      "loss": 1.0252,
+      "step": 8835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642976382944907e-05,
+      "loss": 1.1654,
+      "step": 8836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642853691060845e-05,
+      "loss": 1.0562,
+      "step": 8837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.642730990883438e-05,
+      "loss": 1.0889,
+      "step": 8838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6426082824149836e-05,
+      "loss": 1.0948,
+      "step": 8839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6424855656577795e-05,
+      "loss": 1.0798,
+      "step": 8840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6423628406141246e-05,
+      "loss": 1.0564,
+      "step": 8841
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6422401072863163e-05,
+      "loss": 1.1142,
+      "step": 8842
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6421173656766536e-05,
+      "loss": 1.0402,
+      "step": 8843
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6419946157874342e-05,
+      "loss": 1.1116,
+      "step": 8844
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641871857620958e-05,
+      "loss": 1.0705,
+      "step": 8845
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6417490911795232e-05,
+      "loss": 1.0265,
+      "step": 8846
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6416263164654294e-05,
+      "loss": 1.0435,
+      "step": 8847
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641503533480975e-05,
+      "loss": 1.0924,
+      "step": 8848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.64138074222846e-05,
+      "loss": 1.0588,
+      "step": 8849
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641257942710184e-05,
+      "loss": 1.0327,
+      "step": 8850
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641135134928446e-05,
+      "loss": 1.0425,
+      "step": 8851
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.641012318885547e-05,
+      "loss": 1.0379,
+      "step": 8852
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6408894945837855e-05,
+      "loss": 1.1029,
+      "step": 8853
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6407666620254627e-05,
+      "loss": 1.0829,
+      "step": 8854
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6406438212128783e-05,
+      "loss": 1.0461,
+      "step": 8855
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640520972148333e-05,
+      "loss": 0.9546,
+      "step": 8856
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640398114834128e-05,
+      "loss": 1.0593,
+      "step": 8857
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640275249272563e-05,
+      "loss": 1.0607,
+      "step": 8858
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640152375465939e-05,
+      "loss": 1.0179,
+      "step": 8859
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.640029493416558e-05,
+      "loss": 1.1054,
+      "step": 8860
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6399066031267205e-05,
+      "loss": 1.0493,
+      "step": 8861
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639783704598728e-05,
+      "loss": 1.0951,
+      "step": 8862
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639660797834882e-05,
+      "loss": 1.0513,
+      "step": 8863
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6395378828374846e-05,
+      "loss": 1.0665,
+      "step": 8864
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.639414959608837e-05,
+      "loss": 1.0165,
+      "step": 8865
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6392920281512415e-05,
+      "loss": 1.0322,
+      "step": 8866
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6391690884670003e-05,
+      "loss": 1.081,
+      "step": 8867
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6390461405584154e-05,
+      "loss": 1.0443,
+      "step": 8868
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6389231844277895e-05,
+      "loss": 0.9794,
+      "step": 8869
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6388002200774254e-05,
+      "loss": 1.0539,
+      "step": 8870
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6386772475096254e-05,
+      "loss": 1.015,
+      "step": 8871
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6385542667266927e-05,
+      "loss": 1.0534,
+      "step": 8872
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6384312777309303e-05,
+      "loss": 1.1014,
+      "step": 8873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6383082805246416e-05,
+      "loss": 1.0753,
+      "step": 8874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.6381852751101294e-05,
+      "loss": 1.0613,
+      "step": 8875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6380622614896982e-05,
+      "loss": 1.0398,
+      "step": 8876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6379392396656506e-05,
+      "loss": 1.1381,
+      "step": 8877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637816209640291e-05,
+      "loss": 1.0258,
+      "step": 8878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6376931714159238e-05,
+      "loss": 1.0189,
+      "step": 8879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6375701249948526e-05,
+      "loss": 1.0594,
+      "step": 8880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6374470703793816e-05,
+      "loss": 1.0129,
+      "step": 8881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6373240075718153e-05,
+      "loss": 1.0713,
+      "step": 8882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6372009365744588e-05,
+      "loss": 1.0591,
+      "step": 8883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.637077857389616e-05,
+      "loss": 1.0567,
+      "step": 8884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6369547700195925e-05,
+      "loss": 1.0808,
+      "step": 8885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6368316744666936e-05,
+      "loss": 1.1616,
+      "step": 8886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6367085707332237e-05,
+      "loss": 1.0736,
+      "step": 8887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6365854588214884e-05,
+      "loss": 1.1563,
+      "step": 8888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6364623387337937e-05,
+      "loss": 1.0936,
+      "step": 8889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6363392104724448e-05,
+      "loss": 1.0489,
+      "step": 8890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.636216074039748e-05,
+      "loss": 1.0746,
+      "step": 8891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6360929294380087e-05,
+      "loss": 1.0558,
+      "step": 8892
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6359697766695332e-05,
+      "loss": 1.1378,
+      "step": 8893
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6358466157366277e-05,
+      "loss": 1.0667,
+      "step": 8894
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635723446641599e-05,
+      "loss": 1.0781,
+      "step": 8895
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6356002693867543e-05,
+      "loss": 1.0639,
+      "step": 8896
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.635477083974399e-05,
+      "loss": 1.023,
+      "step": 8897
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6353538904068405e-05,
+      "loss": 1.0955,
+      "step": 8898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6352306886863857e-05,
+      "loss": 1.0585,
+      "step": 8899
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6351074788153426e-05,
+      "loss": 1.0871,
+      "step": 8900
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6349842607960177e-05,
+      "loss": 1.0998,
+      "step": 8901
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6348610346307186e-05,
+      "loss": 1.0031,
+      "step": 8902
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6347378003217534e-05,
+      "loss": 1.0868,
+      "step": 8903
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6346145578714298e-05,
+      "loss": 1.1223,
+      "step": 8904
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6344913072820556e-05,
+      "loss": 1.0394,
+      "step": 8905
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.634368048555939e-05,
+      "loss": 1.0967,
+      "step": 8906
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6342447816953882e-05,
+      "loss": 1.0295,
+      "step": 8907
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6341215067027116e-05,
+      "loss": 1.0824,
+      "step": 8908
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633998223580218e-05,
+      "loss": 1.1121,
+      "step": 8909
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6338749323302156e-05,
+      "loss": 1.0188,
+      "step": 8910
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633751632955014e-05,
+      "loss": 1.0921,
+      "step": 8911
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6336283254569218e-05,
+      "loss": 1.1356,
+      "step": 8912
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.633505009838248e-05,
+      "loss": 1.014,
+      "step": 8913
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6333816861013023e-05,
+      "loss": 1.1114,
+      "step": 8914
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6332583542483945e-05,
+      "loss": 1.1007,
+      "step": 8915
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6331350142818338e-05,
+      "loss": 1.1272,
+      "step": 8916
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63301166620393e-05,
+      "loss": 1.0795,
+      "step": 8917
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.632888310016993e-05,
+      "loss": 1.0961,
+      "step": 8918
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6327649457233328e-05,
+      "loss": 1.0741,
+      "step": 8919
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63264157332526e-05,
+      "loss": 1.0444,
+      "step": 8920
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6325181928250853e-05,
+      "loss": 1.12,
+      "step": 8921
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6323948042251182e-05,
+      "loss": 1.0865,
+      "step": 8922
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63227140752767e-05,
+      "loss": 1.0123,
+      "step": 8923
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.632148002735052e-05,
+      "loss": 1.1129,
+      "step": 8924
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6320245898495745e-05,
+      "loss": 0.9975,
+      "step": 8925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.631901168873549e-05,
+      "loss": 0.9936,
+      "step": 8926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6317777398092866e-05,
+      "loss": 1.0382,
+      "step": 8927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.631654302659099e-05,
+      "loss": 1.0628,
+      "step": 8928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6315308574252974e-05,
+      "loss": 1.001,
+      "step": 8929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6314074041101944e-05,
+      "loss": 0.9821,
+      "step": 8930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.631283942716101e-05,
+      "loss": 1.1006,
+      "step": 8931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.63116047324533e-05,
+      "loss": 1.0858,
+      "step": 8932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6310369957001928e-05,
+      "loss": 1.0668,
+      "step": 8933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6309135100830028e-05,
+      "loss": 1.0271,
+      "step": 8934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6307900163960717e-05,
+      "loss": 1.033,
+      "step": 8935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6306665146417124e-05,
+      "loss": 1.0058,
+      "step": 8936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.630543004822238e-05,
+      "loss": 1.035,
+      "step": 8937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.630419486939961e-05,
+      "loss": 1.0315,
+      "step": 8938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6302959609971947e-05,
+      "loss": 1.0805,
+      "step": 8939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.630172426996253e-05,
+      "loss": 1.0973,
+      "step": 8940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6300488849394483e-05,
+      "loss": 1.0713,
+      "step": 8941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629925334829095e-05,
+      "loss": 1.0005,
+      "step": 8942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629801776667506e-05,
+      "loss": 1.1102,
+      "step": 8943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6296782104569958e-05,
+      "loss": 1.0815,
+      "step": 8944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6295546361998787e-05,
+      "loss": 1.1093,
+      "step": 8945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629431053898468e-05,
+      "loss": 1.0247,
+      "step": 8946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629307463555079e-05,
+      "loss": 1.0928,
+      "step": 8947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629183865172025e-05,
+      "loss": 1.058,
+      "step": 8948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.629060258751622e-05,
+      "loss": 1.0881,
+      "step": 8949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628936644296184e-05,
+      "loss": 1.0949,
+      "step": 8950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6288130218080256e-05,
+      "loss": 1.032,
+      "step": 8951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6286893912894627e-05,
+      "loss": 1.1049,
+      "step": 8952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6285657527428102e-05,
+      "loss": 1.0644,
+      "step": 8953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6284421061703836e-05,
+      "loss": 1.0214,
+      "step": 8954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.628318451574498e-05,
+      "loss": 1.0011,
+      "step": 8955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6281947889574695e-05,
+      "loss": 1.0476,
+      "step": 8956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6280711183216135e-05,
+      "loss": 1.0663,
+      "step": 8957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6279474396692466e-05,
+      "loss": 0.9971,
+      "step": 8958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627823753002685e-05,
+      "loss": 1.0341,
+      "step": 8959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627700058324244e-05,
+      "loss": 1.0692,
+      "step": 8960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6275763556362407e-05,
+      "loss": 1.0951,
+      "step": 8961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.627452644940992e-05,
+      "loss": 1.0413,
+      "step": 8962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6273289262408143e-05,
+      "loss": 1.0018,
+      "step": 8963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6272051995380247e-05,
+      "loss": 1.0365,
+      "step": 8964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6270814648349394e-05,
+      "loss": 1.1059,
+      "step": 8965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6269577221338764e-05,
+      "loss": 1.0212,
+      "step": 8966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6268339714371526e-05,
+      "loss": 0.9982,
+      "step": 8967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626710212747086e-05,
+      "loss": 1.0368,
+      "step": 8968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626586446065994e-05,
+      "loss": 1.0448,
+      "step": 8969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626462671396194e-05,
+      "loss": 1.0123,
+      "step": 8970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.626338888740005e-05,
+      "loss": 1.1082,
+      "step": 8971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6262150980997437e-05,
+      "loss": 1.0327,
+      "step": 8972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6260912994777294e-05,
+      "loss": 1.0673,
+      "step": 8973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62596749287628e-05,
+      "loss": 1.0425,
+      "step": 8974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625843678297714e-05,
+      "loss": 1.044,
+      "step": 8975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62571985574435e-05,
+      "loss": 1.0236,
+      "step": 8976
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6255960252185072e-05,
+      "loss": 1.0263,
+      "step": 8977
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6254721867225047e-05,
+      "loss": 1.0169,
+      "step": 8978
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.625348340258661e-05,
+      "loss": 1.0459,
+      "step": 8979
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6252244858292956e-05,
+      "loss": 1.1029,
+      "step": 8980
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6251006234367285e-05,
+      "loss": 1.042,
+      "step": 8981
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6249767530832785e-05,
+      "loss": 1.0303,
+      "step": 8982
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6248528747712658e-05,
+      "loss": 1.0488,
+      "step": 8983
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62472898850301e-05,
+      "loss": 1.0712,
+      "step": 8984
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.624605094280831e-05,
+      "loss": 1.0313,
+      "step": 8985
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6244811921070495e-05,
+      "loss": 0.9767,
+      "step": 8986
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6243572819839854e-05,
+      "loss": 1.0302,
+      "step": 8987
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6242333639139595e-05,
+      "loss": 1.046,
+      "step": 8988
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6241094378992925e-05,
+      "loss": 1.0408,
+      "step": 8989
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6239855039423045e-05,
+      "loss": 1.1562,
+      "step": 8990
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.623861562045317e-05,
+      "loss": 1.0689,
+      "step": 8991
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6237376122106507e-05,
+      "loss": 1.0876,
+      "step": 8992
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6236136544406273e-05,
+      "loss": 1.0771,
+      "step": 8993
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6234896887375675e-05,
+      "loss": 1.0498,
+      "step": 8994
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6233657151037934e-05,
+      "loss": 1.1017,
+      "step": 8995
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6232417335416266e-05,
+      "loss": 1.0882,
+      "step": 8996
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6231177440533882e-05,
+      "loss": 1.0344,
+      "step": 8997
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6229937466414013e-05,
+      "loss": 1.0366,
+      "step": 8998
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6228697413079873e-05,
+      "loss": 1.0364,
+      "step": 8999
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6227457280554686e-05,
+      "loss": 1.0952,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "eval_text_loss": 0.9748941659927368,
+      "eval_text_runtime": 1.5567,
+      "eval_text_samples_per_second": 47.536,
+      "eval_text_steps_per_second": 1.285,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "eval_svit_loss": 1.1363624334335327,
+      "eval_svit_runtime": 4.1429,
+      "eval_svit_samples_per_second": 164.379,
+      "eval_svit_steps_per_second": 2.655,
+      "step": 9000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6226217068861675e-05,
+      "loss": 0.9893,
+      "step": 9001
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6224976778024064e-05,
+      "loss": 0.9907,
+      "step": 9002
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.622373640806509e-05,
+      "loss": 1.0802,
+      "step": 9003
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6222495959007964e-05,
+      "loss": 1.1269,
+      "step": 9004
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.622125543087593e-05,
+      "loss": 1.0455,
+      "step": 9005
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6220014823692215e-05,
+      "loss": 1.0304,
+      "step": 9006
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.621877413748005e-05,
+      "loss": 0.9729,
+      "step": 9007
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6217533372262675e-05,
+      "loss": 1.0573,
+      "step": 9008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6216292528063322e-05,
+      "loss": 1.0718,
+      "step": 9009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6215051604905224e-05,
+      "loss": 1.1165,
+      "step": 9010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6213810602811632e-05,
+      "loss": 1.0807,
+      "step": 9011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6212569521805773e-05,
+      "loss": 0.9869,
+      "step": 9012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.62113283619109e-05,
+      "loss": 1.0676,
+      "step": 9013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6210087123150246e-05,
+      "loss": 1.075,
+      "step": 9014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620884580554706e-05,
+      "loss": 1.0729,
+      "step": 9015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620760440912459e-05,
+      "loss": 1.1581,
+      "step": 9016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6206362933906083e-05,
+      "loss": 1.0089,
+      "step": 9017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6205121379914788e-05,
+      "loss": 1.0192,
+      "step": 9018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6203879747173958e-05,
+      "loss": 1.0298,
+      "step": 9019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620263803570684e-05,
+      "loss": 1.0662,
+      "step": 9020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.620139624553669e-05,
+      "loss": 1.0743,
+      "step": 9021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6200154376686765e-05,
+      "loss": 1.0129,
+      "step": 9022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619891242918032e-05,
+      "loss": 1.1211,
+      "step": 9023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6197670403040608e-05,
+      "loss": 1.0947,
+      "step": 9024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6196428298290895e-05,
+      "loss": 1.0628,
+      "step": 9025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619518611495444e-05,
+      "loss": 1.0398,
+      "step": 9026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619394385305451e-05,
+      "loss": 1.0696,
+      "step": 9027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6192701512614363e-05,
+      "loss": 1.0746,
+      "step": 9028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6191459093657263e-05,
+      "loss": 1.0426,
+      "step": 9029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.619021659620648e-05,
+      "loss": 1.0862,
+      "step": 9030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6188974020285286e-05,
+      "loss": 1.0395,
+      "step": 9031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6187731365916945e-05,
+      "loss": 1.0676,
+      "step": 9032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618648863312473e-05,
+      "loss": 1.0772,
+      "step": 9033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618524582193191e-05,
+      "loss": 1.0535,
+      "step": 9034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6184002932361767e-05,
+      "loss": 1.0444,
+      "step": 9035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6182759964437573e-05,
+      "loss": 1.0413,
+      "step": 9036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6181516918182605e-05,
+      "loss": 1.1309,
+      "step": 9037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.618027379362014e-05,
+      "loss": 1.0772,
+      "step": 9038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617903059077346e-05,
+      "loss": 1.0771,
+      "step": 9039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6177787309665844e-05,
+      "loss": 1.0495,
+      "step": 9040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617654395032058e-05,
+      "loss": 1.0371,
+      "step": 9041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6175300512760947e-05,
+      "loss": 1.0105,
+      "step": 9042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617405699701024e-05,
+      "loss": 1.0716,
+      "step": 9043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617281340309173e-05,
+      "loss": 1.1273,
+      "step": 9044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.617156973102872e-05,
+      "loss": 1.0675,
+      "step": 9045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6170325980844497e-05,
+      "loss": 1.0384,
+      "step": 9046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6169082152562355e-05,
+      "loss": 1.018,
+      "step": 9047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6167838246205578e-05,
+      "loss": 1.0693,
+      "step": 9048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616659426179747e-05,
+      "loss": 1.0727,
+      "step": 9049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616535019936132e-05,
+      "loss": 1.0345,
+      "step": 9050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616410605892043e-05,
+      "loss": 1.0418,
+      "step": 9051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6162861840498102e-05,
+      "loss": 1.0862,
+      "step": 9052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6161617544117632e-05,
+      "loss": 1.0631,
+      "step": 9053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.616037316980232e-05,
+      "loss": 1.0806,
+      "step": 9054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615912871757548e-05,
+      "loss": 1.0677,
+      "step": 9055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61578841874604e-05,
+      "loss": 1.0682,
+      "step": 9056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6156639579480403e-05,
+      "loss": 0.9881,
+      "step": 9057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615539489365879e-05,
+      "loss": 1.0515,
+      "step": 9058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6154150130018868e-05,
+      "loss": 1.1052,
+      "step": 9059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615290528858395e-05,
+      "loss": 1.0116,
+      "step": 9060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6151660369377343e-05,
+      "loss": 1.0249,
+      "step": 9061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.615041537242237e-05,
+      "loss": 1.0324,
+      "step": 9062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6149170297742343e-05,
+      "loss": 1.0743,
+      "step": 9063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6147925145360574e-05,
+      "loss": 1.0591,
+      "step": 9064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.614667991530039e-05,
+      "loss": 1.0817,
+      "step": 9065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61454346075851e-05,
+      "loss": 1.0586,
+      "step": 9066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6144189222238034e-05,
+      "loss": 1.0159,
+      "step": 9067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6142943759282504e-05,
+      "loss": 1.0815,
+      "step": 9068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6141698218741843e-05,
+      "loss": 1.08,
+      "step": 9069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6140452600639375e-05,
+      "loss": 1.0044,
+      "step": 9070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6139206904998423e-05,
+      "loss": 1.0728,
+      "step": 9071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.613796113184232e-05,
+      "loss": 1.0436,
+      "step": 9072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6136715281194388e-05,
+      "loss": 1.057,
+      "step": 9073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.613546935307797e-05,
+      "loss": 1.0259,
+      "step": 9074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6134223347516387e-05,
+      "loss": 1.07,
+      "step": 9075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.613297726453298e-05,
+      "loss": 1.0495,
+      "step": 9076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.613173110415108e-05,
+      "loss": 1.038,
+      "step": 9077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6130484866394027e-05,
+      "loss": 1.0273,
+      "step": 9078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6129238551285158e-05,
+      "loss": 0.9692,
+      "step": 9079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6127992158847816e-05,
+      "loss": 1.0973,
+      "step": 9080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612674568910534e-05,
+      "loss": 1.1238,
+      "step": 9081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6125499142081067e-05,
+      "loss": 1.0513,
+      "step": 9082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6124252517798352e-05,
+      "loss": 1.0209,
+      "step": 9083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.612300581628053e-05,
+      "loss": 1.0354,
+      "step": 9084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6121759037550956e-05,
+      "loss": 1.0639,
+      "step": 9085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6120512181632977e-05,
+      "loss": 0.9443,
+      "step": 9086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6119265248549937e-05,
+      "loss": 1.0166,
+      "step": 9087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6118018238325195e-05,
+      "loss": 1.0608,
+      "step": 9088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.61167711509821e-05,
+      "loss": 1.0113,
+      "step": 9089
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6115523986544008e-05,
+      "loss": 1.0097,
+      "step": 9090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6114276745034267e-05,
+      "loss": 1.1354,
+      "step": 9091
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.611302942647625e-05,
+      "loss": 1.0533,
+      "step": 9092
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6111782030893302e-05,
+      "loss": 1.0802,
+      "step": 9093
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6110534558308788e-05,
+      "loss": 1.0625,
+      "step": 9094
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6109287008746072e-05,
+      "loss": 1.0906,
+      "step": 9095
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610803938222851e-05,
+      "loss": 1.0075,
+      "step": 9096
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610679167877947e-05,
+      "loss": 1.0953,
+      "step": 9097
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610554389842232e-05,
+      "loss": 1.0819,
+      "step": 9098
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.610429604118042e-05,
+      "loss": 1.0458,
+      "step": 9099
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6103048107077154e-05,
+      "loss": 1.0979,
+      "step": 9100
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6101800096135873e-05,
+      "loss": 1.0366,
+      "step": 9101
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6100552008379963e-05,
+      "loss": 1.0069,
+      "step": 9102
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6099303843832794e-05,
+      "loss": 1.0216,
+      "step": 9103
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6098055602517732e-05,
+      "loss": 1.0148,
+      "step": 9104
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6096807284458164e-05,
+      "loss": 1.1137,
+      "step": 9105
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6095558889677455e-05,
+      "loss": 1.0502,
+      "step": 9106
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6094310418198997e-05,
+      "loss": 1.0844,
+      "step": 9107
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.609306187004616e-05,
+      "loss": 1.0841,
+      "step": 9108
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6091813245242336e-05,
+      "loss": 0.9863,
+      "step": 9109
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.60905645438109e-05,
+      "loss": 1.0695,
+      "step": 9110
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.6089315765775236e-05,
+      "loss": 1.0252,
+      "step": 9111
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.608806691115874e-05,
+      "loss": 1.0561,
+      "step": 9112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6086817979984783e-05,
+      "loss": 1.0547,
+      "step": 9113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.608556897227677e-05,
+      "loss": 1.0613,
+      "step": 9114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.608431988805808e-05,
+      "loss": 1.0554,
+      "step": 9115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6083070727352104e-05,
+      "loss": 1.0945,
+      "step": 9116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6081821490182246e-05,
+      "loss": 1.0707,
+      "step": 9117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.608057217657189e-05,
+      "loss": 1.046,
+      "step": 9118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6079322786544438e-05,
+      "loss": 1.1429,
+      "step": 9119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6078073320123286e-05,
+      "loss": 1.0471,
+      "step": 9120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6076823777331834e-05,
+      "loss": 1.0928,
+      "step": 9121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6075574158193477e-05,
+      "loss": 1.0401,
+      "step": 9122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607432446273162e-05,
+      "loss": 1.0904,
+      "step": 9123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.607307469096967e-05,
+      "loss": 1.0599,
+      "step": 9124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6071824842931027e-05,
+      "loss": 0.9586,
+      "step": 9125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6070574918639097e-05,
+      "loss": 1.0735,
+      "step": 9126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6069324918117288e-05,
+      "loss": 1.0922,
+      "step": 9127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606807484138901e-05,
+      "loss": 1.059,
+      "step": 9128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6066824688477672e-05,
+      "loss": 1.0379,
+      "step": 9129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6065574459406686e-05,
+      "loss": 1.0794,
+      "step": 9130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606432415419947e-05,
+      "loss": 1.0898,
+      "step": 9131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6063073772879422e-05,
+      "loss": 1.0402,
+      "step": 9132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606182331546998e-05,
+      "loss": 1.1098,
+      "step": 9133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.606057278199455e-05,
+      "loss": 1.0499,
+      "step": 9134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6059322172476546e-05,
+      "loss": 1.0548,
+      "step": 9135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.60580714869394e-05,
+      "loss": 1.0453,
+      "step": 9136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6056820725406524e-05,
+      "loss": 1.0603,
+      "step": 9137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6055569887901346e-05,
+      "loss": 1.1019,
+      "step": 9138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605431897444729e-05,
+      "loss": 1.1133,
+      "step": 9139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6053067985067785e-05,
+      "loss": 1.0333,
+      "step": 9140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.605181691978625e-05,
+      "loss": 1.0718,
+      "step": 9141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6050565778626115e-05,
+      "loss": 1.0141,
+      "step": 9142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6049314561610817e-05,
+      "loss": 1.0681,
+      "step": 9143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6048063268763786e-05,
+      "loss": 1.0092,
+      "step": 9144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6046811900108452e-05,
+      "loss": 1.0378,
+      "step": 9145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604556045566825e-05,
+      "loss": 1.0995,
+      "step": 9146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604430893546662e-05,
+      "loss": 1.0115,
+      "step": 9147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6043057339526992e-05,
+      "loss": 1.07,
+      "step": 9148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.604180566787281e-05,
+      "loss": 1.1281,
+      "step": 9149
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6040553920527514e-05,
+      "loss": 1.0269,
+      "step": 9150
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603930209751454e-05,
+      "loss": 1.1323,
+      "step": 9151
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6038050198857337e-05,
+      "loss": 1.0436,
+      "step": 9152
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603679822457935e-05,
+      "loss": 1.0579,
+      "step": 9153
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603554617470402e-05,
+      "loss": 1.032,
+      "step": 9154
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6034294049254797e-05,
+      "loss": 1.0843,
+      "step": 9155
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6033041848255127e-05,
+      "loss": 1.0233,
+      "step": 9156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6031789571728464e-05,
+      "loss": 0.979,
+      "step": 9157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.603053721969826e-05,
+      "loss": 1.0652,
+      "step": 9158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6029284792187966e-05,
+      "loss": 1.0296,
+      "step": 9159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6028032289221033e-05,
+      "loss": 1.1054,
+      "step": 9160
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6026779710820918e-05,
+      "loss": 1.0377,
+      "step": 9161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602552705701108e-05,
+      "loss": 1.0438,
+      "step": 9162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602427432781498e-05,
+      "loss": 1.074,
+      "step": 9163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6023021523256073e-05,
+      "loss": 1.0229,
+      "step": 9164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602176864335782e-05,
+      "loss": 1.0479,
+      "step": 9165
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.602051568814369e-05,
+      "loss": 1.0548,
+      "step": 9166
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6019262657637143e-05,
+      "loss": 1.053,
+      "step": 9167
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6018009551861646e-05,
+      "loss": 1.0763,
+      "step": 9168
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6016756370840662e-05,
+      "loss": 1.1221,
+      "step": 9169
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6015503114597663e-05,
+      "loss": 1.0766,
+      "step": 9170
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6014249783156116e-05,
+      "loss": 1.0424,
+      "step": 9171
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6012996376539498e-05,
+      "loss": 1.0729,
+      "step": 9172
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6011742894771273e-05,
+      "loss": 1.0943,
+      "step": 9173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6010489337874926e-05,
+      "loss": 1.1372,
+      "step": 9174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6009235705873922e-05,
+      "loss": 1.0081,
+      "step": 9175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6007981998791742e-05,
+      "loss": 1.0769,
+      "step": 9176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6006728216651865e-05,
+      "loss": 0.9861,
+      "step": 9177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6005474359477774e-05,
+      "loss": 0.9971,
+      "step": 9178
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.600422042729294e-05,
+      "loss": 1.0406,
+      "step": 9179
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6002966420120853e-05,
+      "loss": 1.0932,
+      "step": 9180
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6001712337984997e-05,
+      "loss": 1.0579,
+      "step": 9181
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.6000458180908855e-05,
+      "loss": 1.0801,
+      "step": 9182
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5999203948915916e-05,
+      "loss": 1.0635,
+      "step": 9183
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5997949642029665e-05,
+      "loss": 1.0459,
+      "step": 9184
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599669526027359e-05,
+      "loss": 1.1035,
+      "step": 9185
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599544080367119e-05,
+      "loss": 1.0606,
+      "step": 9186
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5994186272245952e-05,
+      "loss": 1.0679,
+      "step": 9187
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5992931666021365e-05,
+      "loss": 1.0638,
+      "step": 9188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599167698502093e-05,
+      "loss": 1.0053,
+      "step": 9189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.599042222926814e-05,
+      "loss": 1.0366,
+      "step": 9190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59891673987865e-05,
+      "loss": 1.0381,
+      "step": 9191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59879124935995e-05,
+      "loss": 1.0495,
+      "step": 9192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598665751373065e-05,
+      "loss": 1.0568,
+      "step": 9193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5985402459203442e-05,
+      "loss": 0.9928,
+      "step": 9194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598414733004139e-05,
+      "loss": 1.0748,
+      "step": 9195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.598289212626799e-05,
+      "loss": 1.0349,
+      "step": 9196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5981636847906755e-05,
+      "loss": 1.0192,
+      "step": 9197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5980381494981187e-05,
+      "loss": 1.0411,
+      "step": 9198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5979126067514796e-05,
+      "loss": 1.1186,
+      "step": 9199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5977870565531095e-05,
+      "loss": 1.0562,
+      "step": 9200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5976614989053593e-05,
+      "loss": 1.0526,
+      "step": 9201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597535933810581e-05,
+      "loss": 1.0569,
+      "step": 9202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5974103612711255e-05,
+      "loss": 1.1209,
+      "step": 9203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597284781289344e-05,
+      "loss": 1.0621,
+      "step": 9204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597159193867589e-05,
+      "loss": 1.0509,
+      "step": 9205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.597033599008212e-05,
+      "loss": 1.1029,
+      "step": 9206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5969079967135655e-05,
+      "loss": 1.0104,
+      "step": 9207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596782386986001e-05,
+      "loss": 1.0586,
+      "step": 9208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5966567698278712e-05,
+      "loss": 1.069,
+      "step": 9209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596531145241528e-05,
+      "loss": 1.0863,
+      "step": 9210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5964055132293247e-05,
+      "loss": 1.0236,
+      "step": 9211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5962798737936137e-05,
+      "loss": 1.0165,
+      "step": 9212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.596154226936748e-05,
+      "loss": 1.1486,
+      "step": 9213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5960285726610804e-05,
+      "loss": 1.0097,
+      "step": 9214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.595902910968964e-05,
+      "loss": 1.1218,
+      "step": 9215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.595777241862752e-05,
+      "loss": 1.1233,
+      "step": 9216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5956515653447985e-05,
+      "loss": 1.1032,
+      "step": 9217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5955258814174562e-05,
+      "loss": 1.0333,
+      "step": 9218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.595400190083079e-05,
+      "loss": 1.0571,
+      "step": 9219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.595274491344021e-05,
+      "loss": 1.0403,
+      "step": 9220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5951487852026365e-05,
+      "loss": 1.0511,
+      "step": 9221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5950230716612785e-05,
+      "loss": 1.0787,
+      "step": 9222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594897350722302e-05,
+      "loss": 1.0622,
+      "step": 9223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5947716223880617e-05,
+      "loss": 1.0363,
+      "step": 9224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594645886660911e-05,
+      "loss": 1.0927,
+      "step": 9225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.594520143543206e-05,
+      "loss": 1.0097,
+      "step": 9226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5943943930373003e-05,
+      "loss": 1.0199,
+      "step": 9227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5942686351455492e-05,
+      "loss": 1.0654,
+      "step": 9228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5941428698703078e-05,
+      "loss": 1.0317,
+      "step": 9229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5940170972139312e-05,
+      "loss": 1.0143,
+      "step": 9230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5938913171787755e-05,
+      "loss": 1.0219,
+      "step": 9231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5937655297671954e-05,
+      "loss": 0.9484,
+      "step": 9232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5936397349815467e-05,
+      "loss": 1.0514,
+      "step": 9233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5935139328241848e-05,
+      "loss": 0.9714,
+      "step": 9234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5933881232974665e-05,
+      "loss": 1.0079,
+      "step": 9235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5932623064037473e-05,
+      "loss": 1.028,
+      "step": 9236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5931364821453832e-05,
+      "loss": 1.1272,
+      "step": 9237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5930106505247308e-05,
+      "loss": 1.0046,
+      "step": 9238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5928848115441463e-05,
+      "loss": 1.0518,
+      "step": 9239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592758965205987e-05,
+      "loss": 1.0581,
+      "step": 9240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5926331115126087e-05,
+      "loss": 1.0315,
+      "step": 9241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592507250466369e-05,
+      "loss": 1.0595,
+      "step": 9242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5923813820696247e-05,
+      "loss": 1.0287,
+      "step": 9243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5922555063247326e-05,
+      "loss": 1.091,
+      "step": 9244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.59212962323405e-05,
+      "loss": 1.0259,
+      "step": 9245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.592003732799935e-05,
+      "loss": 1.0503,
+      "step": 9246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5918778350247443e-05,
+      "loss": 0.9922,
+      "step": 9247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591751929910836e-05,
+      "loss": 1.0349,
+      "step": 9248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591626017460568e-05,
+      "loss": 0.9667,
+      "step": 9249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591500097676298e-05,
+      "loss": 1.0683,
+      "step": 9250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591374170560385e-05,
+      "loss": 1.0717,
+      "step": 9251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.591248236115186e-05,
+      "loss": 1.0403,
+      "step": 9252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5911222943430604e-05,
+      "loss": 1.0893,
+      "step": 9253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590996345246366e-05,
+      "loss": 1.0339,
+      "step": 9254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5908703888274616e-05,
+      "loss": 1.0509,
+      "step": 9255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590744425088706e-05,
+      "loss": 1.0296,
+      "step": 9256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590618454032458e-05,
+      "loss": 1.0079,
+      "step": 9257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590492475661077e-05,
+      "loss": 1.0996,
+      "step": 9258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5903664899769225e-05,
+      "loss": 1.0684,
+      "step": 9259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5902404969823532e-05,
+      "loss": 1.0491,
+      "step": 9260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.590114496679729e-05,
+      "loss": 1.0184,
+      "step": 9261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5899884890714094e-05,
+      "loss": 1.0293,
+      "step": 9262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5898624741597538e-05,
+      "loss": 1.0219,
+      "step": 9263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589736451947122e-05,
+      "loss": 1.0165,
+      "step": 9264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5896104224358753e-05,
+      "loss": 1.024,
+      "step": 9265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5894843856283723e-05,
+      "loss": 1.1005,
+      "step": 9266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589358341526974e-05,
+      "loss": 1.1384,
+      "step": 9267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.589232290134041e-05,
+      "loss": 1.0652,
+      "step": 9268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5891062314519336e-05,
+      "loss": 1.0881,
+      "step": 9269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5889801654830124e-05,
+      "loss": 1.0612,
+      "step": 9270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5888540922296387e-05,
+      "loss": 1.0503,
+      "step": 9271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.588728011694173e-05,
+      "loss": 1.0304,
+      "step": 9272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5886019238789764e-05,
+      "loss": 1.0235,
+      "step": 9273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.588475828786411e-05,
+      "loss": 1.1025,
+      "step": 9274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.588349726418837e-05,
+      "loss": 1.0766,
+      "step": 9275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5882236167786168e-05,
+      "loss": 0.9517,
+      "step": 9276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5880974998681113e-05,
+      "loss": 1.0729,
+      "step": 9277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587971375689683e-05,
+      "loss": 1.0567,
+      "step": 9278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587845244245694e-05,
+      "loss": 1.116,
+      "step": 9279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5877191055385055e-05,
+      "loss": 1.0817,
+      "step": 9280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58759295957048e-05,
+      "loss": 1.0835,
+      "step": 9281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58746680634398e-05,
+      "loss": 1.0319,
+      "step": 9282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587340645861368e-05,
+      "loss": 1.0808,
+      "step": 9283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587214478125007e-05,
+      "loss": 1.0084,
+      "step": 9284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.587088303137259e-05,
+      "loss": 1.0624,
+      "step": 9285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586962120900487e-05,
+      "loss": 1.0399,
+      "step": 9286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5868359314170548e-05,
+      "loss": 1.0329,
+      "step": 9287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5867097346893247e-05,
+      "loss": 1.077,
+      "step": 9288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5865835307196602e-05,
+      "loss": 1.0657,
+      "step": 9289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.586457319510425e-05,
+      "loss": 1.024,
+      "step": 9290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5863311010639823e-05,
+      "loss": 1.051,
+      "step": 9291
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5862048753826963e-05,
+      "loss": 1.0775,
+      "step": 9292
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58607864246893e-05,
+      "loss": 1.0149,
+      "step": 9293
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5859524023250487e-05,
+      "loss": 1.1167,
+      "step": 9294
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585826154953415e-05,
+      "loss": 1.0889,
+      "step": 9295
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5856999003563937e-05,
+      "loss": 0.9698,
+      "step": 9296
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5855736385363497e-05,
+      "loss": 1.1067,
+      "step": 9297
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585447369495647e-05,
+      "loss": 1.0066,
+      "step": 9298
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585321093236651e-05,
+      "loss": 1.0327,
+      "step": 9299
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585194809761725e-05,
+      "loss": 1.0687,
+      "step": 9300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.585068519073235e-05,
+      "loss": 1.0904,
+      "step": 9301
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584942221173546e-05,
+      "loss": 1.0059,
+      "step": 9302
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5848159160650227e-05,
+      "loss": 1.1825,
+      "step": 9303
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584689603750031e-05,
+      "loss": 1.0653,
+      "step": 9304
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5845632842309363e-05,
+      "loss": 0.9723,
+      "step": 9305
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5844369575101035e-05,
+      "loss": 1.0582,
+      "step": 9306
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.584310623589899e-05,
+      "loss": 1.0407,
+      "step": 9307
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5841842824726886e-05,
+      "loss": 1.0611,
+      "step": 9308
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5840579341608382e-05,
+      "loss": 1.0144,
+      "step": 9309
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5839315786567138e-05,
+      "loss": 1.0009,
+      "step": 9310
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5838052159626817e-05,
+      "loss": 1.045,
+      "step": 9311
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5836788460811085e-05,
+      "loss": 1.0419,
+      "step": 9312
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5835524690143604e-05,
+      "loss": 1.0898,
+      "step": 9313
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5834260847648048e-05,
+      "loss": 0.9883,
+      "step": 9314
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5832996933348073e-05,
+      "loss": 1.0866,
+      "step": 9315
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5831732947267357e-05,
+      "loss": 1.0086,
+      "step": 9316
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.583046888942957e-05,
+      "loss": 1.0268,
+      "step": 9317
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5829204759858385e-05,
+      "loss": 1.0786,
+      "step": 9318
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5827940558577474e-05,
+      "loss": 1.0266,
+      "step": 9319
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5826676285610507e-05,
+      "loss": 1.0164,
+      "step": 9320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5825411940981166e-05,
+      "loss": 1.037,
+      "step": 9321
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5824147524713128e-05,
+      "loss": 1.0183,
+      "step": 9322
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.582288303683007e-05,
+      "loss": 0.999,
+      "step": 9323
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5821618477355674e-05,
+      "loss": 1.0872,
+      "step": 9324
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5820353846313616e-05,
+      "loss": 1.0141,
+      "step": 9325
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5819089143727585e-05,
+      "loss": 1.1215,
+      "step": 9326
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5817824369621263e-05,
+      "loss": 1.0593,
+      "step": 9327
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5816559524018337e-05,
+      "loss": 0.9496,
+      "step": 9328
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5815294606942493e-05,
+      "loss": 1.0485,
+      "step": 9329
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5814029618417417e-05,
+      "loss": 1.0645,
+      "step": 9330
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.58127645584668e-05,
+      "loss": 1.012,
+      "step": 9331
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5811499427114334e-05,
+      "loss": 1.0044,
+      "step": 9332
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5810234224383713e-05,
+      "loss": 1.1075,
+      "step": 9333
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5808968950298623e-05,
+      "loss": 1.0898,
+      "step": 9334
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5807703604882764e-05,
+      "loss": 1.0644,
+      "step": 9335
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5806438188159833e-05,
+      "loss": 0.9661,
+      "step": 9336
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5805172700153527e-05,
+      "loss": 1.0932,
+      "step": 9337
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5803907140887544e-05,
+      "loss": 1.0025,
+      "step": 9338
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5802641510385583e-05,
+      "loss": 1.0367,
+      "step": 9339
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5801375808671346e-05,
+      "loss": 1.0227,
+      "step": 9340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.580011003576854e-05,
+      "loss": 1.0276,
+      "step": 9341
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579884419170086e-05,
+      "loss": 1.1148,
+      "step": 9342
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5797578276492024e-05,
+      "loss": 1.0619,
+      "step": 9343
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.579631229016573e-05,
+      "loss": 1.0597,
+      "step": 9344
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5795046232745693e-05,
+      "loss": 1.0363,
+      "step": 9345
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5793780104255613e-05,
+      "loss": 1.0535,
+      "step": 9346
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5792513904719205e-05,
+      "loss": 1.0341,
+      "step": 9347
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.5791247634160185e-05,
+      "loss": 1.0065,
+      "step": 9348
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.578998129260226e-05,
+      "loss": 1.096,
+      "step": 9349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578871488006915e-05,
+      "loss": 1.0713,
+      "step": 9350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5787448396584572e-05,
+      "loss": 1.0638,
+      "step": 9351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578618184217224e-05,
+      "loss": 1.1078,
+      "step": 9352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5784915216855874e-05,
+      "loss": 1.046,
+      "step": 9353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5783648520659196e-05,
+      "loss": 1.047,
+      "step": 9354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5782381753605923e-05,
+      "loss": 1.1144,
+      "step": 9355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.578111491571978e-05,
+      "loss": 1.1055,
+      "step": 9356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5779848007024494e-05,
+      "loss": 1.0669,
+      "step": 9357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5778581027543787e-05,
+      "loss": 1.1068,
+      "step": 9358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5777313977301384e-05,
+      "loss": 1.0153,
+      "step": 9359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5776046856321018e-05,
+      "loss": 1.0464,
+      "step": 9360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5774779664626417e-05,
+      "loss": 1.0478,
+      "step": 9361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.577351240224131e-05,
+      "loss": 1.0753,
+      "step": 9362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5772245069189433e-05,
+      "loss": 1.0428,
+      "step": 9363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5770977665494517e-05,
+      "loss": 1.0514,
+      "step": 9364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5769710191180293e-05,
+      "loss": 1.0054,
+      "step": 9365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57684426462705e-05,
+      "loss": 1.043,
+      "step": 9366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576717503078888e-05,
+      "loss": 0.9996,
+      "step": 9367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5765907344759164e-05,
+      "loss": 0.9906,
+      "step": 9368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5764639588205094e-05,
+      "loss": 1.06,
+      "step": 9369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5763371761150416e-05,
+      "loss": 1.0031,
+      "step": 9370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.576210386361887e-05,
+      "loss": 1.0457,
+      "step": 9371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5760835895634203e-05,
+      "loss": 1.0836,
+      "step": 9372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5759567857220152e-05,
+      "loss": 1.0377,
+      "step": 9373
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575829974840047e-05,
+      "loss": 1.0407,
+      "step": 9374
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5757031569198903e-05,
+      "loss": 0.9801,
+      "step": 9375
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57557633196392e-05,
+      "loss": 1.0067,
+      "step": 9376
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5754494999745114e-05,
+      "loss": 1.0516,
+      "step": 9377
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5753226609540394e-05,
+      "loss": 1.0674,
+      "step": 9378
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5751958149048796e-05,
+      "loss": 1.0807,
+      "step": 9379
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.575068961829407e-05,
+      "loss": 1.0962,
+      "step": 9380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5749421017299977e-05,
+      "loss": 1.0212,
+      "step": 9381
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.574815234609027e-05,
+      "loss": 1.0053,
+      "step": 9382
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5746883604688713e-05,
+      "loss": 1.0351,
+      "step": 9383
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5745614793119062e-05,
+      "loss": 1.1197,
+      "step": 9384
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5744345911405074e-05,
+      "loss": 1.0172,
+      "step": 9385
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.574307695957052e-05,
+      "loss": 1.0097,
+      "step": 9386
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5741807937639155e-05,
+      "loss": 0.9996,
+      "step": 9387
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5740538845634753e-05,
+      "loss": 1.0389,
+      "step": 9388
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5739269683581073e-05,
+      "loss": 1.0609,
+      "step": 9389
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5738000451501884e-05,
+      "loss": 1.0226,
+      "step": 9390
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5736731149420965e-05,
+      "loss": 1.0797,
+      "step": 9391
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.573546177736207e-05,
+      "loss": 1.07,
+      "step": 9392
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.573419233534898e-05,
+      "loss": 1.0689,
+      "step": 9393
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5732922823405468e-05,
+      "loss": 1.0989,
+      "step": 9394
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5731653241555303e-05,
+      "loss": 1.0412,
+      "step": 9395
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.573038358982227e-05,
+      "loss": 1.0365,
+      "step": 9396
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5729113868230137e-05,
+      "loss": 1.0059,
+      "step": 9397
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5727844076802682e-05,
+      "loss": 1.0788,
+      "step": 9398
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572657421556369e-05,
+      "loss": 1.0875,
+      "step": 9399
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572530428453694e-05,
+      "loss": 1.0874,
+      "step": 9400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5724034283746212e-05,
+      "loss": 1.1155,
+      "step": 9401
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5722764213215292e-05,
+      "loss": 1.0276,
+      "step": 9402
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.572149407296796e-05,
+      "loss": 1.0876,
+      "step": 9403
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5720223863028008e-05,
+      "loss": 1.1424,
+      "step": 9404
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571895358341922e-05,
+      "loss": 1.0717,
+      "step": 9405
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5717683234165386e-05,
+      "loss": 1.0512,
+      "step": 9406
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5716412815290293e-05,
+      "loss": 1.0347,
+      "step": 9407
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571514232681774e-05,
+      "loss": 1.0067,
+      "step": 9408
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5713871768771507e-05,
+      "loss": 0.9842,
+      "step": 9409
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57126011411754e-05,
+      "loss": 1.0652,
+      "step": 9410
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.571133044405321e-05,
+      "loss": 1.029,
+      "step": 9411
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5710059677428726e-05,
+      "loss": 1.0781,
+      "step": 9412
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5708788841325756e-05,
+      "loss": 1.068,
+      "step": 9413
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570751793576809e-05,
+      "loss": 1.0958,
+      "step": 9414
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5706246960779536e-05,
+      "loss": 1.0633,
+      "step": 9415
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.57049759163839e-05,
+      "loss": 1.0844,
+      "step": 9416
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5703704802604966e-05,
+      "loss": 1.0623,
+      "step": 9417
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570243361946656e-05,
+      "loss": 1.0465,
+      "step": 9418
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.570116236699247e-05,
+      "loss": 1.0661,
+      "step": 9419
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5699891045206515e-05,
+      "loss": 1.0198,
+      "step": 9420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5698619654132497e-05,
+      "loss": 1.0231,
+      "step": 9421
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5697348193794227e-05,
+      "loss": 1.1221,
+      "step": 9422
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5696076664215514e-05,
+      "loss": 1.0306,
+      "step": 9423
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5694805065420174e-05,
+      "loss": 1.0456,
+      "step": 9424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5693533397432018e-05,
+      "loss": 1.0907,
+      "step": 9425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569226166027486e-05,
+      "loss": 1.0148,
+      "step": 9426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.569098985397251e-05,
+      "loss": 1.0846,
+      "step": 9427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56897179785488e-05,
+      "loss": 1.0334,
+      "step": 9428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568844603402754e-05,
+      "loss": 1.0136,
+      "step": 9429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568717402043255e-05,
+      "loss": 1.0665,
+      "step": 9430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5685901937787646e-05,
+      "loss": 0.96,
+      "step": 9431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5684629786116653e-05,
+      "loss": 1.0916,
+      "step": 9432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5683357565443402e-05,
+      "loss": 1.0642,
+      "step": 9433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.568208527579171e-05,
+      "loss": 0.9905,
+      "step": 9434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5680812917185412e-05,
+      "loss": 1.1166,
+      "step": 9435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5679540489648324e-05,
+      "loss": 1.12,
+      "step": 9436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5678267993204283e-05,
+      "loss": 1.0917,
+      "step": 9437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5676995427877117e-05,
+      "loss": 1.0953,
+      "step": 9438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5675722793690654e-05,
+      "loss": 1.0772,
+      "step": 9439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567445009066873e-05,
+      "loss": 1.0092,
+      "step": 9440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567317731883518e-05,
+      "loss": 1.1119,
+      "step": 9441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567190447821384e-05,
+      "loss": 1.0361,
+      "step": 9442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.567063156882854e-05,
+      "loss": 1.05,
+      "step": 9443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5669358590703123e-05,
+      "loss": 1.0744,
+      "step": 9444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566808554386143e-05,
+      "loss": 1.0785,
+      "step": 9445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5666812428327295e-05,
+      "loss": 0.9988,
+      "step": 9446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566553924412457e-05,
+      "loss": 1.1096,
+      "step": 9447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5664265991277087e-05,
+      "loss": 1.008,
+      "step": 9448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.566299266980869e-05,
+      "loss": 1.0025,
+      "step": 9449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5661719279743237e-05,
+      "loss": 1.065,
+      "step": 9450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5660445821104564e-05,
+      "loss": 0.9687,
+      "step": 9451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5659172293916524e-05,
+      "loss": 1.0715,
+      "step": 9452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5657898698202962e-05,
+      "loss": 1.0789,
+      "step": 9453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5656625033987733e-05,
+      "loss": 1.0046,
+      "step": 9454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.565535130129469e-05,
+      "loss": 1.0585,
+      "step": 9455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.565407750014768e-05,
+      "loss": 1.1255,
+      "step": 9456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5652803630570562e-05,
+      "loss": 1.0416,
+      "step": 9457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5651529692587193e-05,
+      "loss": 1.0804,
+      "step": 9458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5650255686221427e-05,
+      "loss": 1.1113,
+      "step": 9459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5648981611497124e-05,
+      "loss": 1.0123,
+      "step": 9460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564770746843814e-05,
+      "loss": 1.0352,
+      "step": 9461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564643325706834e-05,
+      "loss": 1.0797,
+      "step": 9462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5645158977411588e-05,
+      "loss": 1.0581,
+      "step": 9463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5643884629491746e-05,
+      "loss": 1.0975,
+      "step": 9464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5642610213332674e-05,
+      "loss": 1.0199,
+      "step": 9465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5641335728958246e-05,
+      "loss": 1.0383,
+      "step": 9466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.564006117639232e-05,
+      "loss": 1.0675,
+      "step": 9467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5638786555658772e-05,
+      "loss": 1.027,
+      "step": 9468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563751186678147e-05,
+      "loss": 1.0307,
+      "step": 9469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5636237109784285e-05,
+      "loss": 1.0561,
+      "step": 9470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563496228469109e-05,
+      "loss": 1.0283,
+      "step": 9471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.563368739152575e-05,
+      "loss": 1.0454,
+      "step": 9472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5632412430312157e-05,
+      "loss": 1.0697,
+      "step": 9473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5631137401074174e-05,
+      "loss": 1.0671,
+      "step": 9474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5629862303835685e-05,
+      "loss": 1.074,
+      "step": 9475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562858713862057e-05,
+      "loss": 1.0075,
+      "step": 9476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56273119054527e-05,
+      "loss": 1.1192,
+      "step": 9477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5626036604355962e-05,
+      "loss": 1.0924,
+      "step": 9478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.562476123535424e-05,
+      "loss": 1.0398,
+      "step": 9479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5623485798471416e-05,
+      "loss": 1.0728,
+      "step": 9480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5622210293731377e-05,
+      "loss": 1.1444,
+      "step": 9481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5620934721158005e-05,
+      "loss": 1.0617,
+      "step": 9482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5619659080775193e-05,
+      "loss": 1.0602,
+      "step": 9483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561838337260683e-05,
+      "loss": 1.1078,
+      "step": 9484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.56171075966768e-05,
+      "loss": 1.0582,
+      "step": 9485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5615831753009e-05,
+      "loss": 1.0273,
+      "step": 9486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5614555841627323e-05,
+      "loss": 1.089,
+      "step": 9487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5613279862555657e-05,
+      "loss": 1.1077,
+      "step": 9488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5612003815817907e-05,
+      "loss": 1.0806,
+      "step": 9489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.561072770143796e-05,
+      "loss": 0.9896,
+      "step": 9490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5609451519439722e-05,
+      "loss": 1.0973,
+      "step": 9491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5608175269847088e-05,
+      "loss": 1.0337,
+      "step": 9492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560689895268396e-05,
+      "loss": 1.045,
+      "step": 9493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560562256797423e-05,
+      "loss": 1.0359,
+      "step": 9494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560434611574182e-05,
+      "loss": 1.1073,
+      "step": 9495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.560306959601062e-05,
+      "loss": 1.0133,
+      "step": 9496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5601793008804536e-05,
+      "loss": 1.0709,
+      "step": 9497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5600516354147482e-05,
+      "loss": 0.9759,
+      "step": 9498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559923963206336e-05,
+      "loss": 1.0113,
+      "step": 9499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559796284257608e-05,
+      "loss": 0.9983,
+      "step": 9500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5596685985709552e-05,
+      "loss": 1.0793,
+      "step": 9501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559540906148769e-05,
+      "loss": 1.0856,
+      "step": 9502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5594132069934413e-05,
+      "loss": 1.0543,
+      "step": 9503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5592855011073622e-05,
+      "loss": 0.9791,
+      "step": 9504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559157788492924e-05,
+      "loss": 1.08,
+      "step": 9505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.559030069152518e-05,
+      "loss": 1.0674,
+      "step": 9506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5589023430885368e-05,
+      "loss": 1.0786,
+      "step": 9507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558774610303372e-05,
+      "loss": 1.0653,
+      "step": 9508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558646870799415e-05,
+      "loss": 1.136,
+      "step": 9509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5585191245790584e-05,
+      "loss": 1.0246,
+      "step": 9510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558391371644695e-05,
+      "loss": 1.0923,
+      "step": 9511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5582636119987164e-05,
+      "loss": 1.0556,
+      "step": 9512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558135845643516e-05,
+      "loss": 1.0215,
+      "step": 9513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.558008072581486e-05,
+      "loss": 1.0099,
+      "step": 9514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557880292815019e-05,
+      "loss": 1.1174,
+      "step": 9515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557752506346508e-05,
+      "loss": 1.0318,
+      "step": 9516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557624713178347e-05,
+      "loss": 1.0899,
+      "step": 9517
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557496913312928e-05,
+      "loss": 1.1144,
+      "step": 9518
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557369106752645e-05,
+      "loss": 1.0079,
+      "step": 9519
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.557241293499891e-05,
+      "loss": 1.074,
+      "step": 9520
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55711347355706e-05,
+      "loss": 1.0635,
+      "step": 9521
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5569856469265456e-05,
+      "loss": 1.0909,
+      "step": 9522
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5568578136107415e-05,
+      "loss": 1.0099,
+      "step": 9523
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5567299736120417e-05,
+      "loss": 1.0849,
+      "step": 9524
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5566021269328397e-05,
+      "loss": 1.1016,
+      "step": 9525
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5564742735755307e-05,
+      "loss": 1.0433,
+      "step": 9526
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5563464135425084e-05,
+      "loss": 1.0449,
+      "step": 9527
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5562185468361675e-05,
+      "loss": 1.0301,
+      "step": 9528
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.556090673458902e-05,
+      "loss": 1.0634,
+      "step": 9529
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555962793413107e-05,
+      "loss": 1.0742,
+      "step": 9530
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555834906701178e-05,
+      "loss": 1.0968,
+      "step": 9531
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555707013325509e-05,
+      "loss": 0.9992,
+      "step": 9532
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.555579113288495e-05,
+      "loss": 0.9543,
+      "step": 9533
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5554512065925317e-05,
+      "loss": 1.1042,
+      "step": 9534
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5553232932400146e-05,
+      "loss": 1.1168,
+      "step": 9535
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5551953732333384e-05,
+      "loss": 1.0096,
+      "step": 9536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5550674465748993e-05,
+      "loss": 0.9985,
+      "step": 9537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5549395132670923e-05,
+      "loss": 1.0744,
+      "step": 9538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.554811573312314e-05,
+      "loss": 1.011,
+      "step": 9539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55468362671296e-05,
+      "loss": 1.1253,
+      "step": 9540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5545556734714266e-05,
+      "loss": 1.1224,
+      "step": 9541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5544277135901097e-05,
+      "loss": 0.9769,
+      "step": 9542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5542997470714053e-05,
+      "loss": 0.9613,
+      "step": 9543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5541717739177102e-05,
+      "loss": 1.0611,
+      "step": 9544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5540437941314212e-05,
+      "loss": 1.0312,
+      "step": 9545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5539158077149343e-05,
+      "loss": 1.0603,
+      "step": 9546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5537878146706475e-05,
+      "loss": 1.1285,
+      "step": 9547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5536598150009566e-05,
+      "loss": 0.9917,
+      "step": 9548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553531808708259e-05,
+      "loss": 1.0056,
+      "step": 9549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553403795794952e-05,
+      "loss": 1.0641,
+      "step": 9550
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5532757762634327e-05,
+      "loss": 1.0844,
+      "step": 9551
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.553147750116099e-05,
+      "loss": 1.0535,
+      "step": 9552
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5530197173553477e-05,
+      "loss": 0.974,
+      "step": 9553
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552891677983577e-05,
+      "loss": 1.0341,
+      "step": 9554
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5527636320031845e-05,
+      "loss": 1.0905,
+      "step": 9555
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5526355794165685e-05,
+      "loss": 1.0476,
+      "step": 9556
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5525075202261266e-05,
+      "loss": 1.0595,
+      "step": 9557
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552379454434257e-05,
+      "loss": 1.0421,
+      "step": 9558
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5522513820433587e-05,
+      "loss": 1.0315,
+      "step": 9559
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.552123303055829e-05,
+      "loss": 1.0277,
+      "step": 9560
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551995217474067e-05,
+      "loss": 0.9842,
+      "step": 9561
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5518671253004715e-05,
+      "loss": 0.9823,
+      "step": 9562
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551739026537441e-05,
+      "loss": 1.007,
+      "step": 9563
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5516109211873745e-05,
+      "loss": 1.0565,
+      "step": 9564
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551482809252671e-05,
+      "loss": 1.016,
+      "step": 9565
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.55135469073573e-05,
+      "loss": 1.0876,
+      "step": 9566
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5512265656389505e-05,
+      "loss": 1.0043,
+      "step": 9567
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.551098433964732e-05,
+      "loss": 1.1276,
+      "step": 9568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550970295715474e-05,
+      "loss": 1.0671,
+      "step": 9569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5508421508935758e-05,
+      "loss": 1.0134,
+      "step": 9570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550713999501438e-05,
+      "loss": 1.0129,
+      "step": 9571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5505858415414595e-05,
+      "loss": 1.04,
+      "step": 9572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550457677016041e-05,
+      "loss": 1.0605,
+      "step": 9573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5503295059275824e-05,
+      "loss": 1.0416,
+      "step": 9574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.550201328278484e-05,
+      "loss": 1.0617,
+      "step": 9575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5500731440711463e-05,
+      "loss": 1.0672,
+      "step": 9576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5499449533079698e-05,
+      "loss": 1.0177,
+      "step": 9577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5498167559913552e-05,
+      "loss": 0.9755,
+      "step": 9578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5496885521237028e-05,
+      "loss": 1.0487,
+      "step": 9579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5495603417074134e-05,
+      "loss": 1.0602,
+      "step": 9580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549432124744889e-05,
+      "loss": 1.0457,
+      "step": 9581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.54930390123853e-05,
+      "loss": 1.0316,
+      "step": 9582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.549175671190738e-05,
+      "loss": 1.0506,
+      "step": 9583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.5490474346039138e-05,
+      "loss": 0.9636,
+      "step": 9584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.548919191480459e-05,
+      "loss": 1.042,
+      "step": 9585
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5487909418227762e-05,
+      "loss": 1.0476,
+      "step": 9586
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.548662685633266e-05,
+      "loss": 1.0794,
+      "step": 9587
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5485344229143306e-05,
+      "loss": 1.0474,
+      "step": 9588
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.548406153668372e-05,
+      "loss": 1.0445,
+      "step": 9589
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5482778778977926e-05,
+      "loss": 1.1041,
+      "step": 9590
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5481495956049944e-05,
+      "loss": 1.0565,
+      "step": 9591
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5480213067923792e-05,
+      "loss": 0.9898,
+      "step": 9592
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5478930114623504e-05,
+      "loss": 1.1014,
+      "step": 9593
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54776470961731e-05,
+      "loss": 1.005,
+      "step": 9594
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547636401259661e-05,
+      "loss": 1.0306,
+      "step": 9595
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5475080863918063e-05,
+      "loss": 1.1293,
+      "step": 9596
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547379765016149e-05,
+      "loss": 1.0925,
+      "step": 9597
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5472514371350912e-05,
+      "loss": 1.0559,
+      "step": 9598
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.547123102751037e-05,
+      "loss": 1.0274,
+      "step": 9599
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5469947618663897e-05,
+      "loss": 1.0439,
+      "step": 9600
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5468664144835524e-05,
+      "loss": 0.9674,
+      "step": 9601
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5467380606049284e-05,
+      "loss": 0.9375,
+      "step": 9602
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5466097002329225e-05,
+      "loss": 1.0462,
+      "step": 9603
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5464813333699374e-05,
+      "loss": 1.0329,
+      "step": 9604
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5463529600183777e-05,
+      "loss": 1.054,
+      "step": 9605
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5462245801806472e-05,
+      "loss": 1.0365,
+      "step": 9606
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5460961938591503e-05,
+      "loss": 1.0337,
+      "step": 9607
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5459678010562904e-05,
+      "loss": 1.035,
+      "step": 9608
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545839401774473e-05,
+      "loss": 1.0892,
+      "step": 9609
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5457109960161027e-05,
+      "loss": 1.0373,
+      "step": 9610
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545582583783583e-05,
+      "loss": 1.0221,
+      "step": 9611
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5454541650793196e-05,
+      "loss": 1.0326,
+      "step": 9612
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5453257399057173e-05,
+      "loss": 1.0991,
+      "step": 9613
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5451973082651808e-05,
+      "loss": 1.0556,
+      "step": 9614
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.545068870160116e-05,
+      "loss": 1.1009,
+      "step": 9615
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544940425592927e-05,
+      "loss": 1.0714,
+      "step": 9616
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54481197456602e-05,
+      "loss": 1.0037,
+      "step": 9617
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5446835170818006e-05,
+      "loss": 1.0207,
+      "step": 9618
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544555053142674e-05,
+      "loss": 1.0826,
+      "step": 9619
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5444265827510462e-05,
+      "loss": 1.0264,
+      "step": 9620
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.544298105909323e-05,
+      "loss": 1.0414,
+      "step": 9621
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54416962261991e-05,
+      "loss": 1.0103,
+      "step": 9622
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5440411328852138e-05,
+      "loss": 1.0303,
+      "step": 9623
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543912636707641e-05,
+      "loss": 1.0254,
+      "step": 9624
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543784134089597e-05,
+      "loss": 0.9732,
+      "step": 9625
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5436556250334892e-05,
+      "loss": 0.9919,
+      "step": 9626
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5435271095417233e-05,
+      "loss": 1.0321,
+      "step": 9627
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5433985876167066e-05,
+      "loss": 1.0215,
+      "step": 9628
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5432700592608457e-05,
+      "loss": 1.0357,
+      "step": 9629
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.543141524476548e-05,
+      "loss": 0.9822,
+      "step": 9630
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.54301298326622e-05,
+      "loss": 1.0516,
+      "step": 9631
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5428844356322692e-05,
+      "loss": 1.0383,
+      "step": 9632
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.542755881577103e-05,
+      "loss": 1.0654,
+      "step": 9633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5426273211031284e-05,
+      "loss": 1.067,
+      "step": 9634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5424987542127537e-05,
+      "loss": 1.0921,
+      "step": 9635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5423701809083857e-05,
+      "loss": 1.0321,
+      "step": 9636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5422416011924327e-05,
+      "loss": 1.0515,
+      "step": 9637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5421130150673025e-05,
+      "loss": 1.0195,
+      "step": 9638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5419844225354035e-05,
+      "loss": 0.9879,
+      "step": 9639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5418558235991436e-05,
+      "loss": 1.0484,
+      "step": 9640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5417272182609305e-05,
+      "loss": 1.0055,
+      "step": 9641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5415986065231733e-05,
+      "loss": 1.0915,
+      "step": 9642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5414699883882808e-05,
+      "loss": 1.0675,
+      "step": 9643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541341363858661e-05,
+      "loss": 1.0995,
+      "step": 9644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5412127329367225e-05,
+      "loss": 1.0336,
+      "step": 9645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.541084095624875e-05,
+      "loss": 1.0238,
+      "step": 9646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5409554519255266e-05,
+      "loss": 0.9557,
+      "step": 9647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540826801841087e-05,
+      "loss": 1.0065,
+      "step": 9648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5406981453739654e-05,
+      "loss": 1.0366,
+      "step": 9649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540569482526571e-05,
+      "loss": 1.0297,
+      "step": 9650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540440813301313e-05,
+      "loss": 1.0547,
+      "step": 9651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540312137700602e-05,
+      "loss": 1.1071,
+      "step": 9652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5401834557268463e-05,
+      "loss": 1.0875,
+      "step": 9653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.540054767382457e-05,
+      "loss": 0.9859,
+      "step": 9654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539926072669843e-05,
+      "loss": 1.0141,
+      "step": 9655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539797371591415e-05,
+      "loss": 1.0317,
+      "step": 9656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5396686641495835e-05,
+      "loss": 1.0133,
+      "step": 9657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539539950346758e-05,
+      "loss": 1.0263,
+      "step": 9658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5394112301853494e-05,
+      "loss": 1.0786,
+      "step": 9659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5392825036677683e-05,
+      "loss": 1.047,
+      "step": 9660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539153770796425e-05,
+      "loss": 1.0308,
+      "step": 9661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.539025031573731e-05,
+      "loss": 1.0043,
+      "step": 9662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.538896286002096e-05,
+      "loss": 1.0309,
+      "step": 9663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5387675340839322e-05,
+      "loss": 1.1467,
+      "step": 9664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5386387758216505e-05,
+      "loss": 1.0087,
+      "step": 9665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5385100112176615e-05,
+      "loss": 1.0435,
+      "step": 9666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5383812402743773e-05,
+      "loss": 1.053,
+      "step": 9667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5382524629942092e-05,
+      "loss": 1.0315,
+      "step": 9668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5381236793795686e-05,
+      "loss": 1.0283,
+      "step": 9669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537994889432867e-05,
+      "loss": 1.1074,
+      "step": 9670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5378660931565175e-05,
+      "loss": 1.0876,
+      "step": 9671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537737290552931e-05,
+      "loss": 1.0299,
+      "step": 9672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5376084816245197e-05,
+      "loss": 1.0396,
+      "step": 9673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.537479666373696e-05,
+      "loss": 1.0763,
+      "step": 9674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5373508448028716e-05,
+      "loss": 1.0369,
+      "step": 9675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5372220169144602e-05,
+      "loss": 1.0045,
+      "step": 9676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5370931827108734e-05,
+      "loss": 1.0408,
+      "step": 9677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536964342194524e-05,
+      "loss": 1.048,
+      "step": 9678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5368354953678252e-05,
+      "loss": 1.0243,
+      "step": 9679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.53670664223319e-05,
+      "loss": 1.0784,
+      "step": 9680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536577782793031e-05,
+      "loss": 1.0002,
+      "step": 9681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536448917049761e-05,
+      "loss": 1.0715,
+      "step": 9682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.536320045005794e-05,
+      "loss": 1.0666,
+      "step": 9683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5361911666635434e-05,
+      "loss": 0.9875,
+      "step": 9684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5360622820254224e-05,
+      "loss": 1.027,
+      "step": 9685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5359333910938446e-05,
+      "loss": 1.0764,
+      "step": 9686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5358044938712237e-05,
+      "loss": 1.0357,
+      "step": 9687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535675590359974e-05,
+      "loss": 1.0207,
+      "step": 9688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535546680562509e-05,
+      "loss": 1.1162,
+      "step": 9689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5354177644812433e-05,
+      "loss": 1.0532,
+      "step": 9690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5352888421185907e-05,
+      "loss": 1.0367,
+      "step": 9691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.535159913476965e-05,
+      "loss": 1.0567,
+      "step": 9692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5350309785587824e-05,
+      "loss": 1.0779,
+      "step": 9693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534902037366456e-05,
+      "loss": 1.0136,
+      "step": 9694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5347730899024003e-05,
+      "loss": 1.0615,
+      "step": 9695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5346441361690314e-05,
+      "loss": 1.0066,
+      "step": 9696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5345151761687628e-05,
+      "loss": 1.1084,
+      "step": 9697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.534386209904011e-05,
+      "loss": 1.0458,
+      "step": 9698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5342572373771897e-05,
+      "loss": 1.0465,
+      "step": 9699
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5341282585907153e-05,
+      "loss": 1.07,
+      "step": 9700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5339992735470028e-05,
+      "loss": 1.0124,
+      "step": 9701
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5338702822484674e-05,
+      "loss": 1.0635,
+      "step": 9702
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533741284697525e-05,
+      "loss": 1.0317,
+      "step": 9703
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5336122808965915e-05,
+      "loss": 1.0599,
+      "step": 9704
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5334832708480826e-05,
+      "loss": 1.0551,
+      "step": 9705
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5333542545544143e-05,
+      "loss": 1.0563,
+      "step": 9706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5332252320180027e-05,
+      "loss": 0.9976,
+      "step": 9707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.533096203241264e-05,
+      "loss": 1.0895,
+      "step": 9708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5329671682266144e-05,
+      "loss": 1.0754,
+      "step": 9709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5328381269764705e-05,
+      "loss": 1.0093,
+      "step": 9710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.532709079493249e-05,
+      "loss": 1.0256,
+      "step": 9711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5325800257793663e-05,
+      "loss": 1.012,
+      "step": 9712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.532450965837239e-05,
+      "loss": 1.089,
+      "step": 9713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5323218996692846e-05,
+      "loss": 1.0713,
+      "step": 9714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5321928272779198e-05,
+      "loss": 0.9946,
+      "step": 9715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5320637486655616e-05,
+      "loss": 1.0495,
+      "step": 9716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.531934663834628e-05,
+      "loss": 1.0666,
+      "step": 9717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5318055727875356e-05,
+      "loss": 1.0494,
+      "step": 9718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.531676475526702e-05,
+      "loss": 1.0719,
+      "step": 9719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5315473720545448e-05,
+      "loss": 1.0036,
+      "step": 9720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5314182623734822e-05,
+      "loss": 1.0852,
+      "step": 9721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5312891464859314e-05,
+      "loss": 1.0621,
+      "step": 9722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5311600243943108e-05,
+      "loss": 1.0373,
+      "step": 9723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5310308961010385e-05,
+      "loss": 1.1515,
+      "step": 9724
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5309017616085323e-05,
+      "loss": 1.1034,
+      "step": 9725
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530772620919211e-05,
+      "loss": 1.024,
+      "step": 9726
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5306434740354927e-05,
+      "loss": 0.9884,
+      "step": 9727
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530514320959796e-05,
+      "loss": 0.9994,
+      "step": 9728
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5303851616945395e-05,
+      "loss": 1.0537,
+      "step": 9729
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.530255996242142e-05,
+      "loss": 1.0522,
+      "step": 9730
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5301268246050225e-05,
+      "loss": 0.9772,
+      "step": 9731
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5299976467856e-05,
+      "loss": 1.0098,
+      "step": 9732
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5298684627862934e-05,
+      "loss": 1.0899,
+      "step": 9733
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529739272609522e-05,
+      "loss": 1.068,
+      "step": 9734
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5296100762577055e-05,
+      "loss": 1.0163,
+      "step": 9735
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5294808737332633e-05,
+      "loss": 1.0729,
+      "step": 9736
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5293516650386146e-05,
+      "loss": 1.0719,
+      "step": 9737
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5292224501761794e-05,
+      "loss": 1.0115,
+      "step": 9738
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529093229148377e-05,
+      "loss": 1.053,
+      "step": 9739
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528964001957628e-05,
+      "loss": 1.1499,
+      "step": 9740
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528834768606352e-05,
+      "loss": 1.0275,
+      "step": 9741
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5287055290969697e-05,
+      "loss": 1.112,
+      "step": 9742
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5285762834319008e-05,
+      "loss": 1.1123,
+      "step": 9743
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5284470316135658e-05,
+      "loss": 1.0362,
+      "step": 9744
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5283177736443855e-05,
+      "loss": 1.0206,
+      "step": 9745
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.52818850952678e-05,
+      "loss": 1.1102,
+      "step": 9746
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.528059239263171e-05,
+      "loss": 0.998,
+      "step": 9747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5279299628559784e-05,
+      "loss": 1.056,
+      "step": 9748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527800680307623e-05,
+      "loss": 0.9982,
+      "step": 9749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5276713916205267e-05,
+      "loss": 1.0901,
+      "step": 9750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.527542096797111e-05,
+      "loss": 1.0853,
+      "step": 9751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5274127958397956e-05,
+      "loss": 1.055,
+      "step": 9752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5272834887510036e-05,
+      "loss": 1.0106,
+      "step": 9753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5271541755331554e-05,
+      "loss": 1.0332,
+      "step": 9754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5270248561886735e-05,
+      "loss": 1.1106,
+      "step": 9755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526895530719979e-05,
+      "loss": 1.0472,
+      "step": 9756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5267661991294943e-05,
+      "loss": 0.9912,
+      "step": 9757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5266368614196407e-05,
+      "loss": 1.0977,
+      "step": 9758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526507517592841e-05,
+      "loss": 1.0671,
+      "step": 9759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526378167651517e-05,
+      "loss": 0.9452,
+      "step": 9760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526248811598092e-05,
+      "loss": 1.0518,
+      "step": 9761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.526119449434987e-05,
+      "loss": 1.072,
+      "step": 9762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5259900811646255e-05,
+      "loss": 0.9846,
+      "step": 9763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5258607067894298e-05,
+      "loss": 1.0615,
+      "step": 9764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525731326311823e-05,
+      "loss": 1.056,
+      "step": 9765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525601939734228e-05,
+      "loss": 1.0734,
+      "step": 9766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5254725470590673e-05,
+      "loss": 1.1079,
+      "step": 9767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5253431482887646e-05,
+      "loss": 1.0578,
+      "step": 9768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.525213743425743e-05,
+      "loss": 1.0719,
+      "step": 9769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5250843324724262e-05,
+      "loss": 1.0736,
+      "step": 9770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.524954915431237e-05,
+      "loss": 1.1057,
+      "step": 9771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5248254923045996e-05,
+      "loss": 1.0329,
+      "step": 9772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5246960630949373e-05,
+      "loss": 1.0593,
+      "step": 9773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5245666278046743e-05,
+      "loss": 1.026,
+      "step": 9774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5244371864362342e-05,
+      "loss": 1.0629,
+      "step": 9775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5243077389920411e-05,
+      "loss": 1.0536,
+      "step": 9776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5241782854745195e-05,
+      "loss": 1.084,
+      "step": 9777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5240488258860932e-05,
+      "loss": 1.0485,
+      "step": 9778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.523919360229187e-05,
+      "loss": 1.0017,
+      "step": 9779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5237898885062251e-05,
+      "loss": 0.9864,
+      "step": 9780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5236604107196326e-05,
+      "loss": 1.0691,
+      "step": 9781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5235309268718337e-05,
+      "loss": 1.1023,
+      "step": 9782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5234014369652536e-05,
+      "loss": 1.0476,
+      "step": 9783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.523271941002317e-05,
+      "loss": 1.0799,
+      "step": 9784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5231424389854493e-05,
+      "loss": 0.9851,
+      "step": 9785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5230129309170754e-05,
+      "loss": 1.1311,
+      "step": 9786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5228834167996207e-05,
+      "loss": 1.0254,
+      "step": 9787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5227538966355104e-05,
+      "loss": 1.0899,
+      "step": 9788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5226243704271706e-05,
+      "loss": 1.0413,
+      "step": 9789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5224948381770265e-05,
+      "loss": 1.0271,
+      "step": 9790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.522365299887504e-05,
+      "loss": 1.0828,
+      "step": 9791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.522235755561029e-05,
+      "loss": 1.0064,
+      "step": 9792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5221062052000273e-05,
+      "loss": 1.0614,
+      "step": 9793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5219766488069249e-05,
+      "loss": 1.0783,
+      "step": 9794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5218470863841487e-05,
+      "loss": 1.0207,
+      "step": 9795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5217175179341243e-05,
+      "loss": 1.0535,
+      "step": 9796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5215879434592783e-05,
+      "loss": 1.04,
+      "step": 9797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5214583629620374e-05,
+      "loss": 1.0947,
+      "step": 9798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5213287764448282e-05,
+      "loss": 1.0432,
+      "step": 9799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5211991839100775e-05,
+      "loss": 1.0499,
+      "step": 9800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5210695853602123e-05,
+      "loss": 0.9975,
+      "step": 9801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5209399807976593e-05,
+      "loss": 1.071,
+      "step": 9802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5208103702248456e-05,
+      "loss": 0.9515,
+      "step": 9803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.520680753644199e-05,
+      "loss": 0.9752,
+      "step": 9804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.520551131058146e-05,
+      "loss": 1.0492,
+      "step": 9805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5204215024691144e-05,
+      "loss": 1.0848,
+      "step": 9806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5202918678795322e-05,
+      "loss": 1.0209,
+      "step": 9807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5201622272918263e-05,
+      "loss": 1.0694,
+      "step": 9808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5200325807084252e-05,
+      "loss": 1.0621,
+      "step": 9809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5199029281317566e-05,
+      "loss": 1.125,
+      "step": 9810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5197732695642484e-05,
+      "loss": 1.0236,
+      "step": 9811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5196436050083285e-05,
+      "loss": 1.0817,
+      "step": 9812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5195139344664253e-05,
+      "loss": 1.0527,
+      "step": 9813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5193842579409676e-05,
+      "loss": 1.0576,
+      "step": 9814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5192545754343833e-05,
+      "loss": 1.0513,
+      "step": 9815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5191248869491009e-05,
+      "loss": 1.0281,
+      "step": 9816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5189951924875495e-05,
+      "loss": 0.9864,
+      "step": 9817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5188654920521575e-05,
+      "loss": 1.0232,
+      "step": 9818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5187357856453543e-05,
+      "loss": 1.0891,
+      "step": 9819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5186060732695686e-05,
+      "loss": 1.0502,
+      "step": 9820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5184763549272297e-05,
+      "loss": 1.0284,
+      "step": 9821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.5183466306207665e-05,
+      "loss": 1.0585,
+      "step": 9822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5182169003526085e-05,
+      "loss": 1.0048,
+      "step": 9823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5180871641251857e-05,
+      "loss": 1.0466,
+      "step": 9824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5179574219409267e-05,
+      "loss": 1.084,
+      "step": 9825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5178276738022618e-05,
+      "loss": 0.9641,
+      "step": 9826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5176979197116208e-05,
+      "loss": 1.0472,
+      "step": 9827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5175681596714336e-05,
+      "loss": 1.0641,
+      "step": 9828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5174383936841301e-05,
+      "loss": 1.0985,
+      "step": 9829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5173086217521407e-05,
+      "loss": 1.1077,
+      "step": 9830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5171788438778948e-05,
+      "loss": 1.0932,
+      "step": 9831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5170490600638238e-05,
+      "loss": 1.1197,
+      "step": 9832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5169192703123578e-05,
+      "loss": 1.0222,
+      "step": 9833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5167894746259272e-05,
+      "loss": 1.0507,
+      "step": 9834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5166596730069629e-05,
+      "loss": 1.0689,
+      "step": 9835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5165298654578955e-05,
+      "loss": 1.0957,
+      "step": 9836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.516400051981156e-05,
+      "loss": 1.0264,
+      "step": 9837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5162702325791759e-05,
+      "loss": 1.0358,
+      "step": 9838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5161404072543852e-05,
+      "loss": 1.0976,
+      "step": 9839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5160105760092164e-05,
+      "loss": 1.0494,
+      "step": 9840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5158807388460999e-05,
+      "loss": 1.0468,
+      "step": 9841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5157508957674678e-05,
+      "loss": 0.9753,
+      "step": 9842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5156210467757514e-05,
+      "loss": 1.0138,
+      "step": 9843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5154911918733823e-05,
+      "loss": 0.9893,
+      "step": 9844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5153613310627926e-05,
+      "loss": 1.0206,
+      "step": 9845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5152314643464139e-05,
+      "loss": 1.0464,
+      "step": 9846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5151015917266782e-05,
+      "loss": 1.1056,
+      "step": 9847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5149717132060185e-05,
+      "loss": 1.0456,
+      "step": 9848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5148418287868658e-05,
+      "loss": 1.0449,
+      "step": 9849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.514711938471653e-05,
+      "loss": 1.0908,
+      "step": 9850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5145820422628125e-05,
+      "loss": 1.0486,
+      "step": 9851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5144521401627771e-05,
+      "loss": 1.0693,
+      "step": 9852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5143222321739794e-05,
+      "loss": 1.1154,
+      "step": 9853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5141923182988518e-05,
+      "loss": 1.0409,
+      "step": 9854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5140623985398278e-05,
+      "loss": 0.9865,
+      "step": 9855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5139324728993397e-05,
+      "loss": 1.0161,
+      "step": 9856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5138025413798214e-05,
+      "loss": 1.0306,
+      "step": 9857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5136726039837057e-05,
+      "loss": 1.0403,
+      "step": 9858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5135426607134262e-05,
+      "loss": 1.0179,
+      "step": 9859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5134127115714159e-05,
+      "loss": 1.0384,
+      "step": 9860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5132827565601086e-05,
+      "loss": 1.0306,
+      "step": 9861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5131527956819382e-05,
+      "loss": 1.0311,
+      "step": 9862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5130228289393383e-05,
+      "loss": 1.092,
+      "step": 9863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5128928563347427e-05,
+      "loss": 1.047,
+      "step": 9864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5127628778705856e-05,
+      "loss": 1.0929,
+      "step": 9865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5126328935493009e-05,
+      "loss": 0.9687,
+      "step": 9866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5125029033733232e-05,
+      "loss": 1.095,
+      "step": 9867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5123729073450865e-05,
+      "loss": 1.0555,
+      "step": 9868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5122429054670251e-05,
+      "loss": 1.0655,
+      "step": 9869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5121128977415738e-05,
+      "loss": 1.0759,
+      "step": 9870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5119828841711672e-05,
+      "loss": 1.0562,
+      "step": 9871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5118528647582405e-05,
+      "loss": 1.0251,
+      "step": 9872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5117228395052277e-05,
+      "loss": 1.0683,
+      "step": 9873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5115928084145642e-05,
+      "loss": 1.0773,
+      "step": 9874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5114627714886856e-05,
+      "loss": 0.9929,
+      "step": 9875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5113327287300266e-05,
+      "loss": 1.0233,
+      "step": 9876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5112026801410226e-05,
+      "loss": 1.0158,
+      "step": 9877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5110726257241087e-05,
+      "loss": 1.0299,
+      "step": 9878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.510942565481721e-05,
+      "loss": 1.0406,
+      "step": 9879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5108124994162949e-05,
+      "loss": 1.0411,
+      "step": 9880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.510682427530266e-05,
+      "loss": 1.0219,
+      "step": 9881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5105523498260708e-05,
+      "loss": 1.0831,
+      "step": 9882
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5104222663061442e-05,
+      "loss": 1.0833,
+      "step": 9883
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.510292176972923e-05,
+      "loss": 1.0917,
+      "step": 9884
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5101620818288433e-05,
+      "loss": 1.0627,
+      "step": 9885
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5100319808763416e-05,
+      "loss": 1.0286,
+      "step": 9886
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5099018741178542e-05,
+      "loss": 1.0386,
+      "step": 9887
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5097717615558171e-05,
+      "loss": 1.0178,
+      "step": 9888
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5096416431926674e-05,
+      "loss": 1.0896,
+      "step": 9889
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5095115190308418e-05,
+      "loss": 1.0429,
+      "step": 9890
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5093813890727769e-05,
+      "loss": 1.0494,
+      "step": 9891
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5092512533209103e-05,
+      "loss": 1.1093,
+      "step": 9892
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5091211117776783e-05,
+      "loss": 1.0203,
+      "step": 9893
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5089909644455183e-05,
+      "loss": 1.0992,
+      "step": 9894
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5088608113268676e-05,
+      "loss": 1.0835,
+      "step": 9895
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5087306524241639e-05,
+      "loss": 0.9538,
+      "step": 9896
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5086004877398443e-05,
+      "loss": 1.1092,
+      "step": 9897
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5084703172763464e-05,
+      "loss": 1.1039,
+      "step": 9898
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5083401410361083e-05,
+      "loss": 1.0341,
+      "step": 9899
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5082099590215673e-05,
+      "loss": 1.0551,
+      "step": 9900
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5080797712351613e-05,
+      "loss": 1.0774,
+      "step": 9901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5079495776793294e-05,
+      "loss": 1.0917,
+      "step": 9902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5078193783565082e-05,
+      "loss": 1.0338,
+      "step": 9903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5076891732691372e-05,
+      "loss": 1.018,
+      "step": 9904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5075589624196541e-05,
+      "loss": 1.0418,
+      "step": 9905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5074287458104976e-05,
+      "loss": 1.0053,
+      "step": 9906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5072985234441063e-05,
+      "loss": 0.9847,
+      "step": 9907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5071682953229182e-05,
+      "loss": 1.0222,
+      "step": 9908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5070380614493732e-05,
+      "loss": 1.1117,
+      "step": 9909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5069078218259092e-05,
+      "loss": 1.0013,
+      "step": 9910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5067775764549661e-05,
+      "loss": 1.0612,
+      "step": 9911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5066473253389823e-05,
+      "loss": 1.0491,
+      "step": 9912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5065170684803975e-05,
+      "loss": 0.9882,
+      "step": 9913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.506386805881651e-05,
+      "loss": 1.088,
+      "step": 9914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5062565375451819e-05,
+      "loss": 1.0693,
+      "step": 9915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5061262634734299e-05,
+      "loss": 1.0972,
+      "step": 9916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5059959836688345e-05,
+      "loss": 1.0255,
+      "step": 9917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5058656981338358e-05,
+      "loss": 0.9863,
+      "step": 9918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5057354068708734e-05,
+      "loss": 0.9846,
+      "step": 9919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5056051098823872e-05,
+      "loss": 1.0929,
+      "step": 9920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5054748071708176e-05,
+      "loss": 1.039,
+      "step": 9921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5053444987386045e-05,
+      "loss": 1.0773,
+      "step": 9922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5052141845881887e-05,
+      "loss": 1.0381,
+      "step": 9923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5050838647220098e-05,
+      "loss": 1.1218,
+      "step": 9924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5049535391425087e-05,
+      "loss": 1.0529,
+      "step": 9925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5048232078521262e-05,
+      "loss": 0.9821,
+      "step": 9926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5046928708533028e-05,
+      "loss": 1.0613,
+      "step": 9927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5045625281484792e-05,
+      "loss": 1.058,
+      "step": 9928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5044321797400967e-05,
+      "loss": 1.0309,
+      "step": 9929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5043018256305962e-05,
+      "loss": 0.9974,
+      "step": 9930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5041714658224186e-05,
+      "loss": 1.0998,
+      "step": 9931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5040411003180056e-05,
+      "loss": 1.0645,
+      "step": 9932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5039107291197983e-05,
+      "loss": 0.9613,
+      "step": 9933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503780352230238e-05,
+      "loss": 1.0083,
+      "step": 9934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5036499696517666e-05,
+      "loss": 1.0212,
+      "step": 9935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5035195813868257e-05,
+      "loss": 1.049,
+      "step": 9936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.503389187437857e-05,
+      "loss": 0.9767,
+      "step": 9937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5032587878073022e-05,
+      "loss": 1.0532,
+      "step": 9938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5031283824976035e-05,
+      "loss": 1.0579,
+      "step": 9939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5029979715112036e-05,
+      "loss": 0.9748,
+      "step": 9940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5028675548505435e-05,
+      "loss": 1.0183,
+      "step": 9941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5027371325180668e-05,
+      "loss": 1.0621,
+      "step": 9942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5026067045162146e-05,
+      "loss": 1.0331,
+      "step": 9943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5024762708474307e-05,
+      "loss": 1.0698,
+      "step": 9944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.502345831514157e-05,
+      "loss": 1.0289,
+      "step": 9945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5022153865188363e-05,
+      "loss": 0.9496,
+      "step": 9946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5020849358639116e-05,
+      "loss": 1.0734,
+      "step": 9947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5019544795518256e-05,
+      "loss": 1.0699,
+      "step": 9948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5018240175850219e-05,
+      "loss": 1.0953,
+      "step": 9949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5016935499659432e-05,
+      "loss": 1.001,
+      "step": 9950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.501563076697033e-05,
+      "loss": 1.0415,
+      "step": 9951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.501432597780735e-05,
+      "loss": 1.0034,
+      "step": 9952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5013021132194915e-05,
+      "loss": 1.0485,
+      "step": 9953
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5011716230157474e-05,
+      "loss": 0.9797,
+      "step": 9954
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5010411271719458e-05,
+      "loss": 1.0591,
+      "step": 9955
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5009106256905303e-05,
+      "loss": 1.0256,
+      "step": 9956
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5007801185739455e-05,
+      "loss": 0.9922,
+      "step": 9957
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5006496058246346e-05,
+      "loss": 1.116,
+      "step": 9958
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5005190874450424e-05,
+      "loss": 1.0139,
+      "step": 9959
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5003885634376126e-05,
+      "loss": 1.0836,
+      "step": 9960
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5002580338047903e-05,
+      "loss": 1.0309,
+      "step": 9961
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.500127498549019e-05,
+      "loss": 1.0147,
+      "step": 9962
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4999969576727435e-05,
+      "loss": 1.0737,
+      "step": 9963
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4998664111784089e-05,
+      "loss": 1.059,
+      "step": 9964
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4997358590684595e-05,
+      "loss": 1.0399,
+      "step": 9965
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4996053013453404e-05,
+      "loss": 0.9915,
+      "step": 9966
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4994747380114964e-05,
+      "loss": 1.0236,
+      "step": 9967
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4993441690693728e-05,
+      "loss": 1.0865,
+      "step": 9968
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4992135945214146e-05,
+      "loss": 1.0934,
+      "step": 9969
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.499083014370067e-05,
+      "loss": 1.0605,
+      "step": 9970
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4989524286177754e-05,
+      "loss": 1.0545,
+      "step": 9971
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4988218372669854e-05,
+      "loss": 1.0901,
+      "step": 9972
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4986912403201423e-05,
+      "loss": 1.0472,
+      "step": 9973
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4985606377796926e-05,
+      "loss": 1.0418,
+      "step": 9974
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4984300296480812e-05,
+      "loss": 1.0459,
+      "step": 9975
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4982994159277541e-05,
+      "loss": 0.9863,
+      "step": 9976
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4981687966211578e-05,
+      "loss": 0.9817,
+      "step": 9977
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4980381717307381e-05,
+      "loss": 1.0181,
+      "step": 9978
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4979075412589417e-05,
+      "loss": 1.0809,
+      "step": 9979
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4977769052082143e-05,
+      "loss": 1.0182,
+      "step": 9980
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4976462635810022e-05,
+      "loss": 1.0719,
+      "step": 9981
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4975156163797525e-05,
+      "loss": 0.9633,
+      "step": 9982
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4973849636069115e-05,
+      "loss": 1.0478,
+      "step": 9983
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4972543052649264e-05,
+      "loss": 1.0068,
+      "step": 9984
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4971236413562434e-05,
+      "loss": 1.001,
+      "step": 9985
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4969929718833098e-05,
+      "loss": 0.993,
+      "step": 9986
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4968622968485726e-05,
+      "loss": 1.0986,
+      "step": 9987
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4967316162544791e-05,
+      "loss": 1.0838,
+      "step": 9988
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4966009301034764e-05,
+      "loss": 1.0318,
+      "step": 9989
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4964702383980118e-05,
+      "loss": 1.0261,
+      "step": 9990
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4963395411405328e-05,
+      "loss": 1.0542,
+      "step": 9991
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4962088383334871e-05,
+      "loss": 0.9673,
+      "step": 9992
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4960781299793223e-05,
+      "loss": 1.027,
+      "step": 9993
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4959474160804863e-05,
+      "loss": 1.0721,
+      "step": 9994
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.495816696639427e-05,
+      "loss": 1.1008,
+      "step": 9995
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4956859716585918e-05,
+      "loss": 0.9684,
+      "step": 9996
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4955552411404297e-05,
+      "loss": 1.077,
+      "step": 9997
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4954245050873883e-05,
+      "loss": 1.0438,
+      "step": 9998
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4952937635019164e-05,
+      "loss": 1.0098,
+      "step": 9999
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4951630163864617e-05,
+      "loss": 1.0159,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_text_loss": 0.9714189767837524,
+      "eval_text_runtime": 1.5683,
+      "eval_text_samples_per_second": 47.184,
+      "eval_text_steps_per_second": 1.275,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "eval_svit_loss": 1.1284552812576294,
+      "eval_svit_runtime": 3.9785,
+      "eval_svit_samples_per_second": 171.17,
+      "eval_svit_steps_per_second": 2.765,
+      "step": 10000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4950322637434731e-05,
+      "loss": 1.0178,
+      "step": 10001
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4949015055753992e-05,
+      "loss": 1.0586,
+      "step": 10002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4947707418846886e-05,
+      "loss": 1.0576,
+      "step": 10003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4946399726737906e-05,
+      "loss": 1.0182,
+      "step": 10004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4945091979451533e-05,
+      "loss": 1.096,
+      "step": 10005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4943784177012266e-05,
+      "loss": 1.0374,
+      "step": 10006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4942476319444589e-05,
+      "loss": 1.0138,
+      "step": 10007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4941168406773002e-05,
+      "loss": 0.9918,
+      "step": 10008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4939860439021989e-05,
+      "loss": 1.0049,
+      "step": 10009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4938552416216049e-05,
+      "loss": 1.0634,
+      "step": 10010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.493724433837968e-05,
+      "loss": 1.055,
+      "step": 10011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4935936205537373e-05,
+      "loss": 1.0687,
+      "step": 10012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4934628017713632e-05,
+      "loss": 1.0728,
+      "step": 10013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.493331977493295e-05,
+      "loss": 1.0315,
+      "step": 10014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4932011477219828e-05,
+      "loss": 0.9834,
+      "step": 10015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.493070312459877e-05,
+      "loss": 1.0798,
+      "step": 10016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4929394717094271e-05,
+      "loss": 1.0431,
+      "step": 10017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492808625473084e-05,
+      "loss": 1.1139,
+      "step": 10018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4926777737532976e-05,
+      "loss": 1.0711,
+      "step": 10019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4925469165525186e-05,
+      "loss": 1.0518,
+      "step": 10020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4924160538731973e-05,
+      "loss": 1.0868,
+      "step": 10021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4922851857177849e-05,
+      "loss": 1.0069,
+      "step": 10022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4921543120887318e-05,
+      "loss": 0.9926,
+      "step": 10023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.492023432988489e-05,
+      "loss": 1.0636,
+      "step": 10024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4918925484195075e-05,
+      "loss": 1.0334,
+      "step": 10025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4917616583842382e-05,
+      "loss": 1.0282,
+      "step": 10026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4916307628851323e-05,
+      "loss": 1.0623,
+      "step": 10027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4914998619246414e-05,
+      "loss": 0.9802,
+      "step": 10028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4913689555052167e-05,
+      "loss": 1.0522,
+      "step": 10029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4912380436293095e-05,
+      "loss": 1.0489,
+      "step": 10030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4911071262993714e-05,
+      "loss": 1.0386,
+      "step": 10031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4909762035178542e-05,
+      "loss": 1.038,
+      "step": 10032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4908452752872104e-05,
+      "loss": 0.9853,
+      "step": 10033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4907143416098907e-05,
+      "loss": 1.0394,
+      "step": 10034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4905834024883478e-05,
+      "loss": 1.0617,
+      "step": 10035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4904524579250337e-05,
+      "loss": 1.0197,
+      "step": 10036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4903215079224007e-05,
+      "loss": 1.0159,
+      "step": 10037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4901905524829008e-05,
+      "loss": 1.091,
+      "step": 10038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4900595916089865e-05,
+      "loss": 1.02,
+      "step": 10039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4899286253031105e-05,
+      "loss": 1.0431,
+      "step": 10040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4897976535677252e-05,
+      "loss": 1.0394,
+      "step": 10041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4896666764052839e-05,
+      "loss": 1.0385,
+      "step": 10042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4895356938182384e-05,
+      "loss": 1.0123,
+      "step": 10043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4894047058090421e-05,
+      "loss": 1.0293,
+      "step": 10044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4892737123801485e-05,
+      "loss": 0.9707,
+      "step": 10045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4891427135340102e-05,
+      "loss": 0.9856,
+      "step": 10046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4890117092730805e-05,
+      "loss": 0.9839,
+      "step": 10047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4888806995998126e-05,
+      "loss": 1.0861,
+      "step": 10048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.48874968451666e-05,
+      "loss": 1.0431,
+      "step": 10049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4886186640260765e-05,
+      "loss": 1.0769,
+      "step": 10050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4884876381305152e-05,
+      "loss": 1.0164,
+      "step": 10051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4883566068324308e-05,
+      "loss": 0.9104,
+      "step": 10052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.488225570134276e-05,
+      "loss": 0.9585,
+      "step": 10053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4880945280385054e-05,
+      "loss": 1.0209,
+      "step": 10054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4879634805475729e-05,
+      "loss": 0.9935,
+      "step": 10055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4878324276639323e-05,
+      "loss": 1.003,
+      "step": 10056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4877013693900384e-05,
+      "loss": 1.0172,
+      "step": 10057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.487570305728345e-05,
+      "loss": 1.0609,
+      "step": 10058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.487439236681307e-05,
+      "loss": 0.9944,
+      "step": 10059
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4873081622513784e-05,
+      "loss": 0.9988,
+      "step": 10060
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4871770824410146e-05,
+      "loss": 1.0423,
+      "step": 10061
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4870459972526698e-05,
+      "loss": 1.02,
+      "step": 10062
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4869149066887988e-05,
+      "loss": 1.021,
+      "step": 10063
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4867838107518568e-05,
+      "loss": 1.0072,
+      "step": 10064
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4866527094442986e-05,
+      "loss": 0.99,
+      "step": 10065
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4865216027685795e-05,
+      "loss": 1.0011,
+      "step": 10066
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.486390490727155e-05,
+      "loss": 1.0608,
+      "step": 10067
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4862593733224797e-05,
+      "loss": 1.0116,
+      "step": 10068
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4861282505570095e-05,
+      "loss": 1.0774,
+      "step": 10069
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4859971224332001e-05,
+      "loss": 1.0447,
+      "step": 10070
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.485865988953507e-05,
+      "loss": 1.0084,
+      "step": 10071
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.485734850120386e-05,
+      "loss": 1.1067,
+      "step": 10072
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4856037059362928e-05,
+      "loss": 0.9724,
+      "step": 10073
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4854725564036834e-05,
+      "loss": 1.0614,
+      "step": 10074
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4853414015250138e-05,
+      "loss": 1.0369,
+      "step": 10075
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4852102413027402e-05,
+      "loss": 1.014,
+      "step": 10076
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.485079075739319e-05,
+      "loss": 1.0326,
+      "step": 10077
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4849479048372061e-05,
+      "loss": 1.0057,
+      "step": 10078
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4848167285988583e-05,
+      "loss": 1.0356,
+      "step": 10079
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4846855470267322e-05,
+      "loss": 1.0716,
+      "step": 10080
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4845543601232846e-05,
+      "loss": 1.0191,
+      "step": 10081
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4844231678909717e-05,
+      "loss": 0.9591,
+      "step": 10082
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4842919703322505e-05,
+      "loss": 1.0534,
+      "step": 10083
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4841607674495781e-05,
+      "loss": 1.0844,
+      "step": 10084
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4840295592454114e-05,
+      "loss": 1.0457,
+      "step": 10085
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.483898345722208e-05,
+      "loss": 1.041,
+      "step": 10086
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4837671268824245e-05,
+      "loss": 1.0396,
+      "step": 10087
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4836359027285185e-05,
+      "loss": 1.022,
+      "step": 10088
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4835046732629474e-05,
+      "loss": 1.0526,
+      "step": 10089
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4833734384881693e-05,
+      "loss": 1.0626,
+      "step": 10090
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.483242198406641e-05,
+      "loss": 1.0183,
+      "step": 10091
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4831109530208207e-05,
+      "loss": 1.0436,
+      "step": 10092
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.482979702333166e-05,
+      "loss": 1.1028,
+      "step": 10093
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4828484463461348e-05,
+      "loss": 0.9757,
+      "step": 10094
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4827171850621855e-05,
+      "loss": 1.0825,
+      "step": 10095
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4825859184837762e-05,
+      "loss": 1.0348,
+      "step": 10096
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4824546466133646e-05,
+      "loss": 1.0265,
+      "step": 10097
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4823233694534099e-05,
+      "loss": 1.0209,
+      "step": 10098
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4821920870063694e-05,
+      "loss": 1.029,
+      "step": 10099
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4820607992747028e-05,
+      "loss": 1.0642,
+      "step": 10100
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4819295062608684e-05,
+      "loss": 1.0937,
+      "step": 10101
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4817982079673246e-05,
+      "loss": 1.0586,
+      "step": 10102
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4816669043965301e-05,
+      "loss": 0.9508,
+      "step": 10103
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.481535595550944e-05,
+      "loss": 1.0492,
+      "step": 10104
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4814042814330256e-05,
+      "loss": 1.0194,
+      "step": 10105
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4812729620452341e-05,
+      "loss": 1.0468,
+      "step": 10106
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4811416373900283e-05,
+      "loss": 1.0509,
+      "step": 10107
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4810103074698676e-05,
+      "loss": 1.0377,
+      "step": 10108
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4808789722872118e-05,
+      "loss": 1.0298,
+      "step": 10109
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4807476318445202e-05,
+      "loss": 1.0837,
+      "step": 10110
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4806162861442523e-05,
+      "loss": 1.0357,
+      "step": 10111
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4804849351888677e-05,
+      "loss": 1.0295,
+      "step": 10112
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4803535789808262e-05,
+      "loss": 1.0537,
+      "step": 10113
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4802222175225884e-05,
+      "loss": 1.1092,
+      "step": 10114
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4800908508166139e-05,
+      "loss": 0.9995,
+      "step": 10115
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4799594788653624e-05,
+      "loss": 1.1262,
+      "step": 10116
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4798281016712946e-05,
+      "loss": 1.0198,
+      "step": 10117
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4796967192368704e-05,
+      "loss": 1.0885,
+      "step": 10118
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4795653315645508e-05,
+      "loss": 1.0598,
+      "step": 10119
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4794339386567961e-05,
+      "loss": 0.9809,
+      "step": 10120
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4793025405160665e-05,
+      "loss": 0.9589,
+      "step": 10121
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.479171137144823e-05,
+      "loss": 1.0789,
+      "step": 10122
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4790397285455265e-05,
+      "loss": 0.9923,
+      "step": 10123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4789083147206375e-05,
+      "loss": 1.0579,
+      "step": 10124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4787768956726175e-05,
+      "loss": 1.057,
+      "step": 10125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4786454714039275e-05,
+      "loss": 1.0455,
+      "step": 10126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4785140419170283e-05,
+      "loss": 1.1229,
+      "step": 10127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4783826072143816e-05,
+      "loss": 0.9947,
+      "step": 10128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4782511672984485e-05,
+      "loss": 1.0569,
+      "step": 10129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4781197221716907e-05,
+      "loss": 1.0797,
+      "step": 10130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4779882718365695e-05,
+      "loss": 1.0809,
+      "step": 10131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4778568162955468e-05,
+      "loss": 1.0417,
+      "step": 10132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4777253555510844e-05,
+      "loss": 1.0111,
+      "step": 10133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.477593889605644e-05,
+      "loss": 1.0223,
+      "step": 10134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4774624184616878e-05,
+      "loss": 1.0791,
+      "step": 10135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4773309421216778e-05,
+      "loss": 0.9906,
+      "step": 10136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4771994605880762e-05,
+      "loss": 1.0378,
+      "step": 10137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.477067973863345e-05,
+      "loss": 1.0336,
+      "step": 10138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4769364819499468e-05,
+      "loss": 1.0369,
+      "step": 10139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4768049848503441e-05,
+      "loss": 1.0371,
+      "step": 10140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.476673482566999e-05,
+      "loss": 1.0838,
+      "step": 10141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4765419751023746e-05,
+      "loss": 1.0522,
+      "step": 10142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4764104624589336e-05,
+      "loss": 1.0904,
+      "step": 10143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4762789446391388e-05,
+      "loss": 1.0571,
+      "step": 10144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4761474216454531e-05,
+      "loss": 1.0301,
+      "step": 10145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4760158934803397e-05,
+      "loss": 1.0367,
+      "step": 10146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4758843601462615e-05,
+      "loss": 0.9975,
+      "step": 10147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4757528216456817e-05,
+      "loss": 1.0774,
+      "step": 10148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4756212779810641e-05,
+      "loss": 1.0912,
+      "step": 10149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4754897291548715e-05,
+      "loss": 1.0198,
+      "step": 10150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4753581751695676e-05,
+      "loss": 0.9534,
+      "step": 10151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4752266160276163e-05,
+      "loss": 1.1072,
+      "step": 10152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.475095051731481e-05,
+      "loss": 0.9979,
+      "step": 10153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.474963482283626e-05,
+      "loss": 1.0936,
+      "step": 10154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4748319076865144e-05,
+      "loss": 1.0193,
+      "step": 10155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4747003279426111e-05,
+      "loss": 1.0409,
+      "step": 10156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4745687430543796e-05,
+      "loss": 0.9908,
+      "step": 10157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4744371530242841e-05,
+      "loss": 1.0038,
+      "step": 10158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4743055578547894e-05,
+      "loss": 1.051,
+      "step": 10159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4741739575483591e-05,
+      "loss": 1.0614,
+      "step": 10160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4740423521074584e-05,
+      "loss": 1.0313,
+      "step": 10161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4739107415345514e-05,
+      "loss": 0.9565,
+      "step": 10162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4737791258321032e-05,
+      "loss": 1.0017,
+      "step": 10163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4736475050025785e-05,
+      "loss": 1.0158,
+      "step": 10164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4735158790484419e-05,
+      "loss": 1.0382,
+      "step": 10165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4733842479721585e-05,
+      "loss": 1.0657,
+      "step": 10166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4732526117761933e-05,
+      "loss": 1.0292,
+      "step": 10167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4731209704630114e-05,
+      "loss": 1.0578,
+      "step": 10168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4729893240350785e-05,
+      "loss": 1.0899,
+      "step": 10169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4728576724948595e-05,
+      "loss": 1.0929,
+      "step": 10170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4727260158448198e-05,
+      "loss": 1.0062,
+      "step": 10171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4725943540874252e-05,
+      "loss": 1.1246,
+      "step": 10172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4724626872251413e-05,
+      "loss": 1.0748,
+      "step": 10173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.472331015260434e-05,
+      "loss": 1.0116,
+      "step": 10174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4721993381957688e-05,
+      "loss": 0.9675,
+      "step": 10175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4720676560336115e-05,
+      "loss": 1.0791,
+      "step": 10176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4719359687764285e-05,
+      "loss": 1.0364,
+      "step": 10177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4718042764266858e-05,
+      "loss": 1.0593,
+      "step": 10178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4716725789868497e-05,
+      "loss": 1.1456,
+      "step": 10179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4715408764593861e-05,
+      "loss": 1.0682,
+      "step": 10180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4714091688467618e-05,
+      "loss": 1.0215,
+      "step": 10181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.471277456151443e-05,
+      "loss": 0.9923,
+      "step": 10182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4711457383758972e-05,
+      "loss": 0.9968,
+      "step": 10183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4710140155225895e-05,
+      "loss": 1.0474,
+      "step": 10184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470882287593988e-05,
+      "loss": 1.0335,
+      "step": 10185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470750554592559e-05,
+      "loss": 1.1284,
+      "step": 10186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4706188165207695e-05,
+      "loss": 1.0275,
+      "step": 10187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.470487073381087e-05,
+      "loss": 1.0331,
+      "step": 10188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4703553251759778e-05,
+      "loss": 0.9914,
+      "step": 10189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.47022357190791e-05,
+      "loss": 1.0035,
+      "step": 10190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4700918135793504e-05,
+      "loss": 1.0336,
+      "step": 10191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4699600501927666e-05,
+      "loss": 0.9886,
+      "step": 10192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4698282817506267e-05,
+      "loss": 0.9877,
+      "step": 10193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4696965082553977e-05,
+      "loss": 1.0939,
+      "step": 10194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4695647297095472e-05,
+      "loss": 1.022,
+      "step": 10195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4694329461155435e-05,
+      "loss": 1.0561,
+      "step": 10196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.469301157475854e-05,
+      "loss": 1.0244,
+      "step": 10197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4691693637929474e-05,
+      "loss": 1.0608,
+      "step": 10198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4690375650692911e-05,
+      "loss": 1.0747,
+      "step": 10199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4689057613073539e-05,
+      "loss": 1.0831,
+      "step": 10200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4687739525096036e-05,
+      "loss": 1.0505,
+      "step": 10201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4686421386785093e-05,
+      "loss": 1.0594,
+      "step": 10202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4685103198165385e-05,
+      "loss": 1.0228,
+      "step": 10203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4683784959261607e-05,
+      "loss": 1.0138,
+      "step": 10204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4682466670098438e-05,
+      "loss": 1.0922,
+      "step": 10205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4681148330700569e-05,
+      "loss": 1.0076,
+      "step": 10206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4679829941092689e-05,
+      "loss": 0.9856,
+      "step": 10207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.467851150129949e-05,
+      "loss": 1.0765,
+      "step": 10208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4677193011345659e-05,
+      "loss": 0.9836,
+      "step": 10209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.467587447125589e-05,
+      "loss": 1.0124,
+      "step": 10210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4674555881054868e-05,
+      "loss": 1.0931,
+      "step": 10211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4673237240767297e-05,
+      "loss": 0.9891,
+      "step": 10212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4671918550417867e-05,
+      "loss": 1.025,
+      "step": 10213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4670599810031269e-05,
+      "loss": 1.1028,
+      "step": 10214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4669281019632204e-05,
+      "loss": 1.0258,
+      "step": 10215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4667962179245367e-05,
+      "loss": 1.0602,
+      "step": 10216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.466664328889546e-05,
+      "loss": 0.9855,
+      "step": 10217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4665324348607174e-05,
+      "loss": 1.0473,
+      "step": 10218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4664005358405214e-05,
+      "loss": 1.044,
+      "step": 10219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4662686318314283e-05,
+      "loss": 1.0596,
+      "step": 10220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.466136722835908e-05,
+      "loss": 1.0564,
+      "step": 10221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4660048088564305e-05,
+      "loss": 1.0395,
+      "step": 10222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4658728898954667e-05,
+      "loss": 1.0117,
+      "step": 10223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4657409659554867e-05,
+      "loss": 1.0338,
+      "step": 10224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4656090370389608e-05,
+      "loss": 1.0194,
+      "step": 10225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4654771031483603e-05,
+      "loss": 1.0446,
+      "step": 10226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4653451642861556e-05,
+      "loss": 1.0275,
+      "step": 10227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4652132204548175e-05,
+      "loss": 1.0557,
+      "step": 10228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4650812716568167e-05,
+      "loss": 0.9918,
+      "step": 10229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4649493178946246e-05,
+      "loss": 1.0224,
+      "step": 10230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4648173591707125e-05,
+      "loss": 1.063,
+      "step": 10231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.464685395487551e-05,
+      "loss": 1.0095,
+      "step": 10232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4645534268476116e-05,
+      "loss": 0.9808,
+      "step": 10233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4644214532533655e-05,
+      "loss": 1.0419,
+      "step": 10234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4642894747072848e-05,
+      "loss": 1.0442,
+      "step": 10235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4641574912118406e-05,
+      "loss": 1.0196,
+      "step": 10236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4640255027695046e-05,
+      "loss": 1.0478,
+      "step": 10237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4638935093827487e-05,
+      "loss": 1.0267,
+      "step": 10238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4637615110540444e-05,
+      "loss": 1.0701,
+      "step": 10239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4636295077858642e-05,
+      "loss": 1.033,
+      "step": 10240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4634974995806797e-05,
+      "loss": 1.0038,
+      "step": 10241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4633654864409633e-05,
+      "loss": 1.0429,
+      "step": 10242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4632334683691866e-05,
+      "loss": 1.1075,
+      "step": 10243
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4631014453678225e-05,
+      "loss": 1.0759,
+      "step": 10244
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4629694174393433e-05,
+      "loss": 1.0255,
+      "step": 10245
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4628373845862215e-05,
+      "loss": 1.0195,
+      "step": 10246
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.46270534681093e-05,
+      "loss": 1.0037,
+      "step": 10247
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4625733041159406e-05,
+      "loss": 1.0653,
+      "step": 10248
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4624412565037271e-05,
+      "loss": 0.9706,
+      "step": 10249
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4623092039767616e-05,
+      "loss": 1.0122,
+      "step": 10250
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4621771465375177e-05,
+      "loss": 1.0884,
+      "step": 10251
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4620450841884674e-05,
+      "loss": 1.0631,
+      "step": 10252
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4619130169320847e-05,
+      "loss": 1.0323,
+      "step": 10253
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4617809447708428e-05,
+      "loss": 1.0327,
+      "step": 10254
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4616488677072146e-05,
+      "loss": 1.057,
+      "step": 10255
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4615167857436743e-05,
+      "loss": 1.0974,
+      "step": 10256
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4613846988826945e-05,
+      "loss": 1.0414,
+      "step": 10257
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4612526071267493e-05,
+      "loss": 0.9969,
+      "step": 10258
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.461120510478312e-05,
+      "loss": 1.0594,
+      "step": 10259
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460988408939857e-05,
+      "loss": 1.0334,
+      "step": 10260
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4608563025138577e-05,
+      "loss": 0.9867,
+      "step": 10261
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4607241912027881e-05,
+      "loss": 1.0514,
+      "step": 10262
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4605920750091224e-05,
+      "loss": 0.9951,
+      "step": 10263
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4604599539353348e-05,
+      "loss": 1.0531,
+      "step": 10264
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.460327827983899e-05,
+      "loss": 1.0314,
+      "step": 10265
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4601956971572904e-05,
+      "loss": 1.0135,
+      "step": 10266
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4600635614579823e-05,
+      "loss": 1.0461,
+      "step": 10267
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.45993142088845e-05,
+      "loss": 1.0086,
+      "step": 10268
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4597992754511675e-05,
+      "loss": 0.9961,
+      "step": 10269
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4596671251486099e-05,
+      "loss": 1.0827,
+      "step": 10270
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4595349699832519e-05,
+      "loss": 1.0496,
+      "step": 10271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4594028099575683e-05,
+      "loss": 1.0652,
+      "step": 10272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4592706450740341e-05,
+      "loss": 1.0548,
+      "step": 10273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4591384753351246e-05,
+      "loss": 0.9889,
+      "step": 10274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4590063007433144e-05,
+      "loss": 1.0387,
+      "step": 10275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4588741213010796e-05,
+      "loss": 1.0354,
+      "step": 10276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4587419370108943e-05,
+      "loss": 1.0606,
+      "step": 10277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4586097478752352e-05,
+      "loss": 0.9999,
+      "step": 10278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.458477553896577e-05,
+      "loss": 1.0365,
+      "step": 10279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4583453550773953e-05,
+      "loss": 1.0624,
+      "step": 10280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4582131514201664e-05,
+      "loss": 1.0145,
+      "step": 10281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4580809429273657e-05,
+      "loss": 1.0255,
+      "step": 10282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4579487296014689e-05,
+      "loss": 1.0303,
+      "step": 10283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4578165114449523e-05,
+      "loss": 0.9736,
+      "step": 10284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.457684288460292e-05,
+      "loss": 1.0776,
+      "step": 10285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4575520606499635e-05,
+      "loss": 1.1059,
+      "step": 10286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.457419828016444e-05,
+      "loss": 1.1528,
+      "step": 10287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4572875905622088e-05,
+      "loss": 0.983,
+      "step": 10288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4571553482897352e-05,
+      "loss": 1.0062,
+      "step": 10289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4570231012014994e-05,
+      "loss": 0.9714,
+      "step": 10290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4568908492999778e-05,
+      "loss": 0.9967,
+      "step": 10291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.456758592587647e-05,
+      "loss": 1.0473,
+      "step": 10292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4566263310669844e-05,
+      "loss": 1.0318,
+      "step": 10293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4564940647404664e-05,
+      "loss": 1.08,
+      "step": 10294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.4563617936105701e-05,
+      "loss": 1.0473,
+      "step": 10295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4562295176797722e-05,
+      "loss": 1.0158,
+      "step": 10296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4560972369505508e-05,
+      "loss": 1.0978,
+      "step": 10297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4559649514253819e-05,
+      "loss": 1.0646,
+      "step": 10298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4558326611067434e-05,
+      "loss": 1.0601,
+      "step": 10299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.455700365997113e-05,
+      "loss": 1.0473,
+      "step": 10300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4555680660989678e-05,
+      "loss": 1.0922,
+      "step": 10301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4554357614147852e-05,
+      "loss": 0.9577,
+      "step": 10302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4553034519470432e-05,
+      "loss": 1.0167,
+      "step": 10303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4551711376982196e-05,
+      "loss": 1.0269,
+      "step": 10304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4550388186707924e-05,
+      "loss": 1.087,
+      "step": 10305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4549064948672394e-05,
+      "loss": 1.0744,
+      "step": 10306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454774166290038e-05,
+      "loss": 0.9786,
+      "step": 10307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4546418329416674e-05,
+      "loss": 1.0451,
+      "step": 10308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.454509494824605e-05,
+      "loss": 1.0561,
+      "step": 10309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4543771519413297e-05,
+      "loss": 1.0899,
+      "step": 10310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4542448042943191e-05,
+      "loss": 1.053,
+      "step": 10311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4541124518860526e-05,
+      "loss": 1.0275,
+      "step": 10312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4539800947190081e-05,
+      "loss": 1.0333,
+      "step": 10313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4538477327956646e-05,
+      "loss": 1.0832,
+      "step": 10314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4537153661185008e-05,
+      "loss": 1.0249,
+      "step": 10315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4535829946899955e-05,
+      "loss": 1.069,
+      "step": 10316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4534506185126275e-05,
+      "loss": 1.0984,
+      "step": 10317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4533182375888759e-05,
+      "loss": 0.9848,
+      "step": 10318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.45318585192122e-05,
+      "loss": 1.0523,
+      "step": 10319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4530534615121388e-05,
+      "loss": 1.0681,
+      "step": 10320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4529210663641116e-05,
+      "loss": 0.9944,
+      "step": 10321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4527886664796176e-05,
+      "loss": 0.9901,
+      "step": 10322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4526562618611365e-05,
+      "loss": 1.069,
+      "step": 10323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4525238525111482e-05,
+      "loss": 1.1026,
+      "step": 10324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4523914384321318e-05,
+      "loss": 1.0621,
+      "step": 10325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4522590196265669e-05,
+      "loss": 1.0391,
+      "step": 10326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4521265960969336e-05,
+      "loss": 1.011,
+      "step": 10327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.451994167845712e-05,
+      "loss": 1.0724,
+      "step": 10328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4518617348753815e-05,
+      "loss": 1.0349,
+      "step": 10329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4517292971884229e-05,
+      "loss": 1.0499,
+      "step": 10330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4515968547873159e-05,
+      "loss": 1.0213,
+      "step": 10331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4514644076745405e-05,
+      "loss": 1.0116,
+      "step": 10332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.451331955852578e-05,
+      "loss": 1.0723,
+      "step": 10333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4511994993239082e-05,
+      "loss": 1.0199,
+      "step": 10334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4510670380910113e-05,
+      "loss": 1.0426,
+      "step": 10335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4509345721563683e-05,
+      "loss": 0.998,
+      "step": 10336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.45080210152246e-05,
+      "loss": 1.0464,
+      "step": 10337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.450669626191767e-05,
+      "loss": 0.9457,
+      "step": 10338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4505371461667702e-05,
+      "loss": 0.9856,
+      "step": 10339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.450404661449951e-05,
+      "loss": 1.0467,
+      "step": 10340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4502721720437893e-05,
+      "loss": 1.0084,
+      "step": 10341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4501396779507678e-05,
+      "loss": 1.0415,
+      "step": 10342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4500071791733664e-05,
+      "loss": 1.0159,
+      "step": 10343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4498746757140671e-05,
+      "loss": 1.0821,
+      "step": 10344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4497421675753509e-05,
+      "loss": 0.982,
+      "step": 10345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4496096547596994e-05,
+      "loss": 1.0415,
+      "step": 10346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4494771372695946e-05,
+      "loss": 1.0705,
+      "step": 10347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4493446151075177e-05,
+      "loss": 1.0344,
+      "step": 10348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4492120882759505e-05,
+      "loss": 1.0139,
+      "step": 10349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.449079556777375e-05,
+      "loss": 1.0266,
+      "step": 10350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.448947020614273e-05,
+      "loss": 1.0313,
+      "step": 10351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4488144797891268e-05,
+      "loss": 1.0765,
+      "step": 10352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4486819343044183e-05,
+      "loss": 1.0402,
+      "step": 10353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4485493841626292e-05,
+      "loss": 1.0078,
+      "step": 10354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4484168293662423e-05,
+      "loss": 1.0784,
+      "step": 10355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.44828426991774e-05,
+      "loss": 1.0996,
+      "step": 10356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4481517058196046e-05,
+      "loss": 1.132,
+      "step": 10357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4480191370743188e-05,
+      "loss": 0.9899,
+      "step": 10358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4478865636843648e-05,
+      "loss": 0.971,
+      "step": 10359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4477539856522257e-05,
+      "loss": 1.0519,
+      "step": 10360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4476214029803846e-05,
+      "loss": 1.0019,
+      "step": 10361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4474888156713233e-05,
+      "loss": 0.9605,
+      "step": 10362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.447356223727526e-05,
+      "loss": 1.0298,
+      "step": 10363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.447223627151475e-05,
+      "loss": 1.071,
+      "step": 10364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4470910259456534e-05,
+      "loss": 0.9925,
+      "step": 10365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4469584201125448e-05,
+      "loss": 1.0124,
+      "step": 10366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4468258096546325e-05,
+      "loss": 1.0566,
+      "step": 10367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4466931945743999e-05,
+      "loss": 1.0907,
+      "step": 10368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4465605748743301e-05,
+      "loss": 0.9678,
+      "step": 10369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4464279505569073e-05,
+      "loss": 1.0264,
+      "step": 10370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4462953216246144e-05,
+      "loss": 1.0086,
+      "step": 10371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4461626880799357e-05,
+      "loss": 0.965,
+      "step": 10372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4460300499253551e-05,
+      "loss": 1.0554,
+      "step": 10373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4458974071633562e-05,
+      "loss": 1.0273,
+      "step": 10374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4457647597964233e-05,
+      "loss": 1.028,
+      "step": 10375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4456321078270398e-05,
+      "loss": 1.1173,
+      "step": 10376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4454994512576909e-05,
+      "loss": 1.0892,
+      "step": 10377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4453667900908603e-05,
+      "loss": 1.0551,
+      "step": 10378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4452341243290323e-05,
+      "loss": 1.0421,
+      "step": 10379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4451014539746918e-05,
+      "loss": 1.0842,
+      "step": 10380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4449687790303227e-05,
+      "loss": 1.0277,
+      "step": 10381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4448360994984098e-05,
+      "loss": 1.0209,
+      "step": 10382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4447034153814384e-05,
+      "loss": 1.0391,
+      "step": 10383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4445707266818923e-05,
+      "loss": 1.0434,
+      "step": 10384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.444438033402257e-05,
+      "loss": 1.0088,
+      "step": 10385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4443053355450172e-05,
+      "loss": 1.0432,
+      "step": 10386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4441726331126582e-05,
+      "loss": 1.0243,
+      "step": 10387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.444039926107665e-05,
+      "loss": 1.0374,
+      "step": 10388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4439072145325226e-05,
+      "loss": 0.9862,
+      "step": 10389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4437744983897166e-05,
+      "loss": 0.9957,
+      "step": 10390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4436417776817322e-05,
+      "loss": 1.0479,
+      "step": 10391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4435090524110551e-05,
+      "loss": 1.0695,
+      "step": 10392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4433763225801707e-05,
+      "loss": 1.0962,
+      "step": 10393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4432435881915643e-05,
+      "loss": 1.1435,
+      "step": 10394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.443110849247722e-05,
+      "loss": 1.1118,
+      "step": 10395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4429781057511296e-05,
+      "loss": 1.0562,
+      "step": 10396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4428453577042734e-05,
+      "loss": 1.0864,
+      "step": 10397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4427126051096382e-05,
+      "loss": 1.0992,
+      "step": 10398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4425798479697114e-05,
+      "loss": 1.0019,
+      "step": 10399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4424470862869784e-05,
+      "loss": 1.0364,
+      "step": 10400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4423143200639254e-05,
+      "loss": 1.0547,
+      "step": 10401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4421815493030391e-05,
+      "loss": 1.0158,
+      "step": 10402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4420487740068054e-05,
+      "loss": 1.0454,
+      "step": 10403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4419159941777115e-05,
+      "loss": 1.0349,
+      "step": 10404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4417832098182432e-05,
+      "loss": 1.039,
+      "step": 10405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4416504209308876e-05,
+      "loss": 1.038,
+      "step": 10406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4415176275181318e-05,
+      "loss": 0.9408,
+      "step": 10407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4413848295824617e-05,
+      "loss": 1.0545,
+      "step": 10408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4412520271263651e-05,
+      "loss": 1.0433,
+      "step": 10409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4411192201523284e-05,
+      "loss": 1.0031,
+      "step": 10410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.440986408662839e-05,
+      "loss": 1.0625,
+      "step": 10411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4408535926603839e-05,
+      "loss": 0.9808,
+      "step": 10412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4407207721474503e-05,
+      "loss": 1.0432,
+      "step": 10413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4405879471265259e-05,
+      "loss": 1.1166,
+      "step": 10414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4404551176000975e-05,
+      "loss": 1.0457,
+      "step": 10415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4403222835706534e-05,
+      "loss": 1.0761,
+      "step": 10416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4401894450406807e-05,
+      "loss": 1.0827,
+      "step": 10417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4400566020126671e-05,
+      "loss": 0.9941,
+      "step": 10418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4399237544891005e-05,
+      "loss": 1.0054,
+      "step": 10419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4397909024724684e-05,
+      "loss": 1.0334,
+      "step": 10420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4396580459652594e-05,
+      "loss": 1.0625,
+      "step": 10421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4395251849699607e-05,
+      "loss": 0.9912,
+      "step": 10422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4393923194890608e-05,
+      "loss": 1.0706,
+      "step": 10423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.439259449525048e-05,
+      "loss": 1.0391,
+      "step": 10424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4391265750804104e-05,
+      "loss": 1.0336,
+      "step": 10425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4389936961576365e-05,
+      "loss": 1.0532,
+      "step": 10426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4388608127592147e-05,
+      "loss": 1.0818,
+      "step": 10427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4387279248876331e-05,
+      "loss": 1.0326,
+      "step": 10428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4385950325453808e-05,
+      "loss": 1.1059,
+      "step": 10429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4384621357349462e-05,
+      "loss": 1.0759,
+      "step": 10430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4383292344588185e-05,
+      "loss": 1.0208,
+      "step": 10431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4381963287194858e-05,
+      "loss": 1.0304,
+      "step": 10432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4380634185194374e-05,
+      "loss": 0.9386,
+      "step": 10433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4379305038611626e-05,
+      "loss": 0.9959,
+      "step": 10434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4377975847471504e-05,
+      "loss": 1.0326,
+      "step": 10435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4376646611798894e-05,
+      "loss": 1.0467,
+      "step": 10436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4375317331618699e-05,
+      "loss": 1.0528,
+      "step": 10437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4373988006955803e-05,
+      "loss": 1.0547,
+      "step": 10438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4372658637835104e-05,
+      "loss": 0.9574,
+      "step": 10439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4371329224281495e-05,
+      "loss": 1.0928,
+      "step": 10440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.436999976631988e-05,
+      "loss": 1.0143,
+      "step": 10441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4368670263975146e-05,
+      "loss": 1.0863,
+      "step": 10442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4367340717272196e-05,
+      "loss": 1.13,
+      "step": 10443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4366011126235925e-05,
+      "loss": 1.0396,
+      "step": 10444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4364681490891239e-05,
+      "loss": 0.982,
+      "step": 10445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4363351811263034e-05,
+      "loss": 1.0308,
+      "step": 10446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4362022087376206e-05,
+      "loss": 1.002,
+      "step": 10447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4360692319255662e-05,
+      "loss": 1.1241,
+      "step": 10448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4359362506926304e-05,
+      "loss": 1.0237,
+      "step": 10449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4358032650413035e-05,
+      "loss": 1.0464,
+      "step": 10450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4356702749740766e-05,
+      "loss": 1.0482,
+      "step": 10451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4355372804934392e-05,
+      "loss": 1.0457,
+      "step": 10452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.435404281601882e-05,
+      "loss": 1.0163,
+      "step": 10453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4352712783018963e-05,
+      "loss": 1.0022,
+      "step": 10454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4351382705959724e-05,
+      "loss": 1.0476,
+      "step": 10455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4350052584866017e-05,
+      "loss": 1.0236,
+      "step": 10456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4348722419762741e-05,
+      "loss": 1.115,
+      "step": 10457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4347392210674813e-05,
+      "loss": 1.0249,
+      "step": 10458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4346061957627144e-05,
+      "loss": 1.0792,
+      "step": 10459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4344731660644646e-05,
+      "loss": 1.0245,
+      "step": 10460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4343401319752229e-05,
+      "loss": 1.0615,
+      "step": 10461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4342070934974807e-05,
+      "loss": 1.0464,
+      "step": 10462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4340740506337296e-05,
+      "loss": 1.0574,
+      "step": 10463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4339410033864609e-05,
+      "loss": 1.0347,
+      "step": 10464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4338079517581664e-05,
+      "loss": 0.974,
+      "step": 10465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4336748957513375e-05,
+      "loss": 0.9986,
+      "step": 10466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.433541835368466e-05,
+      "loss": 1.0485,
+      "step": 10467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4334087706120438e-05,
+      "loss": 0.979,
+      "step": 10468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4332757014845628e-05,
+      "loss": 1.0409,
+      "step": 10469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4331426279885154e-05,
+      "loss": 1.0547,
+      "step": 10470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.433009550126393e-05,
+      "loss": 0.995,
+      "step": 10471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4328764679006878e-05,
+      "loss": 1.0365,
+      "step": 10472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4327433813138927e-05,
+      "loss": 1.023,
+      "step": 10473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4326102903684993e-05,
+      "loss": 1.0444,
+      "step": 10474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4324771950670006e-05,
+      "loss": 1.008,
+      "step": 10475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4323440954118886e-05,
+      "loss": 1.0743,
+      "step": 10476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4322109914056558e-05,
+      "loss": 1.0881,
+      "step": 10477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4320778830507953e-05,
+      "loss": 1.0753,
+      "step": 10478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4319447703497996e-05,
+      "loss": 1.0934,
+      "step": 10479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4318116533051616e-05,
+      "loss": 1.0418,
+      "step": 10480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.431678531919374e-05,
+      "loss": 1.0224,
+      "step": 10481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4315454061949299e-05,
+      "loss": 1.0891,
+      "step": 10482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4314122761343226e-05,
+      "loss": 1.0377,
+      "step": 10483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4312791417400447e-05,
+      "loss": 0.9971,
+      "step": 10484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4311460030145898e-05,
+      "loss": 1.038,
+      "step": 10485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4310128599604512e-05,
+      "loss": 1.0729,
+      "step": 10486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4308797125801218e-05,
+      "loss": 0.9736,
+      "step": 10487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4307465608760955e-05,
+      "loss": 1.0659,
+      "step": 10488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4306134048508662e-05,
+      "loss": 1.0332,
+      "step": 10489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4304802445069265e-05,
+      "loss": 1.0361,
+      "step": 10490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.430347079846771e-05,
+      "loss": 1.0511,
+      "step": 10491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4302139108728932e-05,
+      "loss": 1.053,
+      "step": 10492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4300807375877868e-05,
+      "loss": 1.0837,
+      "step": 10493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4299475599939461e-05,
+      "loss": 1.1086,
+      "step": 10494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4298143780938645e-05,
+      "loss": 1.039,
+      "step": 10495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4296811918900366e-05,
+      "loss": 0.9596,
+      "step": 10496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4295480013849566e-05,
+      "loss": 1.1837,
+      "step": 10497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4294148065811186e-05,
+      "loss": 1.0756,
+      "step": 10498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4292816074810172e-05,
+      "loss": 1.119,
+      "step": 10499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4291484040871464e-05,
+      "loss": 1.0759,
+      "step": 10500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4290151964020007e-05,
+      "loss": 1.0245,
+      "step": 10501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4288819844280754e-05,
+      "loss": 1.1043,
+      "step": 10502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4287487681678644e-05,
+      "loss": 1.0096,
+      "step": 10503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4286155476238629e-05,
+      "loss": 1.0697,
+      "step": 10504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4284823227985655e-05,
+      "loss": 1.0311,
+      "step": 10505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.428349093694467e-05,
+      "loss": 1.031,
+      "step": 10506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4282158603140627e-05,
+      "loss": 1.0374,
+      "step": 10507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4280826226598474e-05,
+      "loss": 1.0172,
+      "step": 10508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4279493807343167e-05,
+      "loss": 1.0879,
+      "step": 10509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4278161345399653e-05,
+      "loss": 0.9633,
+      "step": 10510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.427682884079289e-05,
+      "loss": 1.0165,
+      "step": 10511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4275496293547823e-05,
+      "loss": 0.9896,
+      "step": 10512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4274163703689416e-05,
+      "loss": 1.0222,
+      "step": 10513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4272831071242623e-05,
+      "loss": 1.0284,
+      "step": 10514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4271498396232396e-05,
+      "loss": 1.0367,
+      "step": 10515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4270165678683694e-05,
+      "loss": 0.9722,
+      "step": 10516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4268832918621476e-05,
+      "loss": 1.0479,
+      "step": 10517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.42675001160707e-05,
+      "loss": 0.9793,
+      "step": 10518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4266167271056327e-05,
+      "loss": 1.0275,
+      "step": 10519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4264834383603318e-05,
+      "loss": 1.0133,
+      "step": 10520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4263501453736628e-05,
+      "loss": 1.0236,
+      "step": 10521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4262168481481221e-05,
+      "loss": 0.9808,
+      "step": 10522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4260835466862068e-05,
+      "loss": 1.0224,
+      "step": 10523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425950240990412e-05,
+      "loss": 1.0407,
+      "step": 10524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425816931063235e-05,
+      "loss": 1.0704,
+      "step": 10525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4256836169071717e-05,
+      "loss": 1.0357,
+      "step": 10526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425550298524719e-05,
+      "loss": 1.0776,
+      "step": 10527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.425416975918374e-05,
+      "loss": 1.1018,
+      "step": 10528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4252836490906327e-05,
+      "loss": 1.0854,
+      "step": 10529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4251503180439924e-05,
+      "loss": 1.0743,
+      "step": 10530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4250169827809497e-05,
+      "loss": 1.0543,
+      "step": 10531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.4248836433040014e-05,
+      "loss": 1.003,
+      "step": 10532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4247502996156453e-05,
+      "loss": 0.9075,
+      "step": 10533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4246169517183777e-05,
+      "loss": 1.0283,
+      "step": 10534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4244835996146962e-05,
+      "loss": 1.0479,
+      "step": 10535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.424350243307098e-05,
+      "loss": 1.0632,
+      "step": 10536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.424216882798081e-05,
+      "loss": 1.0174,
+      "step": 10537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.424083518090142e-05,
+      "loss": 1.1161,
+      "step": 10538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4239501491857788e-05,
+      "loss": 0.9873,
+      "step": 10539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4238167760874887e-05,
+      "loss": 0.9892,
+      "step": 10540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4236833987977696e-05,
+      "loss": 1.0175,
+      "step": 10541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4235500173191195e-05,
+      "loss": 1.0245,
+      "step": 10542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4234166316540358e-05,
+      "loss": 1.0309,
+      "step": 10543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4232832418050168e-05,
+      "loss": 1.0882,
+      "step": 10544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4231498477745602e-05,
+      "loss": 1.0979,
+      "step": 10545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4230164495651644e-05,
+      "loss": 1.0375,
+      "step": 10546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4228830471793273e-05,
+      "loss": 1.0098,
+      "step": 10547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4227496406195473e-05,
+      "loss": 1.0111,
+      "step": 10548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4226162298883226e-05,
+      "loss": 1.025,
+      "step": 10549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4224828149881516e-05,
+      "loss": 1.0209,
+      "step": 10550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4223493959215327e-05,
+      "loss": 1.1238,
+      "step": 10551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4222159726909644e-05,
+      "loss": 1.0338,
+      "step": 10552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4220825452989457e-05,
+      "loss": 1.081,
+      "step": 10553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.421949113747975e-05,
+      "loss": 0.9765,
+      "step": 10554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.421815678040551e-05,
+      "loss": 1.0669,
+      "step": 10555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4216822381791728e-05,
+      "loss": 0.948,
+      "step": 10556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4215487941663393e-05,
+      "loss": 1.0667,
+      "step": 10557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4214153460045495e-05,
+      "loss": 1.0778,
+      "step": 10558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4212818936963021e-05,
+      "loss": 1.0128,
+      "step": 10559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4211484372440969e-05,
+      "loss": 1.0253,
+      "step": 10560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4210149766504326e-05,
+      "loss": 1.0118,
+      "step": 10561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.420881511917809e-05,
+      "loss": 1.0102,
+      "step": 10562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4207480430487251e-05,
+      "loss": 1.0482,
+      "step": 10563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4206145700456805e-05,
+      "loss": 1.103,
+      "step": 10564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.420481092911175e-05,
+      "loss": 0.9906,
+      "step": 10565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.420347611647708e-05,
+      "loss": 1.1347,
+      "step": 10566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.420214126257779e-05,
+      "loss": 1.1293,
+      "step": 10567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4200806367438885e-05,
+      "loss": 1.0132,
+      "step": 10568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4199471431085356e-05,
+      "loss": 1.0156,
+      "step": 10569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4198136453542207e-05,
+      "loss": 1.1184,
+      "step": 10570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4196801434834433e-05,
+      "loss": 0.9899,
+      "step": 10571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4195466374987044e-05,
+      "loss": 1.0296,
+      "step": 10572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4194131274025031e-05,
+      "loss": 1.079,
+      "step": 10573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4192796131973405e-05,
+      "loss": 1.0637,
+      "step": 10574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4191460948857165e-05,
+      "loss": 1.0717,
+      "step": 10575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4190125724701318e-05,
+      "loss": 1.1016,
+      "step": 10576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4188790459530868e-05,
+      "loss": 1.017,
+      "step": 10577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4187455153370817e-05,
+      "loss": 1.0121,
+      "step": 10578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4186119806246173e-05,
+      "loss": 1.0581,
+      "step": 10579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4184784418181945e-05,
+      "loss": 1.0572,
+      "step": 10580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4183448989203138e-05,
+      "loss": 1.0346,
+      "step": 10581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4182113519334768e-05,
+      "loss": 1.0354,
+      "step": 10582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4180778008601836e-05,
+      "loss": 1.0671,
+      "step": 10583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4179442457029356e-05,
+      "loss": 1.0915,
+      "step": 10584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4178106864642337e-05,
+      "loss": 1.076,
+      "step": 10585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4176771231465792e-05,
+      "loss": 1.0606,
+      "step": 10586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4175435557524735e-05,
+      "loss": 1.035,
+      "step": 10587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4174099842844175e-05,
+      "loss": 1.0524,
+      "step": 10588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4172764087449129e-05,
+      "loss": 1.0546,
+      "step": 10589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4171428291364612e-05,
+      "loss": 1.1188,
+      "step": 10590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4170092454615639e-05,
+      "loss": 1.0927,
+      "step": 10591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4168756577227228e-05,
+      "loss": 0.9988,
+      "step": 10592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4167420659224391e-05,
+      "loss": 1.0709,
+      "step": 10593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.416608470063215e-05,
+      "loss": 1.0826,
+      "step": 10594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4164748701475524e-05,
+      "loss": 1.0075,
+      "step": 10595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4163412661779532e-05,
+      "loss": 1.0515,
+      "step": 10596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.416207658156919e-05,
+      "loss": 1.0215,
+      "step": 10597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4160740460869521e-05,
+      "loss": 1.0426,
+      "step": 10598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.415940429970555e-05,
+      "loss": 1.0187,
+      "step": 10599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4158068098102294e-05,
+      "loss": 1.0153,
+      "step": 10600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4156731856084783e-05,
+      "loss": 1.0872,
+      "step": 10601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4155395573678034e-05,
+      "loss": 0.9736,
+      "step": 10602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4154059250907075e-05,
+      "loss": 0.978,
+      "step": 10603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4152722887796932e-05,
+      "loss": 1.0759,
+      "step": 10604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.415138648437263e-05,
+      "loss": 1.0373,
+      "step": 10605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4150050040659193e-05,
+      "loss": 1.0326,
+      "step": 10606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4148713556681655e-05,
+      "loss": 0.981,
+      "step": 10607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4147377032465036e-05,
+      "loss": 0.9575,
+      "step": 10608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4146040468034375e-05,
+      "loss": 1.0288,
+      "step": 10609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4144703863414695e-05,
+      "loss": 0.9958,
+      "step": 10610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.414336721863103e-05,
+      "loss": 1.0931,
+      "step": 10611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4142030533708408e-05,
+      "loss": 1.094,
+      "step": 10612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4140693808671866e-05,
+      "loss": 1.0592,
+      "step": 10613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4139357043546433e-05,
+      "loss": 1.0554,
+      "step": 10614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4138020238357142e-05,
+      "loss": 0.9996,
+      "step": 10615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4136683393129033e-05,
+      "loss": 1.0175,
+      "step": 10616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4135346507887136e-05,
+      "loss": 0.9868,
+      "step": 10617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4134009582656487e-05,
+      "loss": 1.0572,
+      "step": 10618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4132672617462124e-05,
+      "loss": 1.0701,
+      "step": 10619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4131335612329085e-05,
+      "loss": 1.0808,
+      "step": 10620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4129998567282408e-05,
+      "loss": 1.0485,
+      "step": 10621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4128661482347132e-05,
+      "loss": 1.0603,
+      "step": 10622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4127324357548298e-05,
+      "loss": 1.0443,
+      "step": 10623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4125987192910942e-05,
+      "loss": 0.9985,
+      "step": 10624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4124649988460107e-05,
+      "loss": 0.9974,
+      "step": 10625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4123312744220839e-05,
+      "loss": 1.1173,
+      "step": 10626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4121975460218174e-05,
+      "loss": 1.0423,
+      "step": 10627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4120638136477158e-05,
+      "loss": 1.048,
+      "step": 10628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411930077302284e-05,
+      "loss": 1.0415,
+      "step": 10629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4117963369880258e-05,
+      "loss": 1.0588,
+      "step": 10630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.411662592707446e-05,
+      "loss": 1.0492,
+      "step": 10631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4115288444630497e-05,
+      "loss": 1.03,
+      "step": 10632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4113950922573408e-05,
+      "loss": 1.0189,
+      "step": 10633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4112613360928243e-05,
+      "loss": 1.059,
+      "step": 10634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4111275759720057e-05,
+      "loss": 1.0321,
+      "step": 10635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4109938118973891e-05,
+      "loss": 1.0184,
+      "step": 10636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4108600438714802e-05,
+      "loss": 0.9837,
+      "step": 10637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4107262718967833e-05,
+      "loss": 0.9998,
+      "step": 10638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.410592495975804e-05,
+      "loss": 1.0175,
+      "step": 10639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4104587161110481e-05,
+      "loss": 1.0204,
+      "step": 10640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4103249323050199e-05,
+      "loss": 1.0517,
+      "step": 10641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4101911445602254e-05,
+      "loss": 1.064,
+      "step": 10642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.41005735287917e-05,
+      "loss": 1.1055,
+      "step": 10643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4099235572643586e-05,
+      "loss": 1.0169,
+      "step": 10644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4097897577182978e-05,
+      "loss": 1.0623,
+      "step": 10645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4096559542434925e-05,
+      "loss": 1.0124,
+      "step": 10646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4095221468424486e-05,
+      "loss": 0.9989,
+      "step": 10647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4093883355176721e-05,
+      "loss": 1.0779,
+      "step": 10648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4092545202716688e-05,
+      "loss": 1.0351,
+      "step": 10649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4091207011069447e-05,
+      "loss": 0.9882,
+      "step": 10650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.408986878026006e-05,
+      "loss": 1.016,
+      "step": 10651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4088530510313584e-05,
+      "loss": 1.0783,
+      "step": 10652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4087192201255084e-05,
+      "loss": 1.0554,
+      "step": 10653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.408585385310962e-05,
+      "loss": 1.0738,
+      "step": 10654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4084515465902261e-05,
+      "loss": 1.0102,
+      "step": 10655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4083177039658064e-05,
+      "loss": 1.0476,
+      "step": 10656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4081838574402098e-05,
+      "loss": 1.067,
+      "step": 10657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4080500070159426e-05,
+      "loss": 1.0371,
+      "step": 10658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4079161526955118e-05,
+      "loss": 1.1322,
+      "step": 10659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4077822944814237e-05,
+      "loss": 1.0531,
+      "step": 10660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4076484323761853e-05,
+      "loss": 1.0798,
+      "step": 10661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4075145663823033e-05,
+      "loss": 1.0829,
+      "step": 10662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4073806965022847e-05,
+      "loss": 1.0627,
+      "step": 10663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4072468227386366e-05,
+      "loss": 1.056,
+      "step": 10664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4071129450938658e-05,
+      "loss": 1.1019,
+      "step": 10665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4069790635704796e-05,
+      "loss": 1.0739,
+      "step": 10666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4068451781709851e-05,
+      "loss": 1.0553,
+      "step": 10667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4067112888978897e-05,
+      "loss": 1.0625,
+      "step": 10668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4065773957537009e-05,
+      "loss": 1.0651,
+      "step": 10669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4064434987409258e-05,
+      "loss": 1.0696,
+      "step": 10670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4063095978620719e-05,
+      "loss": 0.988,
+      "step": 10671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4061756931196471e-05,
+      "loss": 1.0244,
+      "step": 10672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4060417845161587e-05,
+      "loss": 0.9634,
+      "step": 10673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4059078720541149e-05,
+      "loss": 1.1021,
+      "step": 10674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4057739557360228e-05,
+      "loss": 1.0977,
+      "step": 10675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4056400355643905e-05,
+      "loss": 0.9938,
+      "step": 10676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.405506111541726e-05,
+      "loss": 1.0521,
+      "step": 10677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4053721836705378e-05,
+      "loss": 1.0138,
+      "step": 10678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4052382519533332e-05,
+      "loss": 1.0493,
+      "step": 10679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4051043163926207e-05,
+      "loss": 1.0709,
+      "step": 10680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4049703769909085e-05,
+      "loss": 1.1074,
+      "step": 10681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4048364337507048e-05,
+      "loss": 1.0508,
+      "step": 10682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.404702486674518e-05,
+      "loss": 1.0469,
+      "step": 10683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4045685357648567e-05,
+      "loss": 1.0754,
+      "step": 10684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.404434581024229e-05,
+      "loss": 1.0739,
+      "step": 10685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4043006224551439e-05,
+      "loss": 1.0447,
+      "step": 10686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4041666600601098e-05,
+      "loss": 1.0361,
+      "step": 10687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.404032693841636e-05,
+      "loss": 1.106,
+      "step": 10688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4038987238022303e-05,
+      "loss": 1.0571,
+      "step": 10689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4037647499444022e-05,
+      "loss": 1.0208,
+      "step": 10690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4036307722706606e-05,
+      "loss": 1.0688,
+      "step": 10691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4034967907835142e-05,
+      "loss": 1.0479,
+      "step": 10692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4033628054854722e-05,
+      "loss": 0.9802,
+      "step": 10693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403228816379044e-05,
+      "loss": 0.9624,
+      "step": 10694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.403094823466739e-05,
+      "loss": 1.0322,
+      "step": 10695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4029608267510657e-05,
+      "loss": 1.019,
+      "step": 10696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4028268262345342e-05,
+      "loss": 1.0469,
+      "step": 10697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4026928219196537e-05,
+      "loss": 1.009,
+      "step": 10698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4025588138089334e-05,
+      "loss": 1.0413,
+      "step": 10699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.402424801904883e-05,
+      "loss": 1.1035,
+      "step": 10700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4022907862100129e-05,
+      "loss": 1.0763,
+      "step": 10701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4021567667268317e-05,
+      "loss": 1.0522,
+      "step": 10702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40202274345785e-05,
+      "loss": 1.0671,
+      "step": 10703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.401888716405577e-05,
+      "loss": 1.0052,
+      "step": 10704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.401754685572523e-05,
+      "loss": 1.0099,
+      "step": 10705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4016206509611985e-05,
+      "loss": 1.0276,
+      "step": 10706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4014866125741123e-05,
+      "loss": 0.9901,
+      "step": 10707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4013525704137763e-05,
+      "loss": 1.0521,
+      "step": 10708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.401218524482699e-05,
+      "loss": 1.0848,
+      "step": 10709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4010844747833914e-05,
+      "loss": 1.0139,
+      "step": 10710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400950421318364e-05,
+      "loss": 1.1385,
+      "step": 10711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400816364090127e-05,
+      "loss": 1.061,
+      "step": 10712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4006823031011913e-05,
+      "loss": 1.0355,
+      "step": 10713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400548238354067e-05,
+      "loss": 1.0295,
+      "step": 10714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4004141698512648e-05,
+      "loss": 1.0936,
+      "step": 10715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.4002800975952957e-05,
+      "loss": 0.9674,
+      "step": 10716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.40014602158867e-05,
+      "loss": 0.9967,
+      "step": 10717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.400011941833899e-05,
+      "loss": 0.9849,
+      "step": 10718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3998778583334935e-05,
+      "loss": 1.0445,
+      "step": 10719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3997437710899644e-05,
+      "loss": 1.0319,
+      "step": 10720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3996096801058225e-05,
+      "loss": 1.0466,
+      "step": 10721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3994755853835795e-05,
+      "loss": 1.0416,
+      "step": 10722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3993414869257464e-05,
+      "loss": 1.0679,
+      "step": 10723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3992073847348344e-05,
+      "loss": 1.0302,
+      "step": 10724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.399073278813355e-05,
+      "loss": 0.9825,
+      "step": 10725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.398939169163819e-05,
+      "loss": 1.0342,
+      "step": 10726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3988050557887384e-05,
+      "loss": 1.0404,
+      "step": 10727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.398670938690625e-05,
+      "loss": 1.0668,
+      "step": 10728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3985368178719897e-05,
+      "loss": 0.9743,
+      "step": 10729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3984026933353448e-05,
+      "loss": 1.0908,
+      "step": 10730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3982685650832015e-05,
+      "loss": 1.0062,
+      "step": 10731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3981344331180724e-05,
+      "loss": 1.1121,
+      "step": 10732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3980002974424689e-05,
+      "loss": 1.1166,
+      "step": 10733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3978661580589027e-05,
+      "loss": 1.0937,
+      "step": 10734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3977320149698866e-05,
+      "loss": 0.9963,
+      "step": 10735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.397597868177932e-05,
+      "loss": 0.9979,
+      "step": 10736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3974637176855514e-05,
+      "loss": 1.0544,
+      "step": 10737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3973295634952569e-05,
+      "loss": 1.092,
+      "step": 10738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.397195405609561e-05,
+      "loss": 1.0285,
+      "step": 10739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3970612440309756e-05,
+      "loss": 1.0641,
+      "step": 10740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3969270787620138e-05,
+      "loss": 1.0292,
+      "step": 10741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3967929098051882e-05,
+      "loss": 1.0124,
+      "step": 10742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3966587371630107e-05,
+      "loss": 1.1088,
+      "step": 10743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3965245608379945e-05,
+      "loss": 1.0295,
+      "step": 10744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3963903808326518e-05,
+      "loss": 1.0921,
+      "step": 10745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3962561971494958e-05,
+      "loss": 1.0654,
+      "step": 10746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3961220097910394e-05,
+      "loss": 1.053,
+      "step": 10747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3959878187597953e-05,
+      "loss": 1.0286,
+      "step": 10748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3958536240582767e-05,
+      "loss": 1.0438,
+      "step": 10749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3957194256889966e-05,
+      "loss": 0.9934,
+      "step": 10750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.395585223654468e-05,
+      "loss": 1.0054,
+      "step": 10751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3954510179572044e-05,
+      "loss": 0.9743,
+      "step": 10752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3953168085997188e-05,
+      "loss": 1.089,
+      "step": 10753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.395182595584525e-05,
+      "loss": 1.0312,
+      "step": 10754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3950483789141357e-05,
+      "loss": 1.0857,
+      "step": 10755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3949141585910647e-05,
+      "loss": 0.9787,
+      "step": 10756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.394779934617826e-05,
+      "loss": 1.0542,
+      "step": 10757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3946457069969326e-05,
+      "loss": 0.9644,
+      "step": 10758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3945114757308983e-05,
+      "loss": 1.0566,
+      "step": 10759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.394377240822237e-05,
+      "loss": 1.091,
+      "step": 10760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3942430022734628e-05,
+      "loss": 1.0422,
+      "step": 10761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3941087600870892e-05,
+      "loss": 0.9859,
+      "step": 10762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3939745142656304e-05,
+      "loss": 1.0445,
+      "step": 10763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3938402648116e-05,
+      "loss": 1.0213,
+      "step": 10764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3937060117275123e-05,
+      "loss": 1.0532,
+      "step": 10765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3935717550158818e-05,
+      "loss": 1.0548,
+      "step": 10766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3934374946792224e-05,
+      "loss": 1.0558,
+      "step": 10767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3933032307200482e-05,
+      "loss": 1.0604,
+      "step": 10768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.3931689631408743e-05,
+      "loss": 1.0116,
+      "step": 10769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3930346919442143e-05,
+      "loss": 1.0564,
+      "step": 10770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3929004171325835e-05,
+      "loss": 1.043,
+      "step": 10771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3927661387084957e-05,
+      "loss": 1.0468,
+      "step": 10772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3926318566744661e-05,
+      "loss": 1.0135,
+      "step": 10773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3924975710330091e-05,
+      "loss": 1.0557,
+      "step": 10774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3923632817866395e-05,
+      "loss": 1.0016,
+      "step": 10775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3922289889378725e-05,
+      "loss": 1.0599,
+      "step": 10776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3920946924892224e-05,
+      "loss": 0.9994,
+      "step": 10777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3919603924432048e-05,
+      "loss": 0.994,
+      "step": 10778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3918260888023342e-05,
+      "loss": 1.0536,
+      "step": 10779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.391691781569126e-05,
+      "loss": 1.0047,
+      "step": 10780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3915574707460957e-05,
+      "loss": 1.0324,
+      "step": 10781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.391423156335758e-05,
+      "loss": 0.9795,
+      "step": 10782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3912888383406281e-05,
+      "loss": 1.0231,
+      "step": 10783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.391154516763222e-05,
+      "loss": 1.0981,
+      "step": 10784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3910201916060546e-05,
+      "loss": 1.0055,
+      "step": 10785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3908858628716419e-05,
+      "loss": 1.0739,
+      "step": 10786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3907515305624993e-05,
+      "loss": 1.0758,
+      "step": 10787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3906171946811419e-05,
+      "loss": 1.0414,
+      "step": 10788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3904828552300862e-05,
+      "loss": 1.0468,
+      "step": 10789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3903485122118479e-05,
+      "loss": 1.0742,
+      "step": 10790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3902141656289424e-05,
+      "loss": 1.062,
+      "step": 10791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3900798154838856e-05,
+      "loss": 1.1077,
+      "step": 10792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.389945461779194e-05,
+      "loss": 1.0008,
+      "step": 10793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3898111045173834e-05,
+      "loss": 0.9799,
+      "step": 10794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3896767437009697e-05,
+      "loss": 0.9531,
+      "step": 10795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3895423793324694e-05,
+      "loss": 0.9881,
+      "step": 10796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3894080114143986e-05,
+      "loss": 1.0403,
+      "step": 10797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3892736399492734e-05,
+      "loss": 1.0312,
+      "step": 10798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3891392649396109e-05,
+      "loss": 1.0357,
+      "step": 10799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3890048863879266e-05,
+      "loss": 1.0394,
+      "step": 10800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3888705042967378e-05,
+      "loss": 1.015,
+      "step": 10801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3887361186685606e-05,
+      "loss": 1.0475,
+      "step": 10802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3886017295059118e-05,
+      "loss": 1.0003,
+      "step": 10803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3884673368113082e-05,
+      "loss": 1.1276,
+      "step": 10804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3883329405872665e-05,
+      "loss": 1.0539,
+      "step": 10805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3881985408363035e-05,
+      "loss": 1.0109,
+      "step": 10806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.388064137560936e-05,
+      "loss": 0.9567,
+      "step": 10807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3879297307636811e-05,
+      "loss": 1.0139,
+      "step": 10808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3877953204470562e-05,
+      "loss": 1.036,
+      "step": 10809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3876609066135782e-05,
+      "loss": 1.0719,
+      "step": 10810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3875264892657637e-05,
+      "loss": 1.0147,
+      "step": 10811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3873920684061305e-05,
+      "loss": 1.0254,
+      "step": 10812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.387257644037196e-05,
+      "loss": 0.9464,
+      "step": 10813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3871232161614772e-05,
+      "loss": 0.9539,
+      "step": 10814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3869887847814918e-05,
+      "loss": 1.023,
+      "step": 10815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3868543498997572e-05,
+      "loss": 1.0561,
+      "step": 10816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.386719911518791e-05,
+      "loss": 1.0225,
+      "step": 10817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3865854696411109e-05,
+      "loss": 1.014,
+      "step": 10818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3864510242692344e-05,
+      "loss": 1.0772,
+      "step": 10819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3863165754056796e-05,
+      "loss": 0.9934,
+      "step": 10820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.386182123052964e-05,
+      "loss": 1.0517,
+      "step": 10821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3860476672136054e-05,
+      "loss": 0.9922,
+      "step": 10822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.385913207890122e-05,
+      "loss": 1.075,
+      "step": 10823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3857787450850319e-05,
+      "loss": 0.9807,
+      "step": 10824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3856442788008535e-05,
+      "loss": 0.9703,
+      "step": 10825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3855098090401041e-05,
+      "loss": 1.077,
+      "step": 10826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3853753358053025e-05,
+      "loss": 1.042,
+      "step": 10827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.385240859098967e-05,
+      "loss": 1.0379,
+      "step": 10828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3851063789236156e-05,
+      "loss": 1.024,
+      "step": 10829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3849718952817672e-05,
+      "loss": 1.0592,
+      "step": 10830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.38483740817594e-05,
+      "loss": 0.9994,
+      "step": 10831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384702917608652e-05,
+      "loss": 1.0939,
+      "step": 10832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384568423582423e-05,
+      "loss": 1.0335,
+      "step": 10833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.384433926099771e-05,
+      "loss": 1.0584,
+      "step": 10834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3842994251632146e-05,
+      "loss": 1.0824,
+      "step": 10835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3841649207752727e-05,
+      "loss": 1.0477,
+      "step": 10836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3840304129384646e-05,
+      "loss": 1.0308,
+      "step": 10837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3838959016553087e-05,
+      "loss": 1.0374,
+      "step": 10838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3837613869283246e-05,
+      "loss": 0.9978,
+      "step": 10839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3836268687600305e-05,
+      "loss": 1.0264,
+      "step": 10840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3834923471529462e-05,
+      "loss": 1.0549,
+      "step": 10841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3833578221095905e-05,
+      "loss": 1.082,
+      "step": 10842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.383223293632483e-05,
+      "loss": 1.0628,
+      "step": 10843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3830887617241431e-05,
+      "loss": 0.9822,
+      "step": 10844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3829542263870896e-05,
+      "loss": 1.0861,
+      "step": 10845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3828196876238425e-05,
+      "loss": 1.0352,
+      "step": 10846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3826851454369212e-05,
+      "loss": 1.0501,
+      "step": 10847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3825505998288451e-05,
+      "loss": 0.9975,
+      "step": 10848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.382416050802134e-05,
+      "loss": 1.0817,
+      "step": 10849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3822814983593075e-05,
+      "loss": 1.0401,
+      "step": 10850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3821469425028854e-05,
+      "loss": 1.0673,
+      "step": 10851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3820123832353878e-05,
+      "loss": 1.0319,
+      "step": 10852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.381877820559334e-05,
+      "loss": 1.0931,
+      "step": 10853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3817432544772447e-05,
+      "loss": 0.986,
+      "step": 10854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3816086849916393e-05,
+      "loss": 1.1135,
+      "step": 10855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3814741121050384e-05,
+      "loss": 1.0494,
+      "step": 10856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3813395358199617e-05,
+      "loss": 1.0282,
+      "step": 10857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3812049561389295e-05,
+      "loss": 1.0314,
+      "step": 10858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3810703730644625e-05,
+      "loss": 1.1008,
+      "step": 10859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3809357865990804e-05,
+      "loss": 1.0628,
+      "step": 10860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.380801196745304e-05,
+      "loss": 1.0692,
+      "step": 10861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3806666035056539e-05,
+      "loss": 1.0947,
+      "step": 10862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3805320068826504e-05,
+      "loss": 1.0236,
+      "step": 10863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3803974068788141e-05,
+      "loss": 0.9714,
+      "step": 10864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3802628034966659e-05,
+      "loss": 0.9698,
+      "step": 10865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3801281967387263e-05,
+      "loss": 1.0267,
+      "step": 10866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.379993586607516e-05,
+      "loss": 0.9849,
+      "step": 10867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.379858973105556e-05,
+      "loss": 1.0032,
+      "step": 10868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3797243562353673e-05,
+      "loss": 1.0322,
+      "step": 10869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3795897359994704e-05,
+      "loss": 1.0147,
+      "step": 10870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3794551124003871e-05,
+      "loss": 1.0276,
+      "step": 10871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3793204854406379e-05,
+      "loss": 1.0294,
+      "step": 10872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3791858551227445e-05,
+      "loss": 1.0003,
+      "step": 10873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3790512214492275e-05,
+      "loss": 1.0003,
+      "step": 10874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3789165844226086e-05,
+      "loss": 0.9975,
+      "step": 10875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378781944045409e-05,
+      "loss": 1.102,
+      "step": 10876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3786473003201504e-05,
+      "loss": 0.9466,
+      "step": 10877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378512653249354e-05,
+      "loss": 1.0348,
+      "step": 10878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.378378002835541e-05,
+      "loss": 1.0951,
+      "step": 10879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3782433490812339e-05,
+      "loss": 1.0227,
+      "step": 10880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3781086919889535e-05,
+      "loss": 0.9905,
+      "step": 10881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377974031561222e-05,
+      "loss": 1.0822,
+      "step": 10882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3778393678005613e-05,
+      "loss": 1.0332,
+      "step": 10883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377704700709493e-05,
+      "loss": 1.0816,
+      "step": 10884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377570030290539e-05,
+      "loss": 1.0096,
+      "step": 10885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3774353565462212e-05,
+      "loss": 1.0142,
+      "step": 10886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.377300679479062e-05,
+      "loss": 0.9915,
+      "step": 10887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3771659990915834e-05,
+      "loss": 0.9724,
+      "step": 10888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3770313153863072e-05,
+      "loss": 1.0526,
+      "step": 10889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3768966283657558e-05,
+      "loss": 1.0619,
+      "step": 10890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376761938032452e-05,
+      "loss": 1.0078,
+      "step": 10891
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3766272443889175e-05,
+      "loss": 1.1005,
+      "step": 10892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376492547437675e-05,
+      "loss": 1.0194,
+      "step": 10893
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.376357847181247e-05,
+      "loss": 1.005,
+      "step": 10894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3762231436221558e-05,
+      "loss": 1.0603,
+      "step": 10895
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3760884367629242e-05,
+      "loss": 0.9475,
+      "step": 10896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3759537266060747e-05,
+      "loss": 1.0482,
+      "step": 10897
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3758190131541304e-05,
+      "loss": 1.055,
+      "step": 10898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.375684296409614e-05,
+      "loss": 1.0226,
+      "step": 10899
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3755495763750476e-05,
+      "loss": 1.0561,
+      "step": 10900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3754148530529551e-05,
+      "loss": 1.0391,
+      "step": 10901
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3752801264458593e-05,
+      "loss": 1.0006,
+      "step": 10902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3751453965562828e-05,
+      "loss": 1.0361,
+      "step": 10903
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3750106633867488e-05,
+      "loss": 0.9638,
+      "step": 10904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3748759269397806e-05,
+      "loss": 1.0369,
+      "step": 10905
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3747411872179013e-05,
+      "loss": 1.0606,
+      "step": 10906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3746064442236344e-05,
+      "loss": 1.0166,
+      "step": 10907
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3744716979595031e-05,
+      "loss": 1.001,
+      "step": 10908
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3743369484280306e-05,
+      "loss": 1.0485,
+      "step": 10909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3742021956317406e-05,
+      "loss": 1.0706,
+      "step": 10910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3740674395731568e-05,
+      "loss": 0.9582,
+      "step": 10911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3739326802548022e-05,
+      "loss": 1.0209,
+      "step": 10912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.373797917679201e-05,
+      "loss": 1.0505,
+      "step": 10913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.373663151848877e-05,
+      "loss": 1.1215,
+      "step": 10914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3735283827663532e-05,
+      "loss": 0.9859,
+      "step": 10915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3733936104341541e-05,
+      "loss": 1.0175,
+      "step": 10916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3732588348548036e-05,
+      "loss": 1.0334,
+      "step": 10917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.373124056030825e-05,
+      "loss": 1.0368,
+      "step": 10918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3729892739647428e-05,
+      "loss": 1.0143,
+      "step": 10919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3728544886590813e-05,
+      "loss": 1.111,
+      "step": 10920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3727197001163643e-05,
+      "loss": 1.0495,
+      "step": 10921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.372584908339116e-05,
+      "loss": 1.0174,
+      "step": 10922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3724501133298608e-05,
+      "loss": 1.0527,
+      "step": 10923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3723153150911228e-05,
+      "loss": 0.9692,
+      "step": 10924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3721805136254265e-05,
+      "loss": 0.997,
+      "step": 10925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3720457089352961e-05,
+      "loss": 1.1052,
+      "step": 10926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3719109010232568e-05,
+      "loss": 1.0075,
+      "step": 10927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3717760898918325e-05,
+      "loss": 1.0489,
+      "step": 10928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3716412755435478e-05,
+      "loss": 0.9872,
+      "step": 10929
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3715064579809275e-05,
+      "loss": 1.0412,
+      "step": 10930
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3713716372064966e-05,
+      "loss": 1.0652,
+      "step": 10931
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.37123681322278e-05,
+      "loss": 1.0827,
+      "step": 10932
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3711019860323018e-05,
+      "loss": 1.0085,
+      "step": 10933
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3709671556375874e-05,
+      "loss": 1.0701,
+      "step": 10934
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3708323220411617e-05,
+      "loss": 1.067,
+      "step": 10935
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3706974852455501e-05,
+      "loss": 1.0418,
+      "step": 10936
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3705626452532772e-05,
+      "loss": 0.9649,
+      "step": 10937
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3704278020668681e-05,
+      "loss": 1.0623,
+      "step": 10938
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3702929556888484e-05,
+      "loss": 1.0063,
+      "step": 10939
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3701581061217437e-05,
+      "loss": 1.0883,
+      "step": 10940
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3700232533680787e-05,
+      "loss": 1.004,
+      "step": 10941
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3698883974303785e-05,
+      "loss": 1.078,
+      "step": 10942
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.369753538311169e-05,
+      "loss": 1.0255,
+      "step": 10943
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3696186760129758e-05,
+      "loss": 1.0002,
+      "step": 10944
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3694838105383247e-05,
+      "loss": 1.0728,
+      "step": 10945
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.369348941889741e-05,
+      "loss": 0.9863,
+      "step": 10946
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3692140700697502e-05,
+      "loss": 1.0485,
+      "step": 10947
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3690791950808784e-05,
+      "loss": 0.9448,
+      "step": 10948
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3689443169256513e-05,
+      "loss": 1.0823,
+      "step": 10949
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3688094356065946e-05,
+      "loss": 1.0238,
+      "step": 10950
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3686745511262347e-05,
+      "loss": 1.0271,
+      "step": 10951
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.368539663487097e-05,
+      "loss": 0.9574,
+      "step": 10952
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3684047726917077e-05,
+      "loss": 1.0468,
+      "step": 10953
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3682698787425934e-05,
+      "loss": 1.0522,
+      "step": 10954
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3681349816422796e-05,
+      "loss": 1.0718,
+      "step": 10955
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3680000813932932e-05,
+      "loss": 1.0276,
+      "step": 10956
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3678651779981598e-05,
+      "loss": 1.09,
+      "step": 10957
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3677302714594058e-05,
+      "loss": 1.0456,
+      "step": 10958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3675953617795583e-05,
+      "loss": 1.0426,
+      "step": 10959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.367460448961143e-05,
+      "loss": 1.101,
+      "step": 10960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3673255330066873e-05,
+      "loss": 1.1047,
+      "step": 10961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3671906139187168e-05,
+      "loss": 1.0622,
+      "step": 10962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3670556916997583e-05,
+      "loss": 1.0216,
+      "step": 10963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.366920766352339e-05,
+      "loss": 1.0395,
+      "step": 10964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3667858378789853e-05,
+      "loss": 1.058,
+      "step": 10965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3666509062822243e-05,
+      "loss": 1.0733,
+      "step": 10966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3665159715645825e-05,
+      "loss": 1.0099,
+      "step": 10967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3663810337285874e-05,
+      "loss": 0.9528,
+      "step": 10968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3662460927767651e-05,
+      "loss": 1.0248,
+      "step": 10969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3661111487116436e-05,
+      "loss": 1.0683,
+      "step": 10970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3659762015357492e-05,
+      "loss": 1.0872,
+      "step": 10971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3658412512516095e-05,
+      "loss": 1.0306,
+      "step": 10972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3657062978617514e-05,
+      "loss": 1.0587,
+      "step": 10973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3655713413687027e-05,
+      "loss": 1.0509,
+      "step": 10974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3654363817749905e-05,
+      "loss": 1.0802,
+      "step": 10975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3653014190831422e-05,
+      "loss": 1.0493,
+      "step": 10976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3651664532956852e-05,
+      "loss": 1.0057,
+      "step": 10977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3650314844151467e-05,
+      "loss": 1.0515,
+      "step": 10978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3648965124440548e-05,
+      "loss": 1.0153,
+      "step": 10979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.364761537384937e-05,
+      "loss": 1.0018,
+      "step": 10980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3646265592403208e-05,
+      "loss": 0.9838,
+      "step": 10981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3644915780127336e-05,
+      "loss": 1.0013,
+      "step": 10982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3643565937047042e-05,
+      "loss": 1.0213,
+      "step": 10983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3642216063187596e-05,
+      "loss": 0.9589,
+      "step": 10984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3640866158574284e-05,
+      "loss": 1.0583,
+      "step": 10985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3639516223232378e-05,
+      "loss": 0.9717,
+      "step": 10986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3638166257187166e-05,
+      "loss": 1.0598,
+      "step": 10987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3636816260463922e-05,
+      "loss": 1.003,
+      "step": 10988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.363546623308793e-05,
+      "loss": 0.9886,
+      "step": 10989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3634116175084478e-05,
+      "loss": 1.0925,
+      "step": 10990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3632766086478836e-05,
+      "loss": 1.0432,
+      "step": 10991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3631415967296297e-05,
+      "loss": 1.0641,
+      "step": 10992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3630065817562143e-05,
+      "loss": 1.001,
+      "step": 10993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3628715637301657e-05,
+      "loss": 1.0466,
+      "step": 10994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3627365426540126e-05,
+      "loss": 0.9769,
+      "step": 10995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3626015185302832e-05,
+      "loss": 1.049,
+      "step": 10996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3624664913615065e-05,
+      "loss": 1.0594,
+      "step": 10997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3623314611502104e-05,
+      "loss": 1.0246,
+      "step": 10998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3621964278989245e-05,
+      "loss": 1.0419,
+      "step": 10999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3620613916101778e-05,
+      "loss": 1.0425,
+      "step": 11000
+    },
+    {
+      "epoch": 0.46,
+      "eval_text_loss": 0.9642411470413208,
+      "eval_text_runtime": 1.5671,
+      "eval_text_samples_per_second": 47.22,
+      "eval_text_steps_per_second": 1.276,
+      "step": 11000
+    },
+    {
+      "epoch": 0.46,
+      "eval_svit_loss": 1.1194968223571777,
+      "eval_svit_runtime": 3.7088,
+      "eval_svit_samples_per_second": 183.615,
+      "eval_svit_steps_per_second": 2.966,
+      "step": 11000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3619263522864979e-05,
+      "loss": 0.9591,
+      "step": 11001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3617913099304143e-05,
+      "loss": 1.0279,
+      "step": 11002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3616562645444561e-05,
+      "loss": 1.0194,
+      "step": 11003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3615212161311525e-05,
+      "loss": 1.0431,
+      "step": 11004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.3613861646930321e-05,
+      "loss": 1.029,
+      "step": 11005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3612511102326244e-05,
+      "loss": 1.0082,
+      "step": 11006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3611160527524581e-05,
+      "loss": 1.0214,
+      "step": 11007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3609809922550625e-05,
+      "loss": 1.0276,
+      "step": 11008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3608459287429677e-05,
+      "loss": 1.0182,
+      "step": 11009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3607108622187017e-05,
+      "loss": 1.0869,
+      "step": 11010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3605757926847949e-05,
+      "loss": 1.0366,
+      "step": 11011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3604407201437766e-05,
+      "loss": 0.9665,
+      "step": 11012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360305644598176e-05,
+      "loss": 0.9845,
+      "step": 11013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.360170566050523e-05,
+      "loss": 1.0608,
+      "step": 11014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3600354845033472e-05,
+      "loss": 1.0128,
+      "step": 11015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3599003999591778e-05,
+      "loss": 1.0132,
+      "step": 11016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.359765312420545e-05,
+      "loss": 1.0568,
+      "step": 11017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3596302218899784e-05,
+      "loss": 1.0276,
+      "step": 11018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3594951283700081e-05,
+      "loss": 1.0061,
+      "step": 11019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3593600318631636e-05,
+      "loss": 0.9887,
+      "step": 11020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3592249323719752e-05,
+      "loss": 1.0005,
+      "step": 11021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3590898298989727e-05,
+      "loss": 0.9632,
+      "step": 11022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3589547244466868e-05,
+      "loss": 1.0379,
+      "step": 11023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3588196160176465e-05,
+      "loss": 1.0196,
+      "step": 11024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3586845046143829e-05,
+      "loss": 1.0191,
+      "step": 11025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3585493902394258e-05,
+      "loss": 1.0223,
+      "step": 11026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3584142728953055e-05,
+      "loss": 0.9772,
+      "step": 11027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3582791525845525e-05,
+      "loss": 0.9783,
+      "step": 11028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3581440293096974e-05,
+      "loss": 1.0172,
+      "step": 11029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3580089030732702e-05,
+      "loss": 0.9822,
+      "step": 11030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3578737738778015e-05,
+      "loss": 1.0812,
+      "step": 11031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3577386417258225e-05,
+      "loss": 0.9972,
+      "step": 11032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3576035066198631e-05,
+      "loss": 1.1092,
+      "step": 11033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3574683685624543e-05,
+      "loss": 0.9788,
+      "step": 11034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3573332275561266e-05,
+      "loss": 0.9777,
+      "step": 11035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3571980836034111e-05,
+      "loss": 0.9929,
+      "step": 11036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3570629367068383e-05,
+      "loss": 1.0099,
+      "step": 11037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3569277868689394e-05,
+      "loss": 1.0995,
+      "step": 11038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3567926340922454e-05,
+      "loss": 1.0269,
+      "step": 11039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3566574783792871e-05,
+      "loss": 1.025,
+      "step": 11040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3565223197325959e-05,
+      "loss": 1.0353,
+      "step": 11041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3563871581547024e-05,
+      "loss": 1.0384,
+      "step": 11042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3562519936481381e-05,
+      "loss": 1.0352,
+      "step": 11043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3561168262154343e-05,
+      "loss": 1.0008,
+      "step": 11044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.355981655859122e-05,
+      "loss": 1.0322,
+      "step": 11045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3558464825817327e-05,
+      "loss": 0.9967,
+      "step": 11046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3557113063857978e-05,
+      "loss": 1.0265,
+      "step": 11047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.355576127273849e-05,
+      "loss": 1.0127,
+      "step": 11048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3554409452484174e-05,
+      "loss": 1.0525,
+      "step": 11049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3553057603120347e-05,
+      "loss": 1.1027,
+      "step": 11050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3551705724672326e-05,
+      "loss": 1.0377,
+      "step": 11051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3550353817165428e-05,
+      "loss": 1.0444,
+      "step": 11052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3549001880624973e-05,
+      "loss": 1.0351,
+      "step": 11053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3547649915076269e-05,
+      "loss": 1.0284,
+      "step": 11054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3546297920544642e-05,
+      "loss": 1.0507,
+      "step": 11055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.354494589705541e-05,
+      "loss": 1.0297,
+      "step": 11056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.354359384463389e-05,
+      "loss": 1.0354,
+      "step": 11057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3542241763305408e-05,
+      "loss": 1.0213,
+      "step": 11058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3540889653095277e-05,
+      "loss": 1.0408,
+      "step": 11059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3539537514028821e-05,
+      "loss": 1.1172,
+      "step": 11060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3538185346131367e-05,
+      "loss": 1.0699,
+      "step": 11061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3536833149428226e-05,
+      "loss": 1.1059,
+      "step": 11062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3535480923944733e-05,
+      "loss": 1.0585,
+      "step": 11063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.35341286697062e-05,
+      "loss": 1.0583,
+      "step": 11064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3532776386737955e-05,
+      "loss": 1.0177,
+      "step": 11065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3531424075065325e-05,
+      "loss": 1.0127,
+      "step": 11066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3530071734713632e-05,
+      "loss": 1.0282,
+      "step": 11067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3528719365708203e-05,
+      "loss": 0.959,
+      "step": 11068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3527366968074363e-05,
+      "loss": 0.963,
+      "step": 11069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3526014541837438e-05,
+      "loss": 1.0106,
+      "step": 11070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3524662087022754e-05,
+      "loss": 1.0051,
+      "step": 11071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.352330960365564e-05,
+      "loss": 1.0333,
+      "step": 11072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3521957091761428e-05,
+      "loss": 1.0103,
+      "step": 11073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.352060455136544e-05,
+      "loss": 1.1105,
+      "step": 11074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3519251982493006e-05,
+      "loss": 1.042,
+      "step": 11075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.351789938516946e-05,
+      "loss": 1.048,
+      "step": 11076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3516546759420127e-05,
+      "loss": 1.0767,
+      "step": 11077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3515194105270341e-05,
+      "loss": 1.0241,
+      "step": 11078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3513841422745431e-05,
+      "loss": 0.9676,
+      "step": 11079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3512488711870733e-05,
+      "loss": 0.9682,
+      "step": 11080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3511135972671573e-05,
+      "loss": 1.0718,
+      "step": 11081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.350978320517329e-05,
+      "loss": 1.0162,
+      "step": 11082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3508430409401215e-05,
+      "loss": 1.0302,
+      "step": 11083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3507077585380677e-05,
+      "loss": 0.9855,
+      "step": 11084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3505724733137017e-05,
+      "loss": 1.054,
+      "step": 11085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3504371852695567e-05,
+      "loss": 1.0777,
+      "step": 11086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3503018944081668e-05,
+      "loss": 1.0847,
+      "step": 11087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3501666007320647e-05,
+      "loss": 1.093,
+      "step": 11088
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3500313042437847e-05,
+      "loss": 0.9876,
+      "step": 11089
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3498960049458597e-05,
+      "loss": 0.9797,
+      "step": 11090
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3497607028408245e-05,
+      "loss": 1.0686,
+      "step": 11091
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3496253979312126e-05,
+      "loss": 1.0107,
+      "step": 11092
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3494900902195573e-05,
+      "loss": 1.0231,
+      "step": 11093
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3493547797083932e-05,
+      "loss": 1.0765,
+      "step": 11094
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3492194664002537e-05,
+      "loss": 1.0074,
+      "step": 11095
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3490841502976733e-05,
+      "loss": 0.9953,
+      "step": 11096
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3489488314031862e-05,
+      "loss": 0.9916,
+      "step": 11097
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3488135097193258e-05,
+      "loss": 1.0315,
+      "step": 11098
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3486781852486268e-05,
+      "loss": 1.1179,
+      "step": 11099
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3485428579936233e-05,
+      "loss": 1.0141,
+      "step": 11100
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3484075279568496e-05,
+      "loss": 0.9948,
+      "step": 11101
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3482721951408399e-05,
+      "loss": 1.0285,
+      "step": 11102
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3481368595481287e-05,
+      "loss": 1.0785,
+      "step": 11103
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3480015211812507e-05,
+      "loss": 1.0541,
+      "step": 11104
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3478661800427399e-05,
+      "loss": 0.9805,
+      "step": 11105
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.347730836135131e-05,
+      "loss": 0.9514,
+      "step": 11106
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.347595489460959e-05,
+      "loss": 1.0315,
+      "step": 11107
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3474601400227583e-05,
+      "loss": 0.9716,
+      "step": 11108
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3473247878230631e-05,
+      "loss": 1.0193,
+      "step": 11109
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3471894328644087e-05,
+      "loss": 0.9757,
+      "step": 11110
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.34705407514933e-05,
+      "loss": 1.0814,
+      "step": 11111
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3469187146803613e-05,
+      "loss": 1.029,
+      "step": 11112
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3467833514600376e-05,
+      "loss": 0.992,
+      "step": 11113
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3466479854908944e-05,
+      "loss": 1.028,
+      "step": 11114
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3465126167754662e-05,
+      "loss": 1.0275,
+      "step": 11115
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3463772453162886e-05,
+      "loss": 1.0256,
+      "step": 11116
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.346241871115896e-05,
+      "loss": 1.0152,
+      "step": 11117
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.346106494176824e-05,
+      "loss": 1.0318,
+      "step": 11118
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3459711145016077e-05,
+      "loss": 1.0838,
+      "step": 11119
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3458357320927825e-05,
+      "loss": 1.0644,
+      "step": 11120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3457003469528835e-05,
+      "loss": 1.0677,
+      "step": 11121
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3455649590844459e-05,
+      "loss": 0.9912,
+      "step": 11122
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3454295684900056e-05,
+      "loss": 1.1102,
+      "step": 11123
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3452941751720978e-05,
+      "loss": 1.0298,
+      "step": 11124
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3451587791332579e-05,
+      "loss": 1.0185,
+      "step": 11125
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.345023380376022e-05,
+      "loss": 1.0058,
+      "step": 11126
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3448879789029253e-05,
+      "loss": 0.9895,
+      "step": 11127
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3447525747165033e-05,
+      "loss": 1.0924,
+      "step": 11128
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3446171678192919e-05,
+      "loss": 1.0369,
+      "step": 11129
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3444817582138267e-05,
+      "loss": 0.992,
+      "step": 11130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3443463459026442e-05,
+      "loss": 1.061,
+      "step": 11131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3442109308882796e-05,
+      "loss": 0.9888,
+      "step": 11132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.344075513173269e-05,
+      "loss": 0.9881,
+      "step": 11133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3439400927601482e-05,
+      "loss": 1.0278,
+      "step": 11134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3438046696514539e-05,
+      "loss": 1.0797,
+      "step": 11135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3436692438497215e-05,
+      "loss": 1.0777,
+      "step": 11136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3435338153574873e-05,
+      "loss": 1.0413,
+      "step": 11137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3433983841772873e-05,
+      "loss": 1.0279,
+      "step": 11138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3432629503116581e-05,
+      "loss": 0.9889,
+      "step": 11139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.343127513763136e-05,
+      "loss": 1.0187,
+      "step": 11140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3429920745342567e-05,
+      "loss": 1.0286,
+      "step": 11141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3428566326275574e-05,
+      "loss": 1.046,
+      "step": 11142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.342721188045574e-05,
+      "loss": 1.1122,
+      "step": 11143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3425857407908432e-05,
+      "loss": 1.0542,
+      "step": 11144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3424502908659014e-05,
+      "loss": 1.0061,
+      "step": 11145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3423148382732854e-05,
+      "loss": 0.9816,
+      "step": 11146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3421793830155314e-05,
+      "loss": 1.0731,
+      "step": 11147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3420439250951764e-05,
+      "loss": 1.0034,
+      "step": 11148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3419084645147568e-05,
+      "loss": 1.0211,
+      "step": 11149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3417730012768103e-05,
+      "loss": 0.9907,
+      "step": 11150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3416375353838726e-05,
+      "loss": 1.1046,
+      "step": 11151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.341502066838481e-05,
+      "loss": 1.0157,
+      "step": 11152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3413665956431726e-05,
+      "loss": 1.008,
+      "step": 11153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3412311218004843e-05,
+      "loss": 1.0545,
+      "step": 11154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.341095645312953e-05,
+      "loss": 0.9835,
+      "step": 11155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3409601661831159e-05,
+      "loss": 1.0574,
+      "step": 11156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3408246844135099e-05,
+      "loss": 1.007,
+      "step": 11157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3406892000066726e-05,
+      "loss": 1.0928,
+      "step": 11158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3405537129651407e-05,
+      "loss": 1.0178,
+      "step": 11159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3404182232914521e-05,
+      "loss": 1.0286,
+      "step": 11160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3402827309881434e-05,
+      "loss": 1.0776,
+      "step": 11161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3401472360577525e-05,
+      "loss": 1.0467,
+      "step": 11162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3400117385028168e-05,
+      "loss": 1.0413,
+      "step": 11163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3398762383258736e-05,
+      "loss": 1.0445,
+      "step": 11164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3397407355294604e-05,
+      "loss": 1.0069,
+      "step": 11165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3396052301161149e-05,
+      "loss": 0.9604,
+      "step": 11166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3394697220883744e-05,
+      "loss": 1.0582,
+      "step": 11167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3393342114487768e-05,
+      "loss": 1.0594,
+      "step": 11168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33919869819986e-05,
+      "loss": 0.9598,
+      "step": 11169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3390631823441616e-05,
+      "loss": 1.1126,
+      "step": 11170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3389276638842193e-05,
+      "loss": 1.0533,
+      "step": 11171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3387921428225708e-05,
+      "loss": 1.0594,
+      "step": 11172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3386566191617547e-05,
+      "loss": 1.0407,
+      "step": 11173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3385210929043081e-05,
+      "loss": 1.0366,
+      "step": 11174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3383855640527701e-05,
+      "loss": 1.0046,
+      "step": 11175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3382500326096774e-05,
+      "loss": 1.0723,
+      "step": 11176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.338114498577569e-05,
+      "loss": 1.0431,
+      "step": 11177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.337978961958983e-05,
+      "loss": 1.0337,
+      "step": 11178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3378434227564577e-05,
+      "loss": 1.0001,
+      "step": 11179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3377078809725306e-05,
+      "loss": 1.0053,
+      "step": 11180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3375723366097405e-05,
+      "loss": 1.003,
+      "step": 11181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.337436789670626e-05,
+      "loss": 0.9988,
+      "step": 11182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3373012401577251e-05,
+      "loss": 1.0417,
+      "step": 11183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3371656880735766e-05,
+      "loss": 1.0091,
+      "step": 11184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3370301334207186e-05,
+      "loss": 1.0477,
+      "step": 11185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3368945762016896e-05,
+      "loss": 0.9711,
+      "step": 11186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3367590164190288e-05,
+      "loss": 0.9563,
+      "step": 11187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.336623454075274e-05,
+      "loss": 1.0117,
+      "step": 11188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3364878891729647e-05,
+      "loss": 1.0111,
+      "step": 11189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.336352321714639e-05,
+      "loss": 1.0052,
+      "step": 11190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3362167517028359e-05,
+      "loss": 0.965,
+      "step": 11191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3360811791400947e-05,
+      "loss": 0.9928,
+      "step": 11192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3359456040289533e-05,
+      "loss": 0.9971,
+      "step": 11193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3358100263719515e-05,
+      "loss": 1.0103,
+      "step": 11194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.335674446171628e-05,
+      "loss": 0.9925,
+      "step": 11195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3355388634305213e-05,
+      "loss": 0.9788,
+      "step": 11196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3354032781511712e-05,
+      "loss": 1.001,
+      "step": 11197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3352676903361163e-05,
+      "loss": 1.0256,
+      "step": 11198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3351320999878962e-05,
+      "loss": 0.9798,
+      "step": 11199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33499650710905e-05,
+      "loss": 0.972,
+      "step": 11200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3348609117021167e-05,
+      "loss": 1.0138,
+      "step": 11201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3347253137696356e-05,
+      "loss": 1.0214,
+      "step": 11202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3345897133141464e-05,
+      "loss": 1.0636,
+      "step": 11203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3344541103381882e-05,
+      "loss": 1.0605,
+      "step": 11204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3343185048443005e-05,
+      "loss": 1.0259,
+      "step": 11205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.334182896835023e-05,
+      "loss": 0.9791,
+      "step": 11206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3340472863128947e-05,
+      "loss": 0.9937,
+      "step": 11207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3339116732804557e-05,
+      "loss": 1.0713,
+      "step": 11208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3337760577402458e-05,
+      "loss": 1.0318,
+      "step": 11209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3336404396948041e-05,
+      "loss": 1.0494,
+      "step": 11210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3335048191466709e-05,
+      "loss": 1.0066,
+      "step": 11211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3333691960983852e-05,
+      "loss": 1.0126,
+      "step": 11212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3332335705524877e-05,
+      "loss": 1.0251,
+      "step": 11213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3330979425115176e-05,
+      "loss": 1.0361,
+      "step": 11214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.332962311978015e-05,
+      "loss": 0.9608,
+      "step": 11215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3328266789545202e-05,
+      "loss": 1.0518,
+      "step": 11216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3326910434435726e-05,
+      "loss": 1.0915,
+      "step": 11217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3325554054477128e-05,
+      "loss": 1.0489,
+      "step": 11218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3324197649694808e-05,
+      "loss": 1.0002,
+      "step": 11219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3322841220114164e-05,
+      "loss": 1.0071,
+      "step": 11220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.33214847657606e-05,
+      "loss": 1.0509,
+      "step": 11221
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3320128286659519e-05,
+      "loss": 1.0029,
+      "step": 11222
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3318771782836326e-05,
+      "loss": 1.0185,
+      "step": 11223
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3317415254316421e-05,
+      "loss": 1.0513,
+      "step": 11224
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3316058701125208e-05,
+      "loss": 1.003,
+      "step": 11225
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.331470212328809e-05,
+      "loss": 1.0738,
+      "step": 11226
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3313345520830474e-05,
+      "loss": 1.0266,
+      "step": 11227
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3311988893777768e-05,
+      "loss": 1.0447,
+      "step": 11228
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.331063224215537e-05,
+      "loss": 1.1018,
+      "step": 11229
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3309275565988696e-05,
+      "loss": 0.979,
+      "step": 11230
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3307918865303142e-05,
+      "loss": 1.0435,
+      "step": 11231
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3306562140124122e-05,
+      "loss": 1.0084,
+      "step": 11232
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3305205390477043e-05,
+      "loss": 1.0339,
+      "step": 11233
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.330384861638731e-05,
+      "loss": 1.0232,
+      "step": 11234
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.330249181788033e-05,
+      "loss": 1.0094,
+      "step": 11235
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3301134994981517e-05,
+      "loss": 1.0061,
+      "step": 11236
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3299778147716277e-05,
+      "loss": 1.0076,
+      "step": 11237
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3298421276110022e-05,
+      "loss": 1.0219,
+      "step": 11238
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3297064380188162e-05,
+      "loss": 1.0659,
+      "step": 11239
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3295707459976105e-05,
+      "loss": 1.03,
+      "step": 11240
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3294350515499264e-05,
+      "loss": 1.0347,
+      "step": 11241
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.3292993546783051e-05,
+      "loss": 1.0694,
+      "step": 11242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3291636553852876e-05,
+      "loss": 1.0609,
+      "step": 11243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3290279536734155e-05,
+      "loss": 1.043,
+      "step": 11244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3288922495452294e-05,
+      "loss": 1.0306,
+      "step": 11245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3287565430032714e-05,
+      "loss": 1.0462,
+      "step": 11246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3286208340500826e-05,
+      "loss": 0.9956,
+      "step": 11247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3284851226882047e-05,
+      "loss": 0.9748,
+      "step": 11248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3283494089201784e-05,
+      "loss": 0.9769,
+      "step": 11249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3282136927485458e-05,
+      "loss": 1.0644,
+      "step": 11250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3280779741758484e-05,
+      "loss": 1.006,
+      "step": 11251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3279422532046282e-05,
+      "loss": 1.0626,
+      "step": 11252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.327806529837426e-05,
+      "loss": 1.0558,
+      "step": 11253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3276708040767841e-05,
+      "loss": 1.0568,
+      "step": 11254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.327535075925244e-05,
+      "loss": 1.0565,
+      "step": 11255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3273993453853478e-05,
+      "loss": 1.0176,
+      "step": 11256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3272636124596367e-05,
+      "loss": 1.0485,
+      "step": 11257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3271278771506532e-05,
+      "loss": 1.0716,
+      "step": 11258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3269921394609391e-05,
+      "loss": 0.9962,
+      "step": 11259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.326856399393036e-05,
+      "loss": 1.0762,
+      "step": 11260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3267206569494864e-05,
+      "loss": 1.006,
+      "step": 11261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3265849121328322e-05,
+      "loss": 0.9985,
+      "step": 11262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3264491649456155e-05,
+      "loss": 1.0785,
+      "step": 11263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.326313415390378e-05,
+      "loss": 1.0026,
+      "step": 11264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3261776634696626e-05,
+      "loss": 1.1142,
+      "step": 11265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3260419091860113e-05,
+      "loss": 0.9902,
+      "step": 11266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.325906152541966e-05,
+      "loss": 1.0021,
+      "step": 11267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3257703935400695e-05,
+      "loss": 1.0473,
+      "step": 11268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3256346321828636e-05,
+      "loss": 1.0597,
+      "step": 11269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3254988684728915e-05,
+      "loss": 1.0302,
+      "step": 11270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3253631024126949e-05,
+      "loss": 1.0153,
+      "step": 11271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3252273340048171e-05,
+      "loss": 1.0181,
+      "step": 11272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3250915632518e-05,
+      "loss": 1.0181,
+      "step": 11273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3249557901561862e-05,
+      "loss": 0.9797,
+      "step": 11274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3248200147205188e-05,
+      "loss": 1.0547,
+      "step": 11275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3246842369473398e-05,
+      "loss": 1.0143,
+      "step": 11276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3245484568391927e-05,
+      "loss": 1.0215,
+      "step": 11277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3244126743986194e-05,
+      "loss": 1.0172,
+      "step": 11278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3242768896281634e-05,
+      "loss": 1.0034,
+      "step": 11279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3241411025303671e-05,
+      "loss": 1.0237,
+      "step": 11280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3240053131077739e-05,
+      "loss": 1.0214,
+      "step": 11281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3238695213629264e-05,
+      "loss": 1.0238,
+      "step": 11282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3237337272983675e-05,
+      "loss": 1.0614,
+      "step": 11283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3235979309166402e-05,
+      "loss": 1.0014,
+      "step": 11284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.323462132220288e-05,
+      "loss": 0.9758,
+      "step": 11285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3233263312118539e-05,
+      "loss": 0.9434,
+      "step": 11286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3231905278938803e-05,
+      "loss": 1.0261,
+      "step": 11287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3230547222689112e-05,
+      "loss": 1.0741,
+      "step": 11288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3229189143394894e-05,
+      "loss": 0.9998,
+      "step": 11289
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3227831041081585e-05,
+      "loss": 1.063,
+      "step": 11290
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3226472915774619e-05,
+      "loss": 1.0436,
+      "step": 11291
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3225114767499426e-05,
+      "loss": 0.9871,
+      "step": 11292
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3223756596281442e-05,
+      "loss": 1.027,
+      "step": 11293
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3222398402146104e-05,
+      "loss": 1.0019,
+      "step": 11294
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3221040185118841e-05,
+      "loss": 0.9784,
+      "step": 11295
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3219681945225095e-05,
+      "loss": 1.0366,
+      "step": 11296
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3218323682490295e-05,
+      "loss": 1.0465,
+      "step": 11297
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3216965396939883e-05,
+      "loss": 1.0579,
+      "step": 11298
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.321560708859929e-05,
+      "loss": 1.028,
+      "step": 11299
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3214248757493961e-05,
+      "loss": 1.0646,
+      "step": 11300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.321289040364933e-05,
+      "loss": 0.9392,
+      "step": 11301
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3211532027090832e-05,
+      "loss": 1.0223,
+      "step": 11302
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3210173627843908e-05,
+      "loss": 1.0768,
+      "step": 11303
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3208815205933998e-05,
+      "loss": 0.9994,
+      "step": 11304
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3207456761386537e-05,
+      "loss": 1.0235,
+      "step": 11305
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.320609829422697e-05,
+      "loss": 1.0521,
+      "step": 11306
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3204739804480733e-05,
+      "loss": 1.0093,
+      "step": 11307
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3203381292173266e-05,
+      "loss": 1.0233,
+      "step": 11308
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3202022757330014e-05,
+      "loss": 1.0567,
+      "step": 11309
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3200664199976414e-05,
+      "loss": 1.0241,
+      "step": 11310
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3199305620137913e-05,
+      "loss": 1.081,
+      "step": 11311
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3197947017839948e-05,
+      "loss": 0.9986,
+      "step": 11312
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3196588393107968e-05,
+      "loss": 1.0116,
+      "step": 11313
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3195229745967407e-05,
+      "loss": 1.0014,
+      "step": 11314
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3193871076443715e-05,
+      "loss": 1.074,
+      "step": 11315
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3192512384562335e-05,
+      "loss": 0.9606,
+      "step": 11316
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3191153670348708e-05,
+      "loss": 1.0547,
+      "step": 11317
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3189794933828284e-05,
+      "loss": 0.9841,
+      "step": 11318
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3188436175026501e-05,
+      "loss": 0.9682,
+      "step": 11319
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3187077393968815e-05,
+      "loss": 1.0084,
+      "step": 11320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3185718590680663e-05,
+      "loss": 1.0943,
+      "step": 11321
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3184359765187492e-05,
+      "loss": 1.0661,
+      "step": 11322
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3183000917514754e-05,
+      "loss": 1.0297,
+      "step": 11323
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3181642047687893e-05,
+      "loss": 1.0094,
+      "step": 11324
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3180283155732359e-05,
+      "loss": 1.0166,
+      "step": 11325
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3178924241673593e-05,
+      "loss": 1.0035,
+      "step": 11326
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3177565305537049e-05,
+      "loss": 1.0557,
+      "step": 11327
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3176206347348177e-05,
+      "loss": 1.061,
+      "step": 11328
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3174847367132424e-05,
+      "loss": 1.0672,
+      "step": 11329
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3173488364915244e-05,
+      "loss": 1.0367,
+      "step": 11330
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3172129340722078e-05,
+      "loss": 1.0592,
+      "step": 11331
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3170770294578385e-05,
+      "loss": 0.9791,
+      "step": 11332
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3169411226509612e-05,
+      "loss": 1.0229,
+      "step": 11333
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3168052136541209e-05,
+      "loss": 1.0055,
+      "step": 11334
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3166693024698633e-05,
+      "loss": 1.0031,
+      "step": 11335
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.316533389100733e-05,
+      "loss": 1.0079,
+      "step": 11336
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3163974735492757e-05,
+      "loss": 1.0206,
+      "step": 11337
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3162615558180365e-05,
+      "loss": 1.0119,
+      "step": 11338
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3161256359095608e-05,
+      "loss": 0.9879,
+      "step": 11339
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3159897138263943e-05,
+      "loss": 0.9765,
+      "step": 11340
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.315853789571082e-05,
+      "loss": 1.0459,
+      "step": 11341
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3157178631461693e-05,
+      "loss": 1.0764,
+      "step": 11342
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3155819345542018e-05,
+      "loss": 1.0765,
+      "step": 11343
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3154460037977252e-05,
+      "loss": 1.022,
+      "step": 11344
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.315310070879285e-05,
+      "loss": 1.0609,
+      "step": 11345
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3151741358014268e-05,
+      "loss": 1.0613,
+      "step": 11346
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3150381985666962e-05,
+      "loss": 1.015,
+      "step": 11347
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3149022591776388e-05,
+      "loss": 1.0102,
+      "step": 11348
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3147663176368011e-05,
+      "loss": 1.0653,
+      "step": 11349
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.314630373946728e-05,
+      "loss": 1.0239,
+      "step": 11350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3144944281099656e-05,
+      "loss": 0.9561,
+      "step": 11351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3143584801290597e-05,
+      "loss": 1.045,
+      "step": 11352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3142225300065563e-05,
+      "loss": 1.0033,
+      "step": 11353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3140865777450016e-05,
+      "loss": 0.9821,
+      "step": 11354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.313950623346941e-05,
+      "loss": 1.0543,
+      "step": 11355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3138146668149209e-05,
+      "loss": 1.0825,
+      "step": 11356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3136787081514875e-05,
+      "loss": 0.9333,
+      "step": 11357
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3135427473591866e-05,
+      "loss": 1.0356,
+      "step": 11358
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3134067844405647e-05,
+      "loss": 1.0196,
+      "step": 11359
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3132708193981677e-05,
+      "loss": 1.0016,
+      "step": 11360
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3131348522345417e-05,
+      "loss": 0.9847,
+      "step": 11361
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3129988829522334e-05,
+      "loss": 1.0642,
+      "step": 11362
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3128629115537884e-05,
+      "loss": 1.0366,
+      "step": 11363
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3127269380417539e-05,
+      "loss": 1.0628,
+      "step": 11364
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3125909624186757e-05,
+      "loss": 0.9767,
+      "step": 11365
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3124549846871003e-05,
+      "loss": 1.0243,
+      "step": 11366
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3123190048495744e-05,
+      "loss": 1.0175,
+      "step": 11367
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3121830229086445e-05,
+      "loss": 0.9845,
+      "step": 11368
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.312047038866857e-05,
+      "loss": 1.0298,
+      "step": 11369
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3119110527267585e-05,
+      "loss": 0.9959,
+      "step": 11370
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3117750644908953e-05,
+      "loss": 1.0247,
+      "step": 11371
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3116390741618145e-05,
+      "loss": 1.0186,
+      "step": 11372
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3115030817420626e-05,
+      "loss": 1.0361,
+      "step": 11373
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3113670872341867e-05,
+      "loss": 0.9891,
+      "step": 11374
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.311231090640733e-05,
+      "loss": 0.9899,
+      "step": 11375
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3110950919642487e-05,
+      "loss": 1.0932,
+      "step": 11376
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3109590912072804e-05,
+      "loss": 0.9687,
+      "step": 11377
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3108230883723756e-05,
+      "loss": 0.9712,
+      "step": 11378
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3106870834620805e-05,
+      "loss": 1.0331,
+      "step": 11379
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3105510764789423e-05,
+      "loss": 0.9904,
+      "step": 11380
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.310415067425508e-05,
+      "loss": 1.0486,
+      "step": 11381
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3102790563043248e-05,
+      "loss": 1.085,
+      "step": 11382
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3101430431179398e-05,
+      "loss": 1.0337,
+      "step": 11383
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3100070278688996e-05,
+      "loss": 1.0419,
+      "step": 11384
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.309871010559752e-05,
+      "loss": 0.9796,
+      "step": 11385
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3097349911930441e-05,
+      "loss": 1.0921,
+      "step": 11386
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3095989697713232e-05,
+      "loss": 1.0786,
+      "step": 11387
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3094629462971362e-05,
+      "loss": 1.0375,
+      "step": 11388
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3093269207730308e-05,
+      "loss": 1.0179,
+      "step": 11389
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3091908932015537e-05,
+      "loss": 1.0625,
+      "step": 11390
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3090548635852532e-05,
+      "loss": 1.002,
+      "step": 11391
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.308918831926676e-05,
+      "loss": 1.0098,
+      "step": 11392
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3087827982283703e-05,
+      "loss": 1.0365,
+      "step": 11393
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.308646762492883e-05,
+      "loss": 0.9967,
+      "step": 11394
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3085107247227618e-05,
+      "loss": 1.0812,
+      "step": 11395
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3083746849205543e-05,
+      "loss": 1.0256,
+      "step": 11396
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3082386430888085e-05,
+      "loss": 1.0567,
+      "step": 11397
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3081025992300715e-05,
+      "loss": 0.9712,
+      "step": 11398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3079665533468913e-05,
+      "loss": 1.0586,
+      "step": 11399
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3078305054418152e-05,
+      "loss": 1.0055,
+      "step": 11400
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3076944555173915e-05,
+      "loss": 1.0422,
+      "step": 11401
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3075584035761679e-05,
+      "loss": 1.0078,
+      "step": 11402
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3074223496206926e-05,
+      "loss": 1.0788,
+      "step": 11403
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3072862936535127e-05,
+      "loss": 1.0597,
+      "step": 11404
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3071502356771764e-05,
+      "loss": 1.0327,
+      "step": 11405
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.307014175694232e-05,
+      "loss": 1.0694,
+      "step": 11406
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3068781137072273e-05,
+      "loss": 1.04,
+      "step": 11407
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3067420497187104e-05,
+      "loss": 0.9745,
+      "step": 11408
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3066059837312292e-05,
+      "loss": 0.9894,
+      "step": 11409
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.306469915747332e-05,
+      "loss": 1.0318,
+      "step": 11410
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3063338457695667e-05,
+      "loss": 1.0462,
+      "step": 11411
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3061977738004816e-05,
+      "loss": 1.0778,
+      "step": 11412
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3060616998426252e-05,
+      "loss": 1.1212,
+      "step": 11413
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3059256238985455e-05,
+      "loss": 1.03,
+      "step": 11414
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3057895459707908e-05,
+      "loss": 1.0746,
+      "step": 11415
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3056534660619097e-05,
+      "loss": 1.0842,
+      "step": 11416
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3055173841744503e-05,
+      "loss": 0.972,
+      "step": 11417
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3053813003109611e-05,
+      "loss": 0.9966,
+      "step": 11418
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3052452144739904e-05,
+      "loss": 1.0162,
+      "step": 11419
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.305109126666087e-05,
+      "loss": 1.0707,
+      "step": 11420
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304973036889799e-05,
+      "loss": 1.0109,
+      "step": 11421
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3048369451476754e-05,
+      "loss": 0.9897,
+      "step": 11422
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3047008514422646e-05,
+      "loss": 1.0012,
+      "step": 11423
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304564755776115e-05,
+      "loss": 1.0719,
+      "step": 11424
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304428658151776e-05,
+      "loss": 1.1075,
+      "step": 11425
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3042925585717954e-05,
+      "loss": 1.0414,
+      "step": 11426
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3041564570387227e-05,
+      "loss": 1.008,
+      "step": 11427
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.304020353555106e-05,
+      "loss": 1.0274,
+      "step": 11428
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3038842481234945e-05,
+      "loss": 0.9736,
+      "step": 11429
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3037481407464372e-05,
+      "loss": 1.0558,
+      "step": 11430
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3036120314264826e-05,
+      "loss": 1.0103,
+      "step": 11431
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.30347592016618e-05,
+      "loss": 0.995,
+      "step": 11432
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.303339806968078e-05,
+      "loss": 1.0145,
+      "step": 11433
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3032036918347262e-05,
+      "loss": 1.1231,
+      "step": 11434
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.303067574768673e-05,
+      "loss": 1.0163,
+      "step": 11435
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3029314557724674e-05,
+      "loss": 0.9952,
+      "step": 11436
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3027953348486593e-05,
+      "loss": 1.05,
+      "step": 11437
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3026592119997971e-05,
+      "loss": 0.9651,
+      "step": 11438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3025230872284303e-05,
+      "loss": 0.9947,
+      "step": 11439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3023869605371077e-05,
+      "loss": 1.0646,
+      "step": 11440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3022508319283792e-05,
+      "loss": 1.0302,
+      "step": 11441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3021147014047941e-05,
+      "loss": 1.0459,
+      "step": 11442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.301978568968901e-05,
+      "loss": 1.0876,
+      "step": 11443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3018424346232501e-05,
+      "loss": 0.994,
+      "step": 11444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3017062983703899e-05,
+      "loss": 1.0661,
+      "step": 11445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3015701602128703e-05,
+      "loss": 1.0115,
+      "step": 11446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.301434020153241e-05,
+      "loss": 1.0222,
+      "step": 11447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3012978781940512e-05,
+      "loss": 1.0073,
+      "step": 11448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3011617343378506e-05,
+      "loss": 1.0335,
+      "step": 11449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3010255885871885e-05,
+      "loss": 1.045,
+      "step": 11450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3008894409446149e-05,
+      "loss": 1.0281,
+      "step": 11451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.300753291412679e-05,
+      "loss": 1.0555,
+      "step": 11452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3006171399939308e-05,
+      "loss": 1.0072,
+      "step": 11453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.30048098669092e-05,
+      "loss": 1.1316,
+      "step": 11454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3003448315061961e-05,
+      "loss": 0.984,
+      "step": 11455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3002086744423095e-05,
+      "loss": 1.072,
+      "step": 11456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.3000725155018092e-05,
+      "loss": 1.0386,
+      "step": 11457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2999363546872453e-05,
+      "loss": 1.0239,
+      "step": 11458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.299800192001168e-05,
+      "loss": 0.9802,
+      "step": 11459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2996640274461272e-05,
+      "loss": 1.0234,
+      "step": 11460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2995278610246725e-05,
+      "loss": 1.0344,
+      "step": 11461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2993916927393545e-05,
+      "loss": 1.0143,
+      "step": 11462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2992555225927226e-05,
+      "loss": 1.0355,
+      "step": 11463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.299119350587327e-05,
+      "loss": 1.0701,
+      "step": 11464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2989831767257179e-05,
+      "loss": 1.0567,
+      "step": 11465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2988470010104457e-05,
+      "loss": 0.9929,
+      "step": 11466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2987108234440602e-05,
+      "loss": 1.0007,
+      "step": 11467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2985746440291115e-05,
+      "loss": 0.9519,
+      "step": 11468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2984384627681502e-05,
+      "loss": 0.9662,
+      "step": 11469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2983022796637263e-05,
+      "loss": 0.9907,
+      "step": 11470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2981660947183907e-05,
+      "loss": 1.0142,
+      "step": 11471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2980299079346932e-05,
+      "loss": 1.0394,
+      "step": 11472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2978937193151838e-05,
+      "loss": 1.0567,
+      "step": 11473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2977575288624138e-05,
+      "loss": 1.1246,
+      "step": 11474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.297621336578933e-05,
+      "loss": 1.0653,
+      "step": 11475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2974851424672922e-05,
+      "loss": 1.0348,
+      "step": 11476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2973489465300418e-05,
+      "loss": 1.0082,
+      "step": 11477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2972127487697323e-05,
+      "loss": 1.018,
+      "step": 11478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.2970765491889143e-05,
+      "loss": 1.0575,
+      "step": 11479
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2969403477901388e-05,
+      "loss": 1.0132,
+      "step": 11480
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2968041445759558e-05,
+      "loss": 0.9955,
+      "step": 11481
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2966679395489165e-05,
+      "loss": 1.0392,
+      "step": 11482
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2965317327115712e-05,
+      "loss": 1.0469,
+      "step": 11483
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.296395524066471e-05,
+      "loss": 1.0578,
+      "step": 11484
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2962593136161667e-05,
+      "loss": 1.0503,
+      "step": 11485
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2961231013632088e-05,
+      "loss": 1.0318,
+      "step": 11486
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2959868873101482e-05,
+      "loss": 0.9484,
+      "step": 11487
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295850671459536e-05,
+      "loss": 1.046,
+      "step": 11488
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295714453813923e-05,
+      "loss": 1.0275,
+      "step": 11489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2955782343758604e-05,
+      "loss": 1.0149,
+      "step": 11490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295442013147899e-05,
+      "loss": 0.9874,
+      "step": 11491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2953057901325898e-05,
+      "loss": 1.0048,
+      "step": 11492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2951695653324837e-05,
+      "loss": 1.048,
+      "step": 11493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.295033338750132e-05,
+      "loss": 1.0156,
+      "step": 11494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2948971103880859e-05,
+      "loss": 1.0634,
+      "step": 11495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2947608802488962e-05,
+      "loss": 1.0344,
+      "step": 11496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2946246483351146e-05,
+      "loss": 1.0312,
+      "step": 11497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2944884146492917e-05,
+      "loss": 1.0314,
+      "step": 11498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2943521791939795e-05,
+      "loss": 1.0017,
+      "step": 11499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2942159419717286e-05,
+      "loss": 1.0581,
+      "step": 11500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2940797029850908e-05,
+      "loss": 0.9753,
+      "step": 11501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2939434622366174e-05,
+      "loss": 0.9998,
+      "step": 11502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.293807219728859e-05,
+      "loss": 1.0509,
+      "step": 11503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2936709754643683e-05,
+      "loss": 0.9718,
+      "step": 11504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2935347294456962e-05,
+      "loss": 1.0105,
+      "step": 11505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2933984816753937e-05,
+      "loss": 1.054,
+      "step": 11506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2932622321560128e-05,
+      "loss": 1.0061,
+      "step": 11507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2931259808901051e-05,
+      "loss": 1.0414,
+      "step": 11508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.292989727880222e-05,
+      "loss": 0.9684,
+      "step": 11509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2928534731289156e-05,
+      "loss": 1.0184,
+      "step": 11510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2927172166387367e-05,
+      "loss": 1.0098,
+      "step": 11511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2925809584122374e-05,
+      "loss": 0.9911,
+      "step": 11512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2924446984519697e-05,
+      "loss": 1.0384,
+      "step": 11513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2923084367604848e-05,
+      "loss": 1.0451,
+      "step": 11514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.292172173340335e-05,
+      "loss": 1.0043,
+      "step": 11515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2920359081940716e-05,
+      "loss": 0.9824,
+      "step": 11516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2918996413242467e-05,
+      "loss": 0.9995,
+      "step": 11517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2917633727334129e-05,
+      "loss": 1.0384,
+      "step": 11518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2916271024241206e-05,
+      "loss": 0.9873,
+      "step": 11519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2914908303989233e-05,
+      "loss": 0.9676,
+      "step": 11520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2913545566603716e-05,
+      "loss": 1.078,
+      "step": 11521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2912182812110183e-05,
+      "loss": 1.0779,
+      "step": 11522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2910820040534153e-05,
+      "loss": 1.0153,
+      "step": 11523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2909457251901147e-05,
+      "loss": 1.0138,
+      "step": 11524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2908094446236685e-05,
+      "loss": 1.06,
+      "step": 11525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2906731623566287e-05,
+      "loss": 1.0219,
+      "step": 11526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.290536878391548e-05,
+      "loss": 1.0333,
+      "step": 11527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2904005927309779e-05,
+      "loss": 0.9753,
+      "step": 11528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2902643053774712e-05,
+      "loss": 1.0037,
+      "step": 11529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.29012801633358e-05,
+      "loss": 1.0331,
+      "step": 11530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2899917256018561e-05,
+      "loss": 1.024,
+      "step": 11531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2898554331848525e-05,
+      "loss": 1.0594,
+      "step": 11532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2897191390851212e-05,
+      "loss": 1.0286,
+      "step": 11533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2895828433052148e-05,
+      "loss": 0.9977,
+      "step": 11534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2894465458476855e-05,
+      "loss": 1.0514,
+      "step": 11535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.289310246715086e-05,
+      "loss": 1.0164,
+      "step": 11536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2891739459099686e-05,
+      "loss": 1.0238,
+      "step": 11537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2890376434348857e-05,
+      "loss": 1.0506,
+      "step": 11538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2889013392923904e-05,
+      "loss": 1.1161,
+      "step": 11539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2887650334850344e-05,
+      "loss": 1.029,
+      "step": 11540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2886287260153708e-05,
+      "loss": 0.996,
+      "step": 11541
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2884924168859523e-05,
+      "loss": 1.0212,
+      "step": 11542
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2883561060993314e-05,
+      "loss": 1.0435,
+      "step": 11543
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2882197936580613e-05,
+      "loss": 0.9961,
+      "step": 11544
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.288083479564694e-05,
+      "loss": 0.9671,
+      "step": 11545
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2879471638217826e-05,
+      "loss": 1.0408,
+      "step": 11546
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.28781084643188e-05,
+      "loss": 1.0785,
+      "step": 11547
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2876745273975387e-05,
+      "loss": 0.9905,
+      "step": 11548
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2875382067213118e-05,
+      "loss": 1.0676,
+      "step": 11549
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.287401884405752e-05,
+      "loss": 1.0519,
+      "step": 11550
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2872655604534123e-05,
+      "loss": 1.0647,
+      "step": 11551
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2871292348668458e-05,
+      "loss": 1.0746,
+      "step": 11552
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2869929076486053e-05,
+      "loss": 0.9855,
+      "step": 11553
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2868565788012442e-05,
+      "loss": 1.0173,
+      "step": 11554
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2867202483273149e-05,
+      "loss": 1.1324,
+      "step": 11555
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.286583916229371e-05,
+      "loss": 1.0661,
+      "step": 11556
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2864475825099651e-05,
+      "loss": 1.0998,
+      "step": 11557
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2863112471716507e-05,
+      "loss": 1.0149,
+      "step": 11558
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2861749102169809e-05,
+      "loss": 0.9976,
+      "step": 11559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2860385716485089e-05,
+      "loss": 1.0396,
+      "step": 11560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2859022314687877e-05,
+      "loss": 0.9835,
+      "step": 11561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2857658896803708e-05,
+      "loss": 0.9543,
+      "step": 11562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2856295462858115e-05,
+      "loss": 1.0255,
+      "step": 11563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.285493201287663e-05,
+      "loss": 0.9733,
+      "step": 11564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2853568546884787e-05,
+      "loss": 1.0454,
+      "step": 11565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2852205064908119e-05,
+      "loss": 1.0252,
+      "step": 11566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2850841566972158e-05,
+      "loss": 1.0037,
+      "step": 11567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2849478053102445e-05,
+      "loss": 0.9837,
+      "step": 11568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2848114523324504e-05,
+      "loss": 1.0045,
+      "step": 11569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.284675097766388e-05,
+      "loss": 1.0187,
+      "step": 11570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2845387416146101e-05,
+      "loss": 1.0065,
+      "step": 11571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2844023838796708e-05,
+      "loss": 1.0187,
+      "step": 11572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2842660245641236e-05,
+      "loss": 0.9801,
+      "step": 11573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2841296636705218e-05,
+      "loss": 1.0174,
+      "step": 11574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2839933012014194e-05,
+      "loss": 1.0874,
+      "step": 11575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2838569371593696e-05,
+      "loss": 0.983,
+      "step": 11576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2837205715469262e-05,
+      "loss": 0.9591,
+      "step": 11577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2835842043666437e-05,
+      "loss": 1.0028,
+      "step": 11578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2834478356210748e-05,
+      "loss": 1.0156,
+      "step": 11579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2833114653127738e-05,
+      "loss": 1.0669,
+      "step": 11580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2831750934442944e-05,
+      "loss": 1.0168,
+      "step": 11581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2830387200181903e-05,
+      "loss": 1.0338,
+      "step": 11582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2829023450370159e-05,
+      "loss": 1.0521,
+      "step": 11583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.282765968503325e-05,
+      "loss": 1.006,
+      "step": 11584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2826295904196708e-05,
+      "loss": 1.0287,
+      "step": 11585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2824932107886079e-05,
+      "loss": 1.0133,
+      "step": 11586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.28235682961269e-05,
+      "loss": 1.005,
+      "step": 11587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2822204468944714e-05,
+      "loss": 1.0486,
+      "step": 11588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2820840626365061e-05,
+      "loss": 0.9761,
+      "step": 11589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.281947676841348e-05,
+      "loss": 0.9981,
+      "step": 11590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.281811289511551e-05,
+      "loss": 1.0148,
+      "step": 11591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.28167490064967e-05,
+      "loss": 0.9843,
+      "step": 11592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2815385102582583e-05,
+      "loss": 1.0641,
+      "step": 11593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2814021183398708e-05,
+      "loss": 0.9639,
+      "step": 11594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.281265724897061e-05,
+      "loss": 1.0561,
+      "step": 11595
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2811293299323837e-05,
+      "loss": 1.0237,
+      "step": 11596
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2809929334483933e-05,
+      "loss": 1.0177,
+      "step": 11597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2808565354476435e-05,
+      "loss": 0.996,
+      "step": 11598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2807201359326888e-05,
+      "loss": 1.1553,
+      "step": 11599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2805837349060839e-05,
+      "loss": 1.0526,
+      "step": 11600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.280447332370383e-05,
+      "loss": 0.9744,
+      "step": 11601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2803109283281407e-05,
+      "loss": 1.0044,
+      "step": 11602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2801745227819111e-05,
+      "loss": 1.0734,
+      "step": 11603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2800381157342487e-05,
+      "loss": 1.0562,
+      "step": 11604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2799017071877084e-05,
+      "loss": 1.0633,
+      "step": 11605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2797652971448441e-05,
+      "loss": 0.9867,
+      "step": 11606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2796288856082112e-05,
+      "loss": 1.0036,
+      "step": 11607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2794924725803634e-05,
+      "loss": 1.0618,
+      "step": 11608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2793560580638559e-05,
+      "loss": 0.9582,
+      "step": 11609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.279219642061243e-05,
+      "loss": 1.0657,
+      "step": 11610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.27908322457508e-05,
+      "loss": 0.9341,
+      "step": 11611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.278946805607921e-05,
+      "loss": 0.9985,
+      "step": 11612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2788103851623205e-05,
+      "loss": 1.0247,
+      "step": 11613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2786739632408338e-05,
+      "loss": 1.0374,
+      "step": 11614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2785375398460156e-05,
+      "loss": 1.0408,
+      "step": 11615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2784011149804202e-05,
+      "loss": 1.0126,
+      "step": 11616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2782646886466035e-05,
+      "loss": 0.9841,
+      "step": 11617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2781282608471193e-05,
+      "loss": 1.0532,
+      "step": 11618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.277991831584523e-05,
+      "loss": 1.0688,
+      "step": 11619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2778554008613694e-05,
+      "loss": 0.9422,
+      "step": 11620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2777189686802134e-05,
+      "loss": 1.0027,
+      "step": 11621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2775825350436101e-05,
+      "loss": 1.0526,
+      "step": 11622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2774460999541145e-05,
+      "loss": 1.0115,
+      "step": 11623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2773096634142814e-05,
+      "loss": 0.9488,
+      "step": 11624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2771732254266659e-05,
+      "loss": 1.0588,
+      "step": 11625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2770367859938235e-05,
+      "loss": 1.0948,
+      "step": 11626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2769003451183088e-05,
+      "loss": 0.9913,
+      "step": 11627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2767639028026773e-05,
+      "loss": 1.0954,
+      "step": 11628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2766274590494839e-05,
+      "loss": 1.0414,
+      "step": 11629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2764910138612838e-05,
+      "loss": 1.0748,
+      "step": 11630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2763545672406325e-05,
+      "loss": 0.982,
+      "step": 11631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2762181191900851e-05,
+      "loss": 1.0433,
+      "step": 11632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2760816697121965e-05,
+      "loss": 1.0021,
+      "step": 11633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2759452188095225e-05,
+      "loss": 1.0083,
+      "step": 11634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2758087664846181e-05,
+      "loss": 1.0402,
+      "step": 11635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2756723127400392e-05,
+      "loss": 1.0865,
+      "step": 11636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2755358575783402e-05,
+      "loss": 1.0532,
+      "step": 11637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2753994010020776e-05,
+      "loss": 1.0153,
+      "step": 11638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2752629430138062e-05,
+      "loss": 1.046,
+      "step": 11639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2751264836160813e-05,
+      "loss": 0.9719,
+      "step": 11640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2749900228114588e-05,
+      "loss": 1.0444,
+      "step": 11641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.274853560602494e-05,
+      "loss": 1.0729,
+      "step": 11642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2747170969917426e-05,
+      "loss": 1.0572,
+      "step": 11643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.27458063198176e-05,
+      "loss": 1.0271,
+      "step": 11644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2744441655751017e-05,
+      "loss": 1.0231,
+      "step": 11645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2743076977743236e-05,
+      "loss": 1.0725,
+      "step": 11646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2741712285819811e-05,
+      "loss": 1.0269,
+      "step": 11647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2740347580006298e-05,
+      "loss": 1.0744,
+      "step": 11648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.273898286032826e-05,
+      "loss": 1.0221,
+      "step": 11649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2737618126811246e-05,
+      "loss": 1.0113,
+      "step": 11650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2736253379480818e-05,
+      "loss": 1.0668,
+      "step": 11651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2734888618362533e-05,
+      "loss": 1.0323,
+      "step": 11652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2733523843481947e-05,
+      "loss": 1.0614,
+      "step": 11653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.273215905486462e-05,
+      "loss": 1.0,
+      "step": 11654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2730794252536114e-05,
+      "loss": 1.0159,
+      "step": 11655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2729429436521983e-05,
+      "loss": 1.0979,
+      "step": 11656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2728064606847786e-05,
+      "loss": 1.0189,
+      "step": 11657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2726699763539085e-05,
+      "loss": 1.0457,
+      "step": 11658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2725334906621435e-05,
+      "loss": 1.122,
+      "step": 11659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.27239700361204e-05,
+      "loss": 1.0594,
+      "step": 11660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.272260515206154e-05,
+      "loss": 1.0166,
+      "step": 11661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2721240254470412e-05,
+      "loss": 1.0497,
+      "step": 11662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2719875343372578e-05,
+      "loss": 1.0292,
+      "step": 11663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2718510418793598e-05,
+      "loss": 0.989,
+      "step": 11664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2717145480759038e-05,
+      "loss": 1.0052,
+      "step": 11665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2715780529294454e-05,
+      "loss": 0.9788,
+      "step": 11666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2714415564425408e-05,
+      "loss": 0.9505,
+      "step": 11667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2713050586177464e-05,
+      "loss": 1.0445,
+      "step": 11668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2711685594576181e-05,
+      "loss": 1.0236,
+      "step": 11669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2710320589647127e-05,
+      "loss": 1.053,
+      "step": 11670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2708955571415856e-05,
+      "loss": 1.0396,
+      "step": 11671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2707590539907936e-05,
+      "loss": 1.0703,
+      "step": 11672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.270622549514893e-05,
+      "loss": 1.0262,
+      "step": 11673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.27048604371644e-05,
+      "loss": 0.9994,
+      "step": 11674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2703495365979913e-05,
+      "loss": 1.0192,
+      "step": 11675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2702130281621028e-05,
+      "loss": 1.0399,
+      "step": 11676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2700765184113312e-05,
+      "loss": 1.0088,
+      "step": 11677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2699400073482326e-05,
+      "loss": 0.9929,
+      "step": 11678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2698034949753636e-05,
+      "loss": 0.9602,
+      "step": 11679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2696669812952812e-05,
+      "loss": 1.0764,
+      "step": 11680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2695304663105412e-05,
+      "loss": 1.0696,
+      "step": 11681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2693939500237002e-05,
+      "loss": 1.0374,
+      "step": 11682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2692574324373151e-05,
+      "loss": 1.0911,
+      "step": 11683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.269120913553942e-05,
+      "loss": 1.016,
+      "step": 11684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.268984393376138e-05,
+      "loss": 1.0149,
+      "step": 11685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2688478719064594e-05,
+      "loss": 0.9832,
+      "step": 11686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2687113491474631e-05,
+      "loss": 1.0035,
+      "step": 11687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2685748251017056e-05,
+      "loss": 1.1144,
+      "step": 11688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2684382997717434e-05,
+      "loss": 1.0446,
+      "step": 11689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2683017731601339e-05,
+      "loss": 0.9556,
+      "step": 11690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2681652452694329e-05,
+      "loss": 1.0553,
+      "step": 11691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2680287161021977e-05,
+      "loss": 1.073,
+      "step": 11692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2678921856609851e-05,
+      "loss": 0.9923,
+      "step": 11693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2677556539483519e-05,
+      "loss": 1.0111,
+      "step": 11694
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2676191209668547e-05,
+      "loss": 1.0282,
+      "step": 11695
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2674825867190508e-05,
+      "loss": 1.0777,
+      "step": 11696
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2673460512074966e-05,
+      "loss": 1.0079,
+      "step": 11697
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.267209514434749e-05,
+      "loss": 1.02,
+      "step": 11698
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2670729764033657e-05,
+      "loss": 1.0282,
+      "step": 11699
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2669364371159027e-05,
+      "loss": 0.9957,
+      "step": 11700
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2667998965749174e-05,
+      "loss": 0.989,
+      "step": 11701
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.266663354782967e-05,
+      "loss": 1.0806,
+      "step": 11702
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2665268117426081e-05,
+      "loss": 1.0419,
+      "step": 11703
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2663902674563983e-05,
+      "loss": 1.0529,
+      "step": 11704
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2662537219268941e-05,
+      "loss": 1.0193,
+      "step": 11705
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.266117175156653e-05,
+      "loss": 1.0408,
+      "step": 11706
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2659806271482318e-05,
+      "loss": 0.987,
+      "step": 11707
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2658440779041875e-05,
+      "loss": 1.0274,
+      "step": 11708
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2657075274270782e-05,
+      "loss": 1.037,
+      "step": 11709
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2655709757194601e-05,
+      "loss": 1.0405,
+      "step": 11710
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2654344227838905e-05,
+      "loss": 1.0241,
+      "step": 11711
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.265297868622927e-05,
+      "loss": 1.0241,
+      "step": 11712
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2651613132391273e-05,
+      "loss": 0.9854,
+      "step": 11713
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.2650247566350477e-05,
+      "loss": 0.9971,
+      "step": 11714
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.264888198813246e-05,
+      "loss": 0.9888,
+      "step": 11715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2647516397762794e-05,
+      "loss": 1.1102,
+      "step": 11716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2646150795267053e-05,
+      "loss": 1.0269,
+      "step": 11717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2644785180670812e-05,
+      "loss": 1.0186,
+      "step": 11718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2643419553999643e-05,
+      "loss": 1.0072,
+      "step": 11719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.264205391527912e-05,
+      "loss": 1.0265,
+      "step": 11720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.264068826453482e-05,
+      "loss": 1.0291,
+      "step": 11721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2639322601792314e-05,
+      "loss": 0.9887,
+      "step": 11722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.263795692707718e-05,
+      "loss": 0.9405,
+      "step": 11723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2636591240414992e-05,
+      "loss": 1.0181,
+      "step": 11724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2635225541831326e-05,
+      "loss": 1.0268,
+      "step": 11725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2633859831351754e-05,
+      "loss": 1.0297,
+      "step": 11726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2632494109001856e-05,
+      "loss": 1.0044,
+      "step": 11727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2631128374807207e-05,
+      "loss": 1.0568,
+      "step": 11728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.262976262879338e-05,
+      "loss": 0.9624,
+      "step": 11729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2628396870985955e-05,
+      "loss": 0.9803,
+      "step": 11730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2627031101410507e-05,
+      "loss": 1.0537,
+      "step": 11731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2625665320092615e-05,
+      "loss": 0.9767,
+      "step": 11732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2624299527057853e-05,
+      "loss": 1.1208,
+      "step": 11733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2622933722331802e-05,
+      "loss": 0.9969,
+      "step": 11734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2621567905940036e-05,
+      "loss": 1.0516,
+      "step": 11735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2620202077908133e-05,
+      "loss": 1.0078,
+      "step": 11736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.261883623826167e-05,
+      "loss": 0.9781,
+      "step": 11737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2617470387026232e-05,
+      "loss": 0.964,
+      "step": 11738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2616104524227389e-05,
+      "loss": 1.0263,
+      "step": 11739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.261473864989072e-05,
+      "loss": 0.9852,
+      "step": 11740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2613372764041811e-05,
+      "loss": 1.024,
+      "step": 11741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2612006866706236e-05,
+      "loss": 1.0535,
+      "step": 11742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2610640957909577e-05,
+      "loss": 1.0234,
+      "step": 11743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.260927503767741e-05,
+      "loss": 1.0635,
+      "step": 11744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2607909106035314e-05,
+      "loss": 1.0293,
+      "step": 11745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2606543163008873e-05,
+      "loss": 1.0661,
+      "step": 11746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2605177208623662e-05,
+      "loss": 0.9861,
+      "step": 11747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2603811242905271e-05,
+      "loss": 1.0387,
+      "step": 11748
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2602445265879268e-05,
+      "loss": 0.9921,
+      "step": 11749
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.260107927757124e-05,
+      "loss": 0.9858,
+      "step": 11750
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.259971327800677e-05,
+      "loss": 1.0541,
+      "step": 11751
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2598347267211437e-05,
+      "loss": 1.0268,
+      "step": 11752
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2596981245210822e-05,
+      "loss": 0.9925,
+      "step": 11753
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2595615212030505e-05,
+      "loss": 0.9928,
+      "step": 11754
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2594249167696069e-05,
+      "loss": 1.0714,
+      "step": 11755
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2592883112233097e-05,
+      "loss": 1.0666,
+      "step": 11756
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.259151704566717e-05,
+      "loss": 0.9634,
+      "step": 11757
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2590150968023872e-05,
+      "loss": 1.0466,
+      "step": 11758
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2588784879328784e-05,
+      "loss": 1.0191,
+      "step": 11759
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.258741877960749e-05,
+      "loss": 1.0927,
+      "step": 11760
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2586052668885572e-05,
+      "loss": 1.0174,
+      "step": 11761
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2584686547188619e-05,
+      "loss": 0.9935,
+      "step": 11762
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2583320414542202e-05,
+      "loss": 1.0563,
+      "step": 11763
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2581954270971911e-05,
+      "loss": 1.0111,
+      "step": 11764
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2580588116503334e-05,
+      "loss": 1.0314,
+      "step": 11765
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.257922195116205e-05,
+      "loss": 1.0287,
+      "step": 11766
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2577855774973648e-05,
+      "loss": 1.0492,
+      "step": 11767
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2576489587963708e-05,
+      "loss": 1.05,
+      "step": 11768
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2575123390157815e-05,
+      "loss": 1.0524,
+      "step": 11769
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2573757181581555e-05,
+      "loss": 0.9945,
+      "step": 11770
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2572390962260514e-05,
+      "loss": 1.0203,
+      "step": 11771
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2571024732220278e-05,
+      "loss": 1.0622,
+      "step": 11772
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2569658491486427e-05,
+      "loss": 0.9871,
+      "step": 11773
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.256829224008455e-05,
+      "loss": 0.99,
+      "step": 11774
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2566925978040233e-05,
+      "loss": 1.0118,
+      "step": 11775
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2565559705379064e-05,
+      "loss": 1.0282,
+      "step": 11776
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.256419342212663e-05,
+      "loss": 1.0122,
+      "step": 11777
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2562827128308512e-05,
+      "loss": 1.0282,
+      "step": 11778
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.25614608239503e-05,
+      "loss": 1.0174,
+      "step": 11779
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2560094509077581e-05,
+      "loss": 0.9709,
+      "step": 11780
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.255872818371594e-05,
+      "loss": 0.9897,
+      "step": 11781
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2557361847890969e-05,
+      "loss": 1.0519,
+      "step": 11782
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.255599550162825e-05,
+      "loss": 0.972,
+      "step": 11783
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2554629144953371e-05,
+      "loss": 0.992,
+      "step": 11784
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2553262777891926e-05,
+      "loss": 1.0188,
+      "step": 11785
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2551896400469495e-05,
+      "loss": 1.014,
+      "step": 11786
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2550530012711672e-05,
+      "loss": 0.9836,
+      "step": 11787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2549163614644044e-05,
+      "loss": 1.043,
+      "step": 11788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2547797206292198e-05,
+      "loss": 0.9858,
+      "step": 11789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2546430787681724e-05,
+      "loss": 1.0756,
+      "step": 11790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.254506435883821e-05,
+      "loss": 1.038,
+      "step": 11791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2543697919787248e-05,
+      "loss": 1.037,
+      "step": 11792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2542331470554422e-05,
+      "loss": 1.0588,
+      "step": 11793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2540965011165329e-05,
+      "loss": 0.9939,
+      "step": 11794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2539598541645549e-05,
+      "loss": 1.0535,
+      "step": 11795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2538232062020685e-05,
+      "loss": 1.0324,
+      "step": 11796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2536865572316315e-05,
+      "loss": 1.0542,
+      "step": 11797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2535499072558038e-05,
+      "loss": 1.0444,
+      "step": 11798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2534132562771436e-05,
+      "loss": 0.95,
+      "step": 11799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2532766042982108e-05,
+      "loss": 1.026,
+      "step": 11800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2531399513215639e-05,
+      "loss": 0.9403,
+      "step": 11801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2530032973497624e-05,
+      "loss": 1.001,
+      "step": 11802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.252866642385365e-05,
+      "loss": 1.0448,
+      "step": 11803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.252729986430931e-05,
+      "loss": 0.9933,
+      "step": 11804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2525933294890199e-05,
+      "loss": 0.9423,
+      "step": 11805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2524566715621909e-05,
+      "loss": 0.9995,
+      "step": 11806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2523200126530026e-05,
+      "loss": 0.9964,
+      "step": 11807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2521833527640148e-05,
+      "loss": 1.0228,
+      "step": 11808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2520466918977863e-05,
+      "loss": 1.0496,
+      "step": 11809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2519100300568767e-05,
+      "loss": 1.0182,
+      "step": 11810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2517733672438453e-05,
+      "loss": 1.0573,
+      "step": 11811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2516367034612508e-05,
+      "loss": 1.0257,
+      "step": 11812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2515000387116532e-05,
+      "loss": 0.9793,
+      "step": 11813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2513633729976113e-05,
+      "loss": 1.0465,
+      "step": 11814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2512267063216847e-05,
+      "loss": 1.0493,
+      "step": 11815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2510900386864332e-05,
+      "loss": 0.9986,
+      "step": 11816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2509533700944158e-05,
+      "loss": 0.9544,
+      "step": 11817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2508167005481914e-05,
+      "loss": 1.0428,
+      "step": 11818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.25068003005032e-05,
+      "loss": 1.0708,
+      "step": 11819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.250543358603361e-05,
+      "loss": 0.9734,
+      "step": 11820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2504066862098737e-05,
+      "loss": 0.9887,
+      "step": 11821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.250270012872418e-05,
+      "loss": 1.1142,
+      "step": 11822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2501333385935526e-05,
+      "loss": 1.0733,
+      "step": 11823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2499966633758373e-05,
+      "loss": 1.0219,
+      "step": 11824
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2498599872218322e-05,
+      "loss": 0.9726,
+      "step": 11825
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2497233101340961e-05,
+      "loss": 1.0121,
+      "step": 11826
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.249586632115189e-05,
+      "loss": 1.0017,
+      "step": 11827
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2494499531676702e-05,
+      "loss": 1.013,
+      "step": 11828
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2493132732940996e-05,
+      "loss": 0.9845,
+      "step": 11829
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2491765924970367e-05,
+      "loss": 1.0223,
+      "step": 11830
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.249039910779041e-05,
+      "loss": 0.9758,
+      "step": 11831
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2489032281426718e-05,
+      "loss": 0.9778,
+      "step": 11832
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2487665445904894e-05,
+      "loss": 0.935,
+      "step": 11833
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2486298601250534e-05,
+      "loss": 1.0151,
+      "step": 11834
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2484931747489235e-05,
+      "loss": 1.0671,
+      "step": 11835
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2483564884646592e-05,
+      "loss": 1.021,
+      "step": 11836
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2482198012748203e-05,
+      "loss": 1.0307,
+      "step": 11837
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2480831131819664e-05,
+      "loss": 1.0645,
+      "step": 11838
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2479464241886573e-05,
+      "loss": 1.0388,
+      "step": 11839
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2478097342974533e-05,
+      "loss": 1.0853,
+      "step": 11840
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2476730435109135e-05,
+      "loss": 0.9869,
+      "step": 11841
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2475363518315979e-05,
+      "loss": 0.9928,
+      "step": 11842
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2473996592620666e-05,
+      "loss": 1.0263,
+      "step": 11843
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2472629658048793e-05,
+      "loss": 1.0265,
+      "step": 11844
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.247126271462596e-05,
+      "loss": 0.9748,
+      "step": 11845
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2469895762377764e-05,
+      "loss": 1.049,
+      "step": 11846
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2468528801329801e-05,
+      "loss": 1.0159,
+      "step": 11847
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2467161831507677e-05,
+      "loss": 1.0013,
+      "step": 11848
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2465794852936986e-05,
+      "loss": 1.0718,
+      "step": 11849
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2464427865643331e-05,
+      "loss": 0.9917,
+      "step": 11850
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.246306086965231e-05,
+      "loss": 1.0245,
+      "step": 11851
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2461693864989521e-05,
+      "loss": 1.0095,
+      "step": 11852
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2460326851680566e-05,
+      "loss": 1.0215,
+      "step": 11853
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2458959829751048e-05,
+      "loss": 1.0099,
+      "step": 11854
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2457592799226562e-05,
+      "loss": 1.0493,
+      "step": 11855
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2456225760132711e-05,
+      "loss": 0.9934,
+      "step": 11856
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2454858712495094e-05,
+      "loss": 0.9928,
+      "step": 11857
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2453491656339314e-05,
+      "loss": 1.0505,
+      "step": 11858
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.245212459169097e-05,
+      "loss": 1.0152,
+      "step": 11859
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2450757518575666e-05,
+      "loss": 0.9972,
+      "step": 11860
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2449390437019001e-05,
+      "loss": 1.0628,
+      "step": 11861
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2448023347046576e-05,
+      "loss": 1.0457,
+      "step": 11862
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2446656248683995e-05,
+      "loss": 1.025,
+      "step": 11863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2445289141956858e-05,
+      "loss": 0.9897,
+      "step": 11864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2443922026890765e-05,
+      "loss": 1.0552,
+      "step": 11865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2442554903511321e-05,
+      "loss": 1.0416,
+      "step": 11866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2441187771844124e-05,
+      "loss": 0.9784,
+      "step": 11867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2439820631914782e-05,
+      "loss": 1.0033,
+      "step": 11868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2438453483748898e-05,
+      "loss": 1.0363,
+      "step": 11869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2437086327372068e-05,
+      "loss": 1.0312,
+      "step": 11870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2435719162809897e-05,
+      "loss": 1.0564,
+      "step": 11871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2434351990087994e-05,
+      "loss": 1.0625,
+      "step": 11872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2432984809231956e-05,
+      "loss": 1.0293,
+      "step": 11873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2431617620267388e-05,
+      "loss": 1.0244,
+      "step": 11874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2430250423219894e-05,
+      "loss": 1.0678,
+      "step": 11875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2428883218115072e-05,
+      "loss": 1.0704,
+      "step": 11876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2427516004978534e-05,
+      "loss": 1.0289,
+      "step": 11877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2426148783835881e-05,
+      "loss": 0.992,
+      "step": 11878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2424781554712717e-05,
+      "loss": 1.0297,
+      "step": 11879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2423414317634645e-05,
+      "loss": 0.9971,
+      "step": 11880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2422047072627267e-05,
+      "loss": 0.9426,
+      "step": 11881
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2420679819716194e-05,
+      "loss": 0.9524,
+      "step": 11882
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2419312558927027e-05,
+      "loss": 1.0574,
+      "step": 11883
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.241794529028537e-05,
+      "loss": 0.9895,
+      "step": 11884
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2416578013816829e-05,
+      "loss": 1.0379,
+      "step": 11885
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.241521072954701e-05,
+      "loss": 0.9883,
+      "step": 11886
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2413843437501513e-05,
+      "loss": 1.0438,
+      "step": 11887
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2412476137705949e-05,
+      "loss": 1.0104,
+      "step": 11888
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2411108830185925e-05,
+      "loss": 0.9479,
+      "step": 11889
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.240974151496704e-05,
+      "loss": 1.0352,
+      "step": 11890
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2408374192074906e-05,
+      "loss": 1.03,
+      "step": 11891
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2407006861535122e-05,
+      "loss": 1.0155,
+      "step": 11892
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.24056395233733e-05,
+      "loss": 1.0145,
+      "step": 11893
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2404272177615048e-05,
+      "loss": 1.0188,
+      "step": 11894
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2402904824285966e-05,
+      "loss": 1.0092,
+      "step": 11895
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2401537463411664e-05,
+      "loss": 1.032,
+      "step": 11896
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2400170095017745e-05,
+      "loss": 0.9927,
+      "step": 11897
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2398802719129823e-05,
+      "loss": 0.9611,
+      "step": 11898
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.23974353357735e-05,
+      "loss": 1.0598,
+      "step": 11899
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.239606794497438e-05,
+      "loss": 1.0413,
+      "step": 11900
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.239470054675808e-05,
+      "loss": 1.0347,
+      "step": 11901
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2393333141150196e-05,
+      "loss": 0.9859,
+      "step": 11902
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2391965728176344e-05,
+      "loss": 1.0041,
+      "step": 11903
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.239059830786213e-05,
+      "loss": 1.0451,
+      "step": 11904
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2389230880233156e-05,
+      "loss": 0.9693,
+      "step": 11905
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2387863445315033e-05,
+      "loss": 1.0786,
+      "step": 11906
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2386496003133373e-05,
+      "loss": 0.9796,
+      "step": 11907
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2385128553713784e-05,
+      "loss": 1.0366,
+      "step": 11908
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2383761097081868e-05,
+      "loss": 1.0736,
+      "step": 11909
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2382393633263239e-05,
+      "loss": 0.9797,
+      "step": 11910
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.23810261622835e-05,
+      "loss": 1.0678,
+      "step": 11911
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2379658684168266e-05,
+      "loss": 1.0547,
+      "step": 11912
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2378291198943142e-05,
+      "loss": 0.9329,
+      "step": 11913
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.237692370663374e-05,
+      "loss": 1.0365,
+      "step": 11914
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2375556207265664e-05,
+      "loss": 0.952,
+      "step": 11915
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2374188700864527e-05,
+      "loss": 1.0213,
+      "step": 11916
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2372821187455939e-05,
+      "loss": 1.0947,
+      "step": 11917
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2371453667065508e-05,
+      "loss": 1.0188,
+      "step": 11918
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2370086139718844e-05,
+      "loss": 1.0558,
+      "step": 11919
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2368718605441558e-05,
+      "loss": 1.0427,
+      "step": 11920
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2367351064259256e-05,
+      "loss": 1.0435,
+      "step": 11921
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2365983516197553e-05,
+      "loss": 1.0373,
+      "step": 11922
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2364615961282055e-05,
+      "loss": 0.9716,
+      "step": 11923
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2363248399538374e-05,
+      "loss": 0.9544,
+      "step": 11924
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2361880830992122e-05,
+      "loss": 1.0319,
+      "step": 11925
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2360513255668904e-05,
+      "loss": 0.9612,
+      "step": 11926
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2359145673594338e-05,
+      "loss": 1.0263,
+      "step": 11927
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2357778084794033e-05,
+      "loss": 1.0753,
+      "step": 11928
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2356410489293598e-05,
+      "loss": 1.0752,
+      "step": 11929
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.235504288711864e-05,
+      "loss": 1.0238,
+      "step": 11930
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2353675278294777e-05,
+      "loss": 1.0437,
+      "step": 11931
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.235230766284762e-05,
+      "loss": 1.0062,
+      "step": 11932
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2350940040802775e-05,
+      "loss": 0.9615,
+      "step": 11933
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2349572412185857e-05,
+      "loss": 1.034,
+      "step": 11934
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2348204777022477e-05,
+      "loss": 1.0594,
+      "step": 11935
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2346837135338248e-05,
+      "loss": 1.1535,
+      "step": 11936
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2345469487158781e-05,
+      "loss": 0.9831,
+      "step": 11937
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2344101832509687e-05,
+      "loss": 0.9813,
+      "step": 11938
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2342734171416581e-05,
+      "loss": 0.9717,
+      "step": 11939
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2341366503905074e-05,
+      "loss": 1.0705,
+      "step": 11940
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2339998830000777e-05,
+      "loss": 1.0225,
+      "step": 11941
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2338631149729303e-05,
+      "loss": 1.0243,
+      "step": 11942
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2337263463116265e-05,
+      "loss": 1.0804,
+      "step": 11943
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2335895770187274e-05,
+      "loss": 0.9907,
+      "step": 11944
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2334528070967946e-05,
+      "loss": 1.0057,
+      "step": 11945
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2333160365483891e-05,
+      "loss": 1.0404,
+      "step": 11946
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2331792653760728e-05,
+      "loss": 1.0537,
+      "step": 11947
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2330424935824066e-05,
+      "loss": 1.0052,
+      "step": 11948
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2329057211699513e-05,
+      "loss": 1.0109,
+      "step": 11949
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.232768948141269e-05,
+      "loss": 0.9926,
+      "step": 11950
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2326321744989208e-05,
+      "loss": 1.0145,
+      "step": 11951
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.2324954002454683e-05,
+      "loss": 0.9903,
+      "step": 11952
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2323586253834722e-05,
+      "loss": 1.0532,
+      "step": 11953
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2322218499154947e-05,
+      "loss": 1.0069,
+      "step": 11954
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2320850738440968e-05,
+      "loss": 1.0422,
+      "step": 11955
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2319482971718401e-05,
+      "loss": 1.0687,
+      "step": 11956
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2318115199012858e-05,
+      "loss": 1.0645,
+      "step": 11957
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2316747420349955e-05,
+      "loss": 1.0405,
+      "step": 11958
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2315379635755302e-05,
+      "loss": 0.9412,
+      "step": 11959
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2314011845254521e-05,
+      "loss": 1.035,
+      "step": 11960
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2312644048873222e-05,
+      "loss": 1.0175,
+      "step": 11961
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2311276246637022e-05,
+      "loss": 1.0148,
+      "step": 11962
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2309908438571531e-05,
+      "loss": 1.0344,
+      "step": 11963
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2308540624702372e-05,
+      "loss": 1.0678,
+      "step": 11964
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2307172805055155e-05,
+      "loss": 0.9782,
+      "step": 11965
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2305804979655495e-05,
+      "loss": 1.0131,
+      "step": 11966
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2304437148529011e-05,
+      "loss": 0.9731,
+      "step": 11967
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2303069311701312e-05,
+      "loss": 1.0304,
+      "step": 11968
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.230170146919802e-05,
+      "loss": 1.0558,
+      "step": 11969
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2300333621044747e-05,
+      "loss": 1.0698,
+      "step": 11970
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.229896576726711e-05,
+      "loss": 1.0671,
+      "step": 11971
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2297597907890725e-05,
+      "loss": 0.9703,
+      "step": 11972
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2296230042941206e-05,
+      "loss": 1.04,
+      "step": 11973
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2294862172444172e-05,
+      "loss": 0.9874,
+      "step": 11974
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2293494296425242e-05,
+      "loss": 1.0268,
+      "step": 11975
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2292126414910026e-05,
+      "loss": 1.0463,
+      "step": 11976
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.229075852792414e-05,
+      "loss": 1.0836,
+      "step": 11977
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2289390635493203e-05,
+      "loss": 1.0748,
+      "step": 11978
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2288022737642835e-05,
+      "loss": 0.9948,
+      "step": 11979
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2286654834398646e-05,
+      "loss": 1.0603,
+      "step": 11980
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.228528692578626e-05,
+      "loss": 1.0546,
+      "step": 11981
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2283919011831288e-05,
+      "loss": 1.0065,
+      "step": 11982
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2282551092559348e-05,
+      "loss": 1.0009,
+      "step": 11983
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2281183167996061e-05,
+      "loss": 0.9681,
+      "step": 11984
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227981523816704e-05,
+      "loss": 1.0513,
+      "step": 11985
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2278447303097907e-05,
+      "loss": 1.0084,
+      "step": 11986
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2277079362814273e-05,
+      "loss": 1.0,
+      "step": 11987
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2275711417341758e-05,
+      "loss": 1.0075,
+      "step": 11988
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227434346670598e-05,
+      "loss": 0.9808,
+      "step": 11989
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2272975510932557e-05,
+      "loss": 1.0635,
+      "step": 11990
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2271607550047108e-05,
+      "loss": 1.0175,
+      "step": 11991
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.227023958407525e-05,
+      "loss": 1.0186,
+      "step": 11992
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2268871613042599e-05,
+      "loss": 0.9845,
+      "step": 11993
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2267503636974777e-05,
+      "loss": 1.0072,
+      "step": 11994
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2266135655897398e-05,
+      "loss": 0.9839,
+      "step": 11995
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2264767669836084e-05,
+      "loss": 1.1024,
+      "step": 11996
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.226339967881645e-05,
+      "loss": 1.0419,
+      "step": 11997
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2262031682864113e-05,
+      "loss": 1.0259,
+      "step": 11998
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2260663682004696e-05,
+      "loss": 0.994,
+      "step": 11999
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2259295676263818e-05,
+      "loss": 1.034,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "eval_text_loss": 0.9588204622268677,
+      "eval_text_runtime": 1.5697,
+      "eval_text_samples_per_second": 47.144,
+      "eval_text_steps_per_second": 1.274,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "eval_svit_loss": 1.113356590270996,
+      "eval_svit_runtime": 3.6948,
+      "eval_svit_samples_per_second": 184.312,
+      "eval_svit_steps_per_second": 2.977,
+      "step": 12000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2257927665667095e-05,
+      "loss": 0.9723,
+      "step": 12001
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2256559650240144e-05,
+      "loss": 0.9782,
+      "step": 12002
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2255191630008591e-05,
+      "loss": 0.9499,
+      "step": 12003
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2253823604998049e-05,
+      "loss": 0.9721,
+      "step": 12004
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2252455575234139e-05,
+      "loss": 0.9534,
+      "step": 12005
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2251087540742477e-05,
+      "loss": 0.9506,
+      "step": 12006
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2249719501548685e-05,
+      "loss": 1.1367,
+      "step": 12007
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2248351457678382e-05,
+      "loss": 1.006,
+      "step": 12008
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.224698340915719e-05,
+      "loss": 0.9931,
+      "step": 12009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2245615356010727e-05,
+      "loss": 1.0223,
+      "step": 12010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2244247298264608e-05,
+      "loss": 1.0386,
+      "step": 12011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2242879235944458e-05,
+      "loss": 1.0281,
+      "step": 12012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2241511169075899e-05,
+      "loss": 0.9931,
+      "step": 12013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2240143097684543e-05,
+      "loss": 0.9921,
+      "step": 12014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2238775021796017e-05,
+      "loss": 0.9898,
+      "step": 12015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2237406941435937e-05,
+      "loss": 1.0611,
+      "step": 12016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2236038856629922e-05,
+      "loss": 1.0002,
+      "step": 12017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2234670767403593e-05,
+      "loss": 1.0372,
+      "step": 12018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2233302673782577e-05,
+      "loss": 0.9912,
+      "step": 12019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2231934575792488e-05,
+      "loss": 1.0507,
+      "step": 12020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2230566473458944e-05,
+      "loss": 1.0796,
+      "step": 12021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2229198366807573e-05,
+      "loss": 1.0739,
+      "step": 12022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.222783025586399e-05,
+      "loss": 1.0106,
+      "step": 12023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2226462140653815e-05,
+      "loss": 1.0406,
+      "step": 12024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2225094021202677e-05,
+      "loss": 1.0468,
+      "step": 12025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2223725897536187e-05,
+      "loss": 1.0647,
+      "step": 12026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2222357769679971e-05,
+      "loss": 1.0346,
+      "step": 12027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2220989637659646e-05,
+      "loss": 1.0318,
+      "step": 12028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2219621501500837e-05,
+      "loss": 0.9969,
+      "step": 12029
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2218253361229168e-05,
+      "loss": 0.9678,
+      "step": 12030
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2216885216870254e-05,
+      "loss": 1.0562,
+      "step": 12031
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.221551706844972e-05,
+      "loss": 1.0354,
+      "step": 12032
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2214148915993184e-05,
+      "loss": 1.0518,
+      "step": 12033
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.221278075952627e-05,
+      "loss": 1.0083,
+      "step": 12034
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2211412599074602e-05,
+      "loss": 0.9485,
+      "step": 12035
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2210044434663795e-05,
+      "loss": 1.0824,
+      "step": 12036
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2208676266319474e-05,
+      "loss": 0.9252,
+      "step": 12037
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2207308094067262e-05,
+      "loss": 0.97,
+      "step": 12038
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2205939917932782e-05,
+      "loss": 1.0324,
+      "step": 12039
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2204571737941652e-05,
+      "loss": 1.0784,
+      "step": 12040
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2203203554119498e-05,
+      "loss": 1.0208,
+      "step": 12041
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2201835366491938e-05,
+      "loss": 1.0631,
+      "step": 12042
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2200467175084596e-05,
+      "loss": 1.0732,
+      "step": 12043
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2199098979923095e-05,
+      "loss": 1.1134,
+      "step": 12044
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2197730781033057e-05,
+      "loss": 1.0459,
+      "step": 12045
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2196362578440102e-05,
+      "loss": 1.0313,
+      "step": 12046
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2194994372169854e-05,
+      "loss": 1.0219,
+      "step": 12047
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2193626162247937e-05,
+      "loss": 1.033,
+      "step": 12048
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2192257948699975e-05,
+      "loss": 1.0035,
+      "step": 12049
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2190889731551588e-05,
+      "loss": 0.9797,
+      "step": 12050
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2189521510828394e-05,
+      "loss": 0.9797,
+      "step": 12051
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2188153286556021e-05,
+      "loss": 0.9825,
+      "step": 12052
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2186785058760092e-05,
+      "loss": 1.0878,
+      "step": 12053
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2185416827466231e-05,
+      "loss": 1.0197,
+      "step": 12054
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2184048592700056e-05,
+      "loss": 1.0911,
+      "step": 12055
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2182680354487195e-05,
+      "loss": 1.0165,
+      "step": 12056
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.218131211285327e-05,
+      "loss": 0.992,
+      "step": 12057
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.21799438678239e-05,
+      "loss": 1.006,
+      "step": 12058
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2178575619424715e-05,
+      "loss": 1.014,
+      "step": 12059
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2177207367681334e-05,
+      "loss": 0.9915,
+      "step": 12060
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.217583911261938e-05,
+      "loss": 1.0391,
+      "step": 12061
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2174470854264475e-05,
+      "loss": 1.1583,
+      "step": 12062
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2173102592642245e-05,
+      "loss": 1.0376,
+      "step": 12063
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2171734327778317e-05,
+      "loss": 1.0004,
+      "step": 12064
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2170366059698308e-05,
+      "loss": 1.0547,
+      "step": 12065
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2168997788427846e-05,
+      "loss": 0.9942,
+      "step": 12066
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2167629513992552e-05,
+      "loss": 1.0951,
+      "step": 12067
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.216626123641805e-05,
+      "loss": 1.0362,
+      "step": 12068
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2164892955729969e-05,
+      "loss": 1.0795,
+      "step": 12069
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2163524671953926e-05,
+      "loss": 1.0107,
+      "step": 12070
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2162156385115547e-05,
+      "loss": 1.0664,
+      "step": 12071
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2160788095240457e-05,
+      "loss": 1.0332,
+      "step": 12072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.215941980235428e-05,
+      "loss": 1.061,
+      "step": 12073
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2158051506482637e-05,
+      "loss": 1.0361,
+      "step": 12074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2156683207651159e-05,
+      "loss": 1.0045,
+      "step": 12075
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2155314905885462e-05,
+      "loss": 1.0794,
+      "step": 12076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2153946601211177e-05,
+      "loss": 1.0251,
+      "step": 12077
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2152578293653925e-05,
+      "loss": 1.0489,
+      "step": 12078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2151209983239331e-05,
+      "loss": 1.0568,
+      "step": 12079
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2149841669993018e-05,
+      "loss": 1.0004,
+      "step": 12080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2148473353940615e-05,
+      "loss": 1.0577,
+      "step": 12081
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.214710503510774e-05,
+      "loss": 1.0542,
+      "step": 12082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2145736713520024e-05,
+      "loss": 1.1157,
+      "step": 12083
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2144368389203087e-05,
+      "loss": 1.0937,
+      "step": 12084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2143000062182554e-05,
+      "loss": 1.036,
+      "step": 12085
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.214163173248405e-05,
+      "loss": 1.0715,
+      "step": 12086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2140263400133208e-05,
+      "loss": 0.9735,
+      "step": 12087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.213889506515564e-05,
+      "loss": 0.9811,
+      "step": 12088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2137526727576978e-05,
+      "loss": 0.921,
+      "step": 12089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2136158387422845e-05,
+      "loss": 0.9669,
+      "step": 12090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2134790044718864e-05,
+      "loss": 1.072,
+      "step": 12091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2133421699490664e-05,
+      "loss": 1.0493,
+      "step": 12092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2132053351763872e-05,
+      "loss": 1.058,
+      "step": 12093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2130685001564107e-05,
+      "loss": 1.0284,
+      "step": 12094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2129316648916996e-05,
+      "loss": 0.9812,
+      "step": 12095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2127948293848166e-05,
+      "loss": 1.0608,
+      "step": 12096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.212657993638324e-05,
+      "loss": 1.041,
+      "step": 12097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2125211576547848e-05,
+      "loss": 1.0413,
+      "step": 12098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.212384321436761e-05,
+      "loss": 1.023,
+      "step": 12099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2122474849868153e-05,
+      "loss": 0.962,
+      "step": 12100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2121106483075102e-05,
+      "loss": 1.0546,
+      "step": 12101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2119738114014083e-05,
+      "loss": 0.9797,
+      "step": 12102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2118369742710725e-05,
+      "loss": 0.9825,
+      "step": 12103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2117001369190649e-05,
+      "loss": 1.0198,
+      "step": 12104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.211563299347948e-05,
+      "loss": 1.0423,
+      "step": 12105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2114264615602848e-05,
+      "loss": 1.0584,
+      "step": 12106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2112896235586378e-05,
+      "loss": 1.0398,
+      "step": 12107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.211152785345569e-05,
+      "loss": 1.02,
+      "step": 12108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2110159469236413e-05,
+      "loss": 1.0304,
+      "step": 12109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2108791082954177e-05,
+      "loss": 0.9956,
+      "step": 12110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.21074226946346e-05,
+      "loss": 1.0087,
+      "step": 12111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2106054304303319e-05,
+      "loss": 1.0321,
+      "step": 12112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.210468591198595e-05,
+      "loss": 1.0246,
+      "step": 12113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.210331751770812e-05,
+      "loss": 0.992,
+      "step": 12114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2101949121495463e-05,
+      "loss": 0.9779,
+      "step": 12115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2100580723373593e-05,
+      "loss": 0.9889,
+      "step": 12116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2099212323368145e-05,
+      "loss": 1.0576,
+      "step": 12117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2097843921504744e-05,
+      "loss": 1.0304,
+      "step": 12118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2096475517809011e-05,
+      "loss": 1.0233,
+      "step": 12119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2095107112306577e-05,
+      "loss": 0.9571,
+      "step": 12120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2093738705023066e-05,
+      "loss": 0.9735,
+      "step": 12121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2092370295984107e-05,
+      "loss": 0.9915,
+      "step": 12122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2091001885215325e-05,
+      "loss": 1.0385,
+      "step": 12123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2089633472742343e-05,
+      "loss": 1.0652,
+      "step": 12124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2088265058590794e-05,
+      "loss": 1.032,
+      "step": 12125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2086896642786297e-05,
+      "loss": 1.0521,
+      "step": 12126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2085528225354486e-05,
+      "loss": 0.9591,
+      "step": 12127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2084159806320978e-05,
+      "loss": 1.0231,
+      "step": 12128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2082791385711407e-05,
+      "loss": 0.9721,
+      "step": 12129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2081422963551397e-05,
+      "loss": 0.9237,
+      "step": 12130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2080054539866575e-05,
+      "loss": 1.0759,
+      "step": 12131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2078686114682567e-05,
+      "loss": 1.0666,
+      "step": 12132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2077317688024999e-05,
+      "loss": 1.0872,
+      "step": 12133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2075949259919502e-05,
+      "loss": 1.0297,
+      "step": 12134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2074580830391696e-05,
+      "loss": 1.0458,
+      "step": 12135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2073212399467209e-05,
+      "loss": 1.0455,
+      "step": 12136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2071843967171675e-05,
+      "loss": 1.0551,
+      "step": 12137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.207047553353071e-05,
+      "loss": 1.0837,
+      "step": 12138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2069107098569948e-05,
+      "loss": 1.0322,
+      "step": 12139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2067738662315011e-05,
+      "loss": 0.9716,
+      "step": 12140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2066370224791533e-05,
+      "loss": 1.016,
+      "step": 12141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2065001786025131e-05,
+      "loss": 1.077,
+      "step": 12142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2063633346041442e-05,
+      "loss": 1.0155,
+      "step": 12143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2062264904866085e-05,
+      "loss": 1.0533,
+      "step": 12144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.206089646252469e-05,
+      "loss": 1.0009,
+      "step": 12145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2059528019042885e-05,
+      "loss": 1.0184,
+      "step": 12146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2058159574446295e-05,
+      "loss": 1.0031,
+      "step": 12147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2056791128760544e-05,
+      "loss": 0.9882,
+      "step": 12148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2055422682011266e-05,
+      "loss": 0.9814,
+      "step": 12149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2054054234224084e-05,
+      "loss": 0.9638,
+      "step": 12150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2052685785424627e-05,
+      "loss": 1.0151,
+      "step": 12151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2051317335638517e-05,
+      "loss": 1.0372,
+      "step": 12152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2049948884891387e-05,
+      "loss": 1.015,
+      "step": 12153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.204858043320886e-05,
+      "loss": 1.0312,
+      "step": 12154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2047211980616564e-05,
+      "loss": 1.0468,
+      "step": 12155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2045843527140131e-05,
+      "loss": 0.9969,
+      "step": 12156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.204447507280518e-05,
+      "loss": 1.0353,
+      "step": 12157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2043106617637341e-05,
+      "loss": 0.9983,
+      "step": 12158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2041738161662241e-05,
+      "loss": 1.0208,
+      "step": 12159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2040369704905512e-05,
+      "loss": 0.9948,
+      "step": 12160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2039001247392776e-05,
+      "loss": 1.0055,
+      "step": 12161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2037632789149664e-05,
+      "loss": 1.0468,
+      "step": 12162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2036264330201797e-05,
+      "loss": 1.0034,
+      "step": 12163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2034895870574807e-05,
+      "loss": 1.0288,
+      "step": 12164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2033527410294318e-05,
+      "loss": 1.0312,
+      "step": 12165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2032158949385964e-05,
+      "loss": 1.0138,
+      "step": 12166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2030790487875366e-05,
+      "loss": 0.9866,
+      "step": 12167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.202942202578815e-05,
+      "loss": 1.0126,
+      "step": 12168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2028053563149949e-05,
+      "loss": 0.9988,
+      "step": 12169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2026685099986385e-05,
+      "loss": 1.0512,
+      "step": 12170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.202531663632309e-05,
+      "loss": 1.0575,
+      "step": 12171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.202394817218569e-05,
+      "loss": 0.9926,
+      "step": 12172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2022579707599808e-05,
+      "loss": 1.0169,
+      "step": 12173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2021211242591076e-05,
+      "loss": 1.0736,
+      "step": 12174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2019842777185123e-05,
+      "loss": 0.9864,
+      "step": 12175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2018474311407568e-05,
+      "loss": 0.9823,
+      "step": 12176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2017105845284045e-05,
+      "loss": 0.9669,
+      "step": 12177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.201573737884018e-05,
+      "loss": 1.0979,
+      "step": 12178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2014368912101599e-05,
+      "loss": 1.0322,
+      "step": 12179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2013000445093937e-05,
+      "loss": 1.0534,
+      "step": 12180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.201163197784281e-05,
+      "loss": 1.0209,
+      "step": 12181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2010263510373849e-05,
+      "loss": 1.0301,
+      "step": 12182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2008895042712684e-05,
+      "loss": 0.9614,
+      "step": 12183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2007526574884942e-05,
+      "loss": 0.9749,
+      "step": 12184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.200615810691625e-05,
+      "loss": 1.0484,
+      "step": 12185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2004789638832232e-05,
+      "loss": 1.0172,
+      "step": 12186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.200342117065852e-05,
+      "loss": 1.0192,
+      "step": 12187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.2002052702420739e-05,
+      "loss": 1.0145,
+      "step": 12188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.200068423414452e-05,
+      "loss": 0.9663,
+      "step": 12189
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1999315765855483e-05,
+      "loss": 1.0167,
+      "step": 12190
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.199794729757926e-05,
+      "loss": 1.0681,
+      "step": 12191
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1996578829341481e-05,
+      "loss": 0.9661,
+      "step": 12192
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.199521036116777e-05,
+      "loss": 1.0146,
+      "step": 12193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1993841893083751e-05,
+      "loss": 1.0527,
+      "step": 12194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1992473425115059e-05,
+      "loss": 1.0793,
+      "step": 12195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1991104957287315e-05,
+      "loss": 1.0087,
+      "step": 12196
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1989736489626152e-05,
+      "loss": 1.0283,
+      "step": 12197
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1988368022157194e-05,
+      "loss": 0.9969,
+      "step": 12198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1986999554906067e-05,
+      "loss": 0.9598,
+      "step": 12199
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1985631087898402e-05,
+      "loss": 1.034,
+      "step": 12200
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.198426262115982e-05,
+      "loss": 0.9804,
+      "step": 12201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1982894154715957e-05,
+      "loss": 1.0296,
+      "step": 12202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1981525688592435e-05,
+      "loss": 0.9863,
+      "step": 12203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1980157222814881e-05,
+      "loss": 0.9946,
+      "step": 12204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1978788757408925e-05,
+      "loss": 1.0135,
+      "step": 12205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1977420292400191e-05,
+      "loss": 1.0353,
+      "step": 12206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1976051827814311e-05,
+      "loss": 1.0196,
+      "step": 12207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1974683363676912e-05,
+      "loss": 1.0167,
+      "step": 12208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1973314900013616e-05,
+      "loss": 1.0801,
+      "step": 12209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1971946436850055e-05,
+      "loss": 0.9986,
+      "step": 12210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.197057797421185e-05,
+      "loss": 1.0638,
+      "step": 12211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1969209512124638e-05,
+      "loss": 1.0499,
+      "step": 12212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.196784105061404e-05,
+      "loss": 1.0416,
+      "step": 12213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1966472589705681e-05,
+      "loss": 1.0202,
+      "step": 12214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1965104129425195e-05,
+      "loss": 1.0348,
+      "step": 12215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1963735669798204e-05,
+      "loss": 1.0716,
+      "step": 12216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1962367210850338e-05,
+      "loss": 1.0619,
+      "step": 12217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1960998752607224e-05,
+      "loss": 0.9949,
+      "step": 12218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1959630295094488e-05,
+      "loss": 0.956,
+      "step": 12219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.195826183833776e-05,
+      "loss": 0.962,
+      "step": 12220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1956893382362658e-05,
+      "loss": 1.018,
+      "step": 12221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1955524927194822e-05,
+      "loss": 0.992,
+      "step": 12222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1954156472859874e-05,
+      "loss": 1.0129,
+      "step": 12223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1952788019383435e-05,
+      "loss": 1.0042,
+      "step": 12224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1951419566791142e-05,
+      "loss": 1.0411,
+      "step": 12225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1950051115108617e-05,
+      "loss": 1.033,
+      "step": 12226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1948682664361485e-05,
+      "loss": 1.044,
+      "step": 12227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1947314214575379e-05,
+      "loss": 1.0553,
+      "step": 12228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1945945765775916e-05,
+      "loss": 1.059,
+      "step": 12229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1944577317988736e-05,
+      "loss": 1.0042,
+      "step": 12230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1943208871239458e-05,
+      "loss": 1.0636,
+      "step": 12231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1941840425553707e-05,
+      "loss": 1.0482,
+      "step": 12232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1940471980957119e-05,
+      "loss": 1.07,
+      "step": 12233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.193910353747531e-05,
+      "loss": 0.9975,
+      "step": 12234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1937735095133916e-05,
+      "loss": 1.054,
+      "step": 12235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1936366653958562e-05,
+      "loss": 0.93,
+      "step": 12236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1934998213974869e-05,
+      "loss": 0.9833,
+      "step": 12237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1933629775208471e-05,
+      "loss": 0.9852,
+      "step": 12238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1932261337684988e-05,
+      "loss": 1.0181,
+      "step": 12239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1930892901430055e-05,
+      "loss": 0.9567,
+      "step": 12240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1929524466469295e-05,
+      "loss": 0.9941,
+      "step": 12241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1928156032828329e-05,
+      "loss": 1.0364,
+      "step": 12242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1926787600532792e-05,
+      "loss": 1.0392,
+      "step": 12243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1925419169608305e-05,
+      "loss": 1.0298,
+      "step": 12244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.19240507400805e-05,
+      "loss": 0.9968,
+      "step": 12245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1922682311975002e-05,
+      "loss": 1.0503,
+      "step": 12246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1921313885317435e-05,
+      "loss": 0.9545,
+      "step": 12247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1919945460133429e-05,
+      "loss": 0.9701,
+      "step": 12248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1918577036448602e-05,
+      "loss": 1.0194,
+      "step": 12249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1917208614288595e-05,
+      "loss": 1.0356,
+      "step": 12250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1915840193679025e-05,
+      "loss": 1.0488,
+      "step": 12251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1914471774645517e-05,
+      "loss": 1.0004,
+      "step": 12252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1913103357213704e-05,
+      "loss": 1.0488,
+      "step": 12253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1911734941409209e-05,
+      "loss": 0.9772,
+      "step": 12254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1910366527257658e-05,
+      "loss": 1.0358,
+      "step": 12255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1908998114784677e-05,
+      "loss": 1.0072,
+      "step": 12256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1907629704015893e-05,
+      "loss": 0.9942,
+      "step": 12257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1906261294976937e-05,
+      "loss": 1.0432,
+      "step": 12258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1904892887693424e-05,
+      "loss": 1.0369,
+      "step": 12259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.190352448219099e-05,
+      "loss": 0.9952,
+      "step": 12260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1902156078495262e-05,
+      "loss": 1.0253,
+      "step": 12261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1900787676631855e-05,
+      "loss": 0.9824,
+      "step": 12262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1899419276626409e-05,
+      "loss": 1.129,
+      "step": 12263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1898050878504541e-05,
+      "loss": 0.9772,
+      "step": 12264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1896682482291881e-05,
+      "loss": 1.0028,
+      "step": 12265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1895314088014055e-05,
+      "loss": 0.9303,
+      "step": 12266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1893945695696684e-05,
+      "loss": 1.0132,
+      "step": 12267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18925773053654e-05,
+      "loss": 1.0038,
+      "step": 12268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1891208917045823e-05,
+      "loss": 0.9685,
+      "step": 12269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1889840530763588e-05,
+      "loss": 0.9331,
+      "step": 12270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1888472146544315e-05,
+      "loss": 0.9911,
+      "step": 12271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1887103764413625e-05,
+      "loss": 1.0358,
+      "step": 12272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1885735384397154e-05,
+      "loss": 1.0167,
+      "step": 12273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1884367006520521e-05,
+      "loss": 0.9939,
+      "step": 12274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1882998630809355e-05,
+      "loss": 1.0442,
+      "step": 12275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1881630257289279e-05,
+      "loss": 1.0098,
+      "step": 12276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1880261885985916e-05,
+      "loss": 1.094,
+      "step": 12277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.18788935169249e-05,
+      "loss": 1.0641,
+      "step": 12278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1877525150131848e-05,
+      "loss": 0.9891,
+      "step": 12279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1876156785632392e-05,
+      "loss": 1.0565,
+      "step": 12280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1874788423452156e-05,
+      "loss": 1.0077,
+      "step": 12281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.187342006361676e-05,
+      "loss": 1.0284,
+      "step": 12282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1872051706151836e-05,
+      "loss": 1.0698,
+      "step": 12283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1870683351083005e-05,
+      "loss": 0.9993,
+      "step": 12284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1869314998435895e-05,
+      "loss": 1.0594,
+      "step": 12285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1867946648236132e-05,
+      "loss": 1.0442,
+      "step": 12286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1866578300509335e-05,
+      "loss": 1.0085,
+      "step": 12287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1865209955281136e-05,
+      "loss": 1.0205,
+      "step": 12288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1863841612577155e-05,
+      "loss": 1.0173,
+      "step": 12289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1862473272423024e-05,
+      "loss": 1.0471,
+      "step": 12290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1861104934844363e-05,
+      "loss": 1.0384,
+      "step": 12291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1859736599866795e-05,
+      "loss": 1.0871,
+      "step": 12292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.185836826751595e-05,
+      "loss": 0.9697,
+      "step": 12293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1856999937817448e-05,
+      "loss": 0.9531,
+      "step": 12294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1855631610796915e-05,
+      "loss": 1.005,
+      "step": 12295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.185426328647998e-05,
+      "loss": 1.0063,
+      "step": 12296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1852894964892259e-05,
+      "loss": 1.051,
+      "step": 12297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1851526646059387e-05,
+      "loss": 1.0494,
+      "step": 12298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1850158330006985e-05,
+      "loss": 1.0502,
+      "step": 12299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.184879001676067e-05,
+      "loss": 1.0342,
+      "step": 12300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1847421706346077e-05,
+      "loss": 1.0546,
+      "step": 12301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1846053398788825e-05,
+      "loss": 1.0116,
+      "step": 12302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.184468509411454e-05,
+      "loss": 1.0351,
+      "step": 12303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1843316792348847e-05,
+      "loss": 0.9774,
+      "step": 12304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1841948493517364e-05,
+      "loss": 1.0395,
+      "step": 12305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1840580197645724e-05,
+      "loss": 1.0195,
+      "step": 12306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1839211904759544e-05,
+      "loss": 0.9594,
+      "step": 12307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1837843614884455e-05,
+      "loss": 1.0699,
+      "step": 12308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1836475328046078e-05,
+      "loss": 1.0304,
+      "step": 12309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1835107044270032e-05,
+      "loss": 1.0807,
+      "step": 12310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.183373876358195e-05,
+      "loss": 1.0064,
+      "step": 12311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1832370486007449e-05,
+      "loss": 1.0026,
+      "step": 12312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1831002211572158e-05,
+      "loss": 1.1051,
+      "step": 12313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1829633940301696e-05,
+      "loss": 1.0104,
+      "step": 12314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1828265672221685e-05,
+      "loss": 0.9565,
+      "step": 12315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1826897407357757e-05,
+      "loss": 0.9881,
+      "step": 12316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1825529145735525e-05,
+      "loss": 1.1221,
+      "step": 12317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1824160887380625e-05,
+      "loss": 1.0013,
+      "step": 12318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1822792632318672e-05,
+      "loss": 1.0443,
+      "step": 12319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1821424380575289e-05,
+      "loss": 1.0595,
+      "step": 12320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1820056132176103e-05,
+      "loss": 0.9761,
+      "step": 12321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181868788714673e-05,
+      "loss": 1.0406,
+      "step": 12322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1817319645512806e-05,
+      "loss": 0.9965,
+      "step": 12323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1815951407299946e-05,
+      "loss": 1.0317,
+      "step": 12324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.181458317253377e-05,
+      "loss": 0.9627,
+      "step": 12325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1813214941239908e-05,
+      "loss": 1.0738,
+      "step": 12326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1811846713443978e-05,
+      "loss": 1.0432,
+      "step": 12327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1810478489171606e-05,
+      "loss": 1.0526,
+      "step": 12328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1809110268448416e-05,
+      "loss": 1.0798,
+      "step": 12329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1807742051300026e-05,
+      "loss": 0.9944,
+      "step": 12330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1806373837752064e-05,
+      "loss": 0.968,
+      "step": 12331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1805005627830143e-05,
+      "loss": 1.0341,
+      "step": 12332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1803637421559898e-05,
+      "loss": 1.0331,
+      "step": 12333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1802269218966947e-05,
+      "loss": 1.0863,
+      "step": 12334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1800901020076905e-05,
+      "loss": 1.0053,
+      "step": 12335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1799532824915406e-05,
+      "loss": 0.9645,
+      "step": 12336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.179816463350806e-05,
+      "loss": 0.9852,
+      "step": 12337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1796796445880504e-05,
+      "loss": 0.9697,
+      "step": 12338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1795428262058349e-05,
+      "loss": 1.0449,
+      "step": 12339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.179406008206722e-05,
+      "loss": 0.9966,
+      "step": 12340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.179269190593274e-05,
+      "loss": 1.0343,
+      "step": 12341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1791323733680526e-05,
+      "loss": 1.0794,
+      "step": 12342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1789955565336209e-05,
+      "loss": 1.0356,
+      "step": 12343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1788587400925404e-05,
+      "loss": 1.0286,
+      "step": 12344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1787219240473731e-05,
+      "loss": 1.0489,
+      "step": 12345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1785851084006818e-05,
+      "loss": 1.0415,
+      "step": 12346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1784482931550282e-05,
+      "loss": 0.9789,
+      "step": 12347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1783114783129748e-05,
+      "loss": 1.0341,
+      "step": 12348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1781746638770834e-05,
+      "loss": 0.9971,
+      "step": 12349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1780378498499162e-05,
+      "loss": 1.0447,
+      "step": 12350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1779010362340355e-05,
+      "loss": 0.9837,
+      "step": 12351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.177764223032003e-05,
+      "loss": 0.983,
+      "step": 12352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1776274102463815e-05,
+      "loss": 0.9691,
+      "step": 12353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1774905978797327e-05,
+      "loss": 1.0534,
+      "step": 12354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1773537859346182e-05,
+      "loss": 1.0125,
+      "step": 12355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.177216974413601e-05,
+      "loss": 0.9765,
+      "step": 12356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1770801633192428e-05,
+      "loss": 1.0131,
+      "step": 12357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1769433526541057e-05,
+      "loss": 0.9602,
+      "step": 12358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1768065424207518e-05,
+      "loss": 1.0744,
+      "step": 12359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1766697326217423e-05,
+      "loss": 0.9931,
+      "step": 12360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1765329232596408e-05,
+      "loss": 1.0194,
+      "step": 12361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1763961143370082e-05,
+      "loss": 1.0013,
+      "step": 12362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1762593058564067e-05,
+      "loss": 1.0014,
+      "step": 12363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1761224978203987e-05,
+      "loss": 0.9146,
+      "step": 12364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1759856902315458e-05,
+      "loss": 1.0109,
+      "step": 12365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1758488830924103e-05,
+      "loss": 0.9993,
+      "step": 12366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1757120764055542e-05,
+      "loss": 0.9526,
+      "step": 12367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1755752701735393e-05,
+      "loss": 1.1059,
+      "step": 12368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1754384643989277e-05,
+      "loss": 1.0274,
+      "step": 12369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.175301659084281e-05,
+      "loss": 0.9705,
+      "step": 12370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1751648542321619e-05,
+      "loss": 0.9444,
+      "step": 12371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1750280498451319e-05,
+      "loss": 1.0336,
+      "step": 12372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1748912459257525e-05,
+      "loss": 1.0348,
+      "step": 12373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1747544424765865e-05,
+      "loss": 1.0723,
+      "step": 12374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1746176395001954e-05,
+      "loss": 0.9546,
+      "step": 12375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.174480836999141e-05,
+      "loss": 0.9505,
+      "step": 12376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1743440349759856e-05,
+      "loss": 1.0056,
+      "step": 12377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1742072334332907e-05,
+      "loss": 0.9983,
+      "step": 12378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1740704323736184e-05,
+      "loss": 1.0139,
+      "step": 12379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1739336317995303e-05,
+      "loss": 1.0167,
+      "step": 12380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1737968317135887e-05,
+      "loss": 0.9596,
+      "step": 12381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1736600321183556e-05,
+      "loss": 1.0595,
+      "step": 12382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1735232330163919e-05,
+      "loss": 1.0025,
+      "step": 12383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1733864344102604e-05,
+      "loss": 0.9723,
+      "step": 12384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1732496363025225e-05,
+      "loss": 1.0446,
+      "step": 12385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1731128386957404e-05,
+      "loss": 1.0532,
+      "step": 12386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1729760415924752e-05,
+      "loss": 0.9867,
+      "step": 12387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1728392449952893e-05,
+      "loss": 1.0765,
+      "step": 12388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1727024489067446e-05,
+      "loss": 0.9847,
+      "step": 12389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.172565653329402e-05,
+      "loss": 0.9465,
+      "step": 12390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1724288582658245e-05,
+      "loss": 1.0162,
+      "step": 12391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1722920637185733e-05,
+      "loss": 1.0494,
+      "step": 12392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1721552696902095e-05,
+      "loss": 1.0086,
+      "step": 12393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.172018476183296e-05,
+      "loss": 1.0456,
+      "step": 12394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1718816832003939e-05,
+      "loss": 1.0122,
+      "step": 12395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1717448907440654e-05,
+      "loss": 1.0154,
+      "step": 12396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1716080988168716e-05,
+      "loss": 0.9504,
+      "step": 12397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.171471307421374e-05,
+      "loss": 1.0261,
+      "step": 12398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1713345165601355e-05,
+      "loss": 1.0202,
+      "step": 12399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1711977262357166e-05,
+      "loss": 1.001,
+      "step": 12400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1710609364506797e-05,
+      "loss": 1.0221,
+      "step": 12401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1709241472075863e-05,
+      "loss": 1.1042,
+      "step": 12402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1707873585089976e-05,
+      "loss": 0.9764,
+      "step": 12403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.170650570357476e-05,
+      "loss": 1.0465,
+      "step": 12404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1705137827555826e-05,
+      "loss": 0.976,
+      "step": 12405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1703769957058795e-05,
+      "loss": 0.996,
+      "step": 12406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1702402092109279e-05,
+      "loss": 0.9758,
+      "step": 12407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1701034232732892e-05,
+      "loss": 1.0723,
+      "step": 12408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1699666378955257e-05,
+      "loss": 0.9719,
+      "step": 12409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.169829853080198e-05,
+      "loss": 1.0078,
+      "step": 12410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.169693068829869e-05,
+      "loss": 1.0323,
+      "step": 12411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1695562851470993e-05,
+      "loss": 1.0104,
+      "step": 12412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1694195020344507e-05,
+      "loss": 0.9772,
+      "step": 12413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.169282719494485e-05,
+      "loss": 1.0075,
+      "step": 12414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1691459375297628e-05,
+      "loss": 1.0259,
+      "step": 12415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.169009156142847e-05,
+      "loss": 1.0297,
+      "step": 12416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1688723753362982e-05,
+      "loss": 1.0433,
+      "step": 12417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1687355951126779e-05,
+      "loss": 1.0405,
+      "step": 12418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1685988154745481e-05,
+      "loss": 0.9939,
+      "step": 12419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1684620364244696e-05,
+      "loss": 0.9837,
+      "step": 12420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1683252579650047e-05,
+      "loss": 1.0206,
+      "step": 12421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1681884800987146e-05,
+      "loss": 1.0053,
+      "step": 12422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1680517028281601e-05,
+      "loss": 0.9714,
+      "step": 12423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1679149261559035e-05,
+      "loss": 1.034,
+      "step": 12424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.1677781500845052e-05,
+      "loss": 1.0425,
+      "step": 12425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1676413746165278e-05,
+      "loss": 0.9776,
+      "step": 12426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1675045997545322e-05,
+      "loss": 0.9291,
+      "step": 12427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1673678255010791e-05,
+      "loss": 1.026,
+      "step": 12428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1672310518587312e-05,
+      "loss": 1.064,
+      "step": 12429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1670942788300486e-05,
+      "loss": 1.0168,
+      "step": 12430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1669575064175937e-05,
+      "loss": 1.0106,
+      "step": 12431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1668207346239275e-05,
+      "loss": 1.012,
+      "step": 12432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1666839634516108e-05,
+      "loss": 1.0052,
+      "step": 12433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1665471929032056e-05,
+      "loss": 1.0046,
+      "step": 12434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.166410422981273e-05,
+      "loss": 1.0048,
+      "step": 12435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1662736536883737e-05,
+      "loss": 1.0294,
+      "step": 12436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1661368850270701e-05,
+      "loss": 1.0061,
+      "step": 12437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1660001169999224e-05,
+      "loss": 1.0301,
+      "step": 12438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1658633496094928e-05,
+      "loss": 1.0534,
+      "step": 12439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1657265828583421e-05,
+      "loss": 1.0463,
+      "step": 12440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1655898167490315e-05,
+      "loss": 1.0226,
+      "step": 12441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1654530512841221e-05,
+      "loss": 1.0256,
+      "step": 12442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1653162864661753e-05,
+      "loss": 0.9784,
+      "step": 12443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1651795222977526e-05,
+      "loss": 0.9905,
+      "step": 12444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1650427587814148e-05,
+      "loss": 1.0838,
+      "step": 12445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1649059959197227e-05,
+      "loss": 0.9785,
+      "step": 12446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1647692337152385e-05,
+      "loss": 1.0101,
+      "step": 12447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1646324721705224e-05,
+      "loss": 1.0149,
+      "step": 12448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.164495711288136e-05,
+      "loss": 0.9781,
+      "step": 12449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1643589510706406e-05,
+      "loss": 1.0249,
+      "step": 12450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1642221915205968e-05,
+      "loss": 0.9655,
+      "step": 12451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1640854326405664e-05,
+      "loss": 0.9376,
+      "step": 12452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1639486744331094e-05,
+      "loss": 1.05,
+      "step": 12453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.163811916900788e-05,
+      "loss": 1.1362,
+      "step": 12454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1636751600461629e-05,
+      "loss": 1.054,
+      "step": 12455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1635384038717945e-05,
+      "loss": 0.9843,
+      "step": 12456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.163401648380245e-05,
+      "loss": 1.0195,
+      "step": 12457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1632648935740742e-05,
+      "loss": 0.9825,
+      "step": 12458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1631281394558443e-05,
+      "loss": 0.9813,
+      "step": 12459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1629913860281158e-05,
+      "loss": 1.003,
+      "step": 12460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1628546332934494e-05,
+      "loss": 1.0078,
+      "step": 12461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1627178812544064e-05,
+      "loss": 1.0585,
+      "step": 12462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1625811299135474e-05,
+      "loss": 1.0046,
+      "step": 12463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1624443792734338e-05,
+      "loss": 1.0496,
+      "step": 12464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1623076293366265e-05,
+      "loss": 0.9916,
+      "step": 12465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1621708801056859e-05,
+      "loss": 0.9925,
+      "step": 12466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1620341315831738e-05,
+      "loss": 1.0203,
+      "step": 12467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.16189738377165e-05,
+      "loss": 1.0485,
+      "step": 12468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1617606366736764e-05,
+      "loss": 1.0074,
+      "step": 12469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1616238902918136e-05,
+      "loss": 0.9955,
+      "step": 12470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1614871446286219e-05,
+      "loss": 1.0595,
+      "step": 12471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1613503996866628e-05,
+      "loss": 0.9878,
+      "step": 12472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1612136554684965e-05,
+      "loss": 1.0175,
+      "step": 12473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1610769119766846e-05,
+      "loss": 1.0539,
+      "step": 12474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1609401692137876e-05,
+      "loss": 1.0342,
+      "step": 12475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1608034271823657e-05,
+      "loss": 0.9678,
+      "step": 12476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1606666858849804e-05,
+      "loss": 1.066,
+      "step": 12477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1605299453241922e-05,
+      "loss": 1.0468,
+      "step": 12478
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.160393205502562e-05,
+      "loss": 0.9724,
+      "step": 12479
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1602564664226503e-05,
+      "loss": 0.9587,
+      "step": 12480
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.160119728087018e-05,
+      "loss": 1.0388,
+      "step": 12481
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1599829904982255e-05,
+      "loss": 1.0618,
+      "step": 12482
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1598462536588337e-05,
+      "loss": 1.0152,
+      "step": 12483
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1597095175714036e-05,
+      "loss": 1.0397,
+      "step": 12484
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1595727822384956e-05,
+      "loss": 1.0309,
+      "step": 12485
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1594360476626697e-05,
+      "loss": 1.0362,
+      "step": 12486
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1592993138464879e-05,
+      "loss": 1.0201,
+      "step": 12487
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1591625807925097e-05,
+      "loss": 1.0721,
+      "step": 12488
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1590258485032963e-05,
+      "loss": 0.996,
+      "step": 12489
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.158889116981408e-05,
+      "loss": 1.0047,
+      "step": 12490
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1587523862294052e-05,
+      "loss": 1.0294,
+      "step": 12491
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1586156562498489e-05,
+      "loss": 0.9454,
+      "step": 12492
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1584789270452991e-05,
+      "loss": 1.0145,
+      "step": 12493
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1583421986183173e-05,
+      "loss": 1.0019,
+      "step": 12494
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1582054709714632e-05,
+      "loss": 1.0497,
+      "step": 12495
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1580687441072973e-05,
+      "loss": 1.0831,
+      "step": 12496
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1579320180283806e-05,
+      "loss": 0.9781,
+      "step": 12497
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1577952927372732e-05,
+      "loss": 1.0654,
+      "step": 12498
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1576585682365357e-05,
+      "loss": 1.0423,
+      "step": 12499
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1575218445287287e-05,
+      "loss": 1.0275,
+      "step": 12500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.157385121616412e-05,
+      "loss": 1.0199,
+      "step": 12501
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1572483995021467e-05,
+      "loss": 0.9838,
+      "step": 12502
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.157111678188493e-05,
+      "loss": 1.001,
+      "step": 12503
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1569749576780109e-05,
+      "loss": 1.0218,
+      "step": 12504
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1568382379732616e-05,
+      "loss": 1.0081,
+      "step": 12505
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1567015190768046e-05,
+      "loss": 1.0154,
+      "step": 12506
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.156564800991201e-05,
+      "loss": 1.0314,
+      "step": 12507
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1564280837190103e-05,
+      "loss": 1.0163,
+      "step": 12508
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1562913672627933e-05,
+      "loss": 0.9808,
+      "step": 12509
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1561546516251106e-05,
+      "loss": 1.0541,
+      "step": 12510
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1560179368085217e-05,
+      "loss": 0.9779,
+      "step": 12511
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1558812228155876e-05,
+      "loss": 1.0022,
+      "step": 12512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1557445096488684e-05,
+      "loss": 0.9695,
+      "step": 12513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1556077973109236e-05,
+      "loss": 1.018,
+      "step": 12514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1554710858043146e-05,
+      "loss": 0.9939,
+      "step": 12515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1553343751316006e-05,
+      "loss": 0.9998,
+      "step": 12516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1551976652953426e-05,
+      "loss": 0.9204,
+      "step": 12517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1550609562981003e-05,
+      "loss": 1.0049,
+      "step": 12518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1549242481424334e-05,
+      "loss": 1.0275,
+      "step": 12519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1547875408309032e-05,
+      "loss": 1.0675,
+      "step": 12520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1546508343660685e-05,
+      "loss": 0.9867,
+      "step": 12521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1545141287504906e-05,
+      "loss": 1.0311,
+      "step": 12522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1543774239867293e-05,
+      "loss": 0.9799,
+      "step": 12523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1542407200773438e-05,
+      "loss": 0.9944,
+      "step": 12524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1541040170248953e-05,
+      "loss": 0.9954,
+      "step": 12525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1539673148319433e-05,
+      "loss": 1.0248,
+      "step": 12526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1538306135010481e-05,
+      "loss": 1.0319,
+      "step": 12527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1536939130347694e-05,
+      "loss": 0.8992,
+      "step": 12528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.153557213435667e-05,
+      "loss": 1.0319,
+      "step": 12529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1534205147063015e-05,
+      "loss": 0.9525,
+      "step": 12530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1532838168492322e-05,
+      "loss": 1.0494,
+      "step": 12531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1531471198670199e-05,
+      "loss": 1.001,
+      "step": 12532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.153010423762224e-05,
+      "loss": 1.0145,
+      "step": 12533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1528737285374043e-05,
+      "loss": 1.0483,
+      "step": 12534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1527370341951208e-05,
+      "loss": 1.0126,
+      "step": 12535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1526003407379335e-05,
+      "loss": 0.9827,
+      "step": 12536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1524636481684023e-05,
+      "loss": 1.0212,
+      "step": 12537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1523269564890869e-05,
+      "loss": 0.9797,
+      "step": 12538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1521902657025469e-05,
+      "loss": 1.0418,
+      "step": 12539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1520535758113429e-05,
+      "loss": 1.0371,
+      "step": 12540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1519168868180337e-05,
+      "loss": 1.0944,
+      "step": 12541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1517801987251799e-05,
+      "loss": 0.9629,
+      "step": 12542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1516435115353412e-05,
+      "loss": 0.8967,
+      "step": 12543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1515068252510766e-05,
+      "loss": 0.9845,
+      "step": 12544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1513701398749466e-05,
+      "loss": 1.0021,
+      "step": 12545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1512334554095103e-05,
+      "loss": 0.9884,
+      "step": 12546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1510967718573282e-05,
+      "loss": 1.1199,
+      "step": 12547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1509600892209596e-05,
+      "loss": 0.9932,
+      "step": 12548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1508234075029635e-05,
+      "loss": 1.0289,
+      "step": 12549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1506867267059007e-05,
+      "loss": 1.0527,
+      "step": 12550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1505500468323295e-05,
+      "loss": 0.9791,
+      "step": 12551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.150413367884811e-05,
+      "loss": 1.0195,
+      "step": 12552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1502766898659041e-05,
+      "loss": 1.013,
+      "step": 12553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1501400127781679e-05,
+      "loss": 1.0429,
+      "step": 12554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1500033366241629e-05,
+      "loss": 1.0603,
+      "step": 12555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1498666614064474e-05,
+      "loss": 0.9657,
+      "step": 12556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1497299871275825e-05,
+      "loss": 1.0096,
+      "step": 12557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1495933137901265e-05,
+      "loss": 1.0112,
+      "step": 12558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1494566413966391e-05,
+      "loss": 1.0253,
+      "step": 12559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.14931996994968e-05,
+      "loss": 0.9794,
+      "step": 12560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1491832994518087e-05,
+      "loss": 0.9953,
+      "step": 12561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1490466299055844e-05,
+      "loss": 1.0005,
+      "step": 12562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.148909961313567e-05,
+      "loss": 1.0016,
+      "step": 12563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1487732936783152e-05,
+      "loss": 1.0233,
+      "step": 12564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.148636627002389e-05,
+      "loss": 0.9593,
+      "step": 12565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1484999612883469e-05,
+      "loss": 0.9756,
+      "step": 12566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1483632965387494e-05,
+      "loss": 1.0026,
+      "step": 12567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1482266327561553e-05,
+      "loss": 1.0073,
+      "step": 12568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1480899699431233e-05,
+      "loss": 0.9522,
+      "step": 12569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1479533081022138e-05,
+      "loss": 1.0132,
+      "step": 12570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1478166472359856e-05,
+      "loss": 1.0655,
+      "step": 12571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1476799873469976e-05,
+      "loss": 1.0137,
+      "step": 12572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1475433284378094e-05,
+      "loss": 1.0067,
+      "step": 12573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1474066705109802e-05,
+      "loss": 1.0057,
+      "step": 12574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1472700135690691e-05,
+      "loss": 0.989,
+      "step": 12575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1471333576146354e-05,
+      "loss": 0.9936,
+      "step": 12576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146996702650238e-05,
+      "loss": 1.0101,
+      "step": 12577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1468600486784365e-05,
+      "loss": 1.0153,
+      "step": 12578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1467233957017893e-05,
+      "loss": 1.0714,
+      "step": 12579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1465867437228566e-05,
+      "loss": 1.0146,
+      "step": 12580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1464500927441966e-05,
+      "loss": 1.0057,
+      "step": 12581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1463134427683686e-05,
+      "loss": 1.0148,
+      "step": 12582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.146176793797932e-05,
+      "loss": 1.0225,
+      "step": 12583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1460401458354448e-05,
+      "loss": 1.001,
+      "step": 12584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1459034988834673e-05,
+      "loss": 0.991,
+      "step": 12585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.145766852944558e-05,
+      "loss": 1.1096,
+      "step": 12586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1456302080212752e-05,
+      "loss": 1.0496,
+      "step": 12587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1454935641161791e-05,
+      "loss": 0.99,
+      "step": 12588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1453569212318276e-05,
+      "loss": 1.009,
+      "step": 12589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1452202793707802e-05,
+      "loss": 1.0763,
+      "step": 12590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.145083638535596e-05,
+      "loss": 0.9657,
+      "step": 12591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.144946998728833e-05,
+      "loss": 1.0253,
+      "step": 12592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1448103599530507e-05,
+      "loss": 1.0413,
+      "step": 12593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1446737222108074e-05,
+      "loss": 1.0198,
+      "step": 12594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.144537085504663e-05,
+      "loss": 0.9828,
+      "step": 12595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1444004498371755e-05,
+      "loss": 0.9892,
+      "step": 12596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1442638152109032e-05,
+      "loss": 1.035,
+      "step": 12597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1441271816284061e-05,
+      "loss": 1.0083,
+      "step": 12598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.143990549092242e-05,
+      "loss": 1.1043,
+      "step": 12599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1438539176049704e-05,
+      "loss": 1.0169,
+      "step": 12600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.143717287169149e-05,
+      "loss": 1.0312,
+      "step": 12601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1435806577873373e-05,
+      "loss": 0.9706,
+      "step": 12602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1434440294620937e-05,
+      "loss": 1.0634,
+      "step": 12603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1433074021959766e-05,
+      "loss": 1.0011,
+      "step": 12604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1431707759915451e-05,
+      "loss": 0.921,
+      "step": 12605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1430341508513579e-05,
+      "loss": 0.9853,
+      "step": 12606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1428975267779726e-05,
+      "loss": 1.0347,
+      "step": 12607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1427609037739487e-05,
+      "loss": 0.9891,
+      "step": 12608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1426242818418445e-05,
+      "loss": 1.0408,
+      "step": 12609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1424876609842188e-05,
+      "loss": 1.0425,
+      "step": 12610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1423510412036296e-05,
+      "loss": 1.0348,
+      "step": 12611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1422144225026353e-05,
+      "loss": 0.9634,
+      "step": 12612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.142077804883795e-05,
+      "loss": 1.0071,
+      "step": 12613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1419411883496665e-05,
+      "loss": 1.0179,
+      "step": 12614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141804572902809e-05,
+      "loss": 1.0641,
+      "step": 12615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1416679585457802e-05,
+      "loss": 1.0352,
+      "step": 12616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1415313452811385e-05,
+      "loss": 1.0198,
+      "step": 12617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1413947331114428e-05,
+      "loss": 0.9993,
+      "step": 12618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.141258122039251e-05,
+      "loss": 1.0112,
+      "step": 12619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1411215120671217e-05,
+      "loss": 1.0359,
+      "step": 12620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.140984903197613e-05,
+      "loss": 0.9792,
+      "step": 12621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1408482954332831e-05,
+      "loss": 0.9318,
+      "step": 12622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1407116887766906e-05,
+      "loss": 1.0321,
+      "step": 12623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.140575083230393e-05,
+      "loss": 1.0004,
+      "step": 12624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1404384787969497e-05,
+      "loss": 0.9904,
+      "step": 12625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1403018754789182e-05,
+      "loss": 0.9892,
+      "step": 12626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1401652732788565e-05,
+      "loss": 1.0172,
+      "step": 12627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.140028672199323e-05,
+      "loss": 1.0174,
+      "step": 12628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.139892072242876e-05,
+      "loss": 0.9893,
+      "step": 12629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1397554734120735e-05,
+      "loss": 1.0163,
+      "step": 12630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1396188757094735e-05,
+      "loss": 1.0168,
+      "step": 12631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1394822791376337e-05,
+      "loss": 0.9812,
+      "step": 12632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.139345683699113e-05,
+      "loss": 1.0348,
+      "step": 12633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1392090893964686e-05,
+      "loss": 0.9918,
+      "step": 12634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1390724962322591e-05,
+      "loss": 1.0463,
+      "step": 12635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1389359042090427e-05,
+      "loss": 0.9653,
+      "step": 12636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1387993133293766e-05,
+      "loss": 1.068,
+      "step": 12637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1386627235958191e-05,
+      "loss": 0.944,
+      "step": 12638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.138526135010928e-05,
+      "loss": 1.0145,
+      "step": 12639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1383895475772615e-05,
+      "loss": 1.0185,
+      "step": 12640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1382529612973774e-05,
+      "loss": 1.0605,
+      "step": 12641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.138116376173833e-05,
+      "loss": 0.9788,
+      "step": 12642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.137979792209187e-05,
+      "loss": 0.9437,
+      "step": 12643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1378432094059968e-05,
+      "loss": 1.0008,
+      "step": 12644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.13770662776682e-05,
+      "loss": 0.999,
+      "step": 12645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1375700472942147e-05,
+      "loss": 1.0483,
+      "step": 12646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1374334679907386e-05,
+      "loss": 0.993,
+      "step": 12647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1372968898589495e-05,
+      "loss": 1.0248,
+      "step": 12648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.137160312901405e-05,
+      "loss": 1.0478,
+      "step": 12649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1370237371206622e-05,
+      "loss": 1.0257,
+      "step": 12650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1368871625192797e-05,
+      "loss": 0.9596,
+      "step": 12651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1367505890998144e-05,
+      "loss": 0.9826,
+      "step": 12652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1366140168648247e-05,
+      "loss": 0.9607,
+      "step": 12653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1364774458168678e-05,
+      "loss": 1.0119,
+      "step": 12654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1363408759585009e-05,
+      "loss": 1.0542,
+      "step": 12655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1362043072922821e-05,
+      "loss": 1.0799,
+      "step": 12656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1360677398207686e-05,
+      "loss": 0.9562,
+      "step": 12657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1359311735465184e-05,
+      "loss": 0.9801,
+      "step": 12658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1357946084720883e-05,
+      "loss": 0.9776,
+      "step": 12659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.135658044600036e-05,
+      "loss": 1.0094,
+      "step": 12660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.135521481932919e-05,
+      "loss": 0.9935,
+      "step": 12661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.1353849204732948e-05,
+      "loss": 0.9979,
+      "step": 12662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1352483602237207e-05,
+      "loss": 1.0141,
+      "step": 12663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1351118011867543e-05,
+      "loss": 1.008,
+      "step": 12664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1349752433649525e-05,
+      "loss": 1.0569,
+      "step": 12665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1348386867608729e-05,
+      "loss": 1.0116,
+      "step": 12666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1347021313770727e-05,
+      "loss": 1.0014,
+      "step": 12667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1345655772161095e-05,
+      "loss": 0.9983,
+      "step": 12668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1344290242805405e-05,
+      "loss": 0.9768,
+      "step": 12669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.134292472572922e-05,
+      "loss": 1.0609,
+      "step": 12670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1341559220958125e-05,
+      "loss": 0.97,
+      "step": 12671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1340193728517682e-05,
+      "loss": 1.0446,
+      "step": 12672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1338828248433471e-05,
+      "loss": 0.9718,
+      "step": 12673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1337462780731063e-05,
+      "loss": 1.0089,
+      "step": 12674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.133609732543602e-05,
+      "loss": 1.0267,
+      "step": 12675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1334731882573921e-05,
+      "loss": 1.0251,
+      "step": 12676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.133336645217033e-05,
+      "loss": 1.0151,
+      "step": 12677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1332001034250826e-05,
+      "loss": 1.0622,
+      "step": 12678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1330635628840975e-05,
+      "loss": 1.0454,
+      "step": 12679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1329270235966343e-05,
+      "loss": 1.0315,
+      "step": 12680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.132790485565251e-05,
+      "loss": 1.0272,
+      "step": 12681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1326539487925034e-05,
+      "loss": 1.0451,
+      "step": 12682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1325174132809494e-05,
+      "loss": 1.0457,
+      "step": 12683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1323808790331454e-05,
+      "loss": 1.0342,
+      "step": 12684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1322443460516483e-05,
+      "loss": 1.0459,
+      "step": 12685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1321078143390151e-05,
+      "loss": 1.0468,
+      "step": 12686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1319712838978024e-05,
+      "loss": 1.0009,
+      "step": 12687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1318347547305672e-05,
+      "loss": 0.9721,
+      "step": 12688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1316982268398665e-05,
+      "loss": 0.9835,
+      "step": 12689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1315617002282564e-05,
+      "loss": 1.0599,
+      "step": 12690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1314251748982945e-05,
+      "loss": 0.9566,
+      "step": 12691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1312886508525369e-05,
+      "loss": 1.0216,
+      "step": 12692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1311521280935407e-05,
+      "loss": 0.9939,
+      "step": 12693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.131015606623862e-05,
+      "loss": 0.9925,
+      "step": 12694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1308790864460581e-05,
+      "loss": 1.0094,
+      "step": 12695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1307425675626853e-05,
+      "loss": 1.0008,
+      "step": 12696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1306060499762999e-05,
+      "loss": 0.952,
+      "step": 12697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1304695336894592e-05,
+      "loss": 1.0177,
+      "step": 12698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1303330187047192e-05,
+      "loss": 0.993,
+      "step": 12699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.130196505024636e-05,
+      "loss": 1.0485,
+      "step": 12700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1300599926517674e-05,
+      "loss": 0.9752,
+      "step": 12701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1299234815886688e-05,
+      "loss": 0.9801,
+      "step": 12702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1297869718378974e-05,
+      "loss": 1.0066,
+      "step": 12703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.129650463402009e-05,
+      "loss": 0.9659,
+      "step": 12704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1295139562835599e-05,
+      "loss": 1.0411,
+      "step": 12705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1293774504851072e-05,
+      "loss": 1.0158,
+      "step": 12706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1292409460092066e-05,
+      "loss": 0.9664,
+      "step": 12707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1291044428584144e-05,
+      "loss": 0.9761,
+      "step": 12708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1289679410352877e-05,
+      "loss": 1.0084,
+      "step": 12709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1288314405423816e-05,
+      "loss": 1.0166,
+      "step": 12710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1286949413822537e-05,
+      "loss": 0.9781,
+      "step": 12711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1285584435574595e-05,
+      "loss": 0.9311,
+      "step": 12712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1284219470705548e-05,
+      "loss": 1.0018,
+      "step": 12713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1282854519240966e-05,
+      "loss": 1.0797,
+      "step": 12714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.12814895812064e-05,
+      "loss": 1.0135,
+      "step": 12715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1280124656627425e-05,
+      "loss": 1.0724,
+      "step": 12716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1278759745529593e-05,
+      "loss": 1.0443,
+      "step": 12717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1277394847938461e-05,
+      "loss": 1.1147,
+      "step": 12718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1276029963879602e-05,
+      "loss": 0.9982,
+      "step": 12719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1274665093378567e-05,
+      "loss": 0.9873,
+      "step": 12720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1273300236460917e-05,
+      "loss": 1.0086,
+      "step": 12721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1271935393152216e-05,
+      "loss": 1.0185,
+      "step": 12722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1270570563478019e-05,
+      "loss": 1.0615,
+      "step": 12723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.126920574746389e-05,
+      "loss": 1.0351,
+      "step": 12724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1267840945135377e-05,
+      "loss": 1.028,
+      "step": 12725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1266476156518054e-05,
+      "loss": 1.0013,
+      "step": 12726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.126511138163747e-05,
+      "loss": 1.0071,
+      "step": 12727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1263746620519182e-05,
+      "loss": 0.9452,
+      "step": 12728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1262381873188756e-05,
+      "loss": 0.9996,
+      "step": 12729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1261017139671743e-05,
+      "loss": 0.9994,
+      "step": 12730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1259652419993705e-05,
+      "loss": 0.9925,
+      "step": 12731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1258287714180191e-05,
+      "loss": 1.0861,
+      "step": 12732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1256923022256767e-05,
+      "loss": 1.0459,
+      "step": 12733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1255558344248987e-05,
+      "loss": 1.0116,
+      "step": 12734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.12541936801824e-05,
+      "loss": 1.0343,
+      "step": 12735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1252829030082576e-05,
+      "loss": 0.9805,
+      "step": 12736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1251464393975063e-05,
+      "loss": 1.0046,
+      "step": 12737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1250099771885412e-05,
+      "loss": 1.0337,
+      "step": 12738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1248735163839188e-05,
+      "loss": 0.9904,
+      "step": 12739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.124737056986194e-05,
+      "loss": 0.9494,
+      "step": 12740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1246005989979228e-05,
+      "loss": 1.0319,
+      "step": 12741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.12446414242166e-05,
+      "loss": 1.0769,
+      "step": 12742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.124327687259961e-05,
+      "loss": 1.0454,
+      "step": 12743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.124191233515382e-05,
+      "loss": 1.0074,
+      "step": 12744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1240547811904775e-05,
+      "loss": 1.0395,
+      "step": 12745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1239183302878036e-05,
+      "loss": 1.0146,
+      "step": 12746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1237818808099154e-05,
+      "loss": 1.0635,
+      "step": 12747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1236454327593676e-05,
+      "loss": 0.9732,
+      "step": 12748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1235089861387162e-05,
+      "loss": 1.0346,
+      "step": 12749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1233725409505162e-05,
+      "loss": 0.9714,
+      "step": 12750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1232360971973231e-05,
+      "loss": 1.0442,
+      "step": 12751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1230996548816916e-05,
+      "loss": 1.0476,
+      "step": 12752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1229632140061767e-05,
+      "loss": 1.0182,
+      "step": 12753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1228267745733343e-05,
+      "loss": 1.0443,
+      "step": 12754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1226903365857187e-05,
+      "loss": 1.0043,
+      "step": 12755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1225539000458858e-05,
+      "loss": 1.0688,
+      "step": 12756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1224174649563903e-05,
+      "loss": 0.9627,
+      "step": 12757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1222810313197869e-05,
+      "loss": 0.9834,
+      "step": 12758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1221445991386308e-05,
+      "loss": 0.9725,
+      "step": 12759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.122008168415477e-05,
+      "loss": 1.0197,
+      "step": 12760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.121871739152881e-05,
+      "loss": 0.9956,
+      "step": 12761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1217353113533969e-05,
+      "loss": 0.9708,
+      "step": 12762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1215988850195795e-05,
+      "loss": 1.0232,
+      "step": 12763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1214624601539848e-05,
+      "loss": 1.0536,
+      "step": 12764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.121326036759166e-05,
+      "loss": 1.0439,
+      "step": 12765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1211896148376797e-05,
+      "loss": 1.0251,
+      "step": 12766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1210531943920794e-05,
+      "loss": 1.0281,
+      "step": 12767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1209167754249201e-05,
+      "loss": 1.0049,
+      "step": 12768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1207803579387571e-05,
+      "loss": 1.0002,
+      "step": 12769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1206439419361442e-05,
+      "loss": 1.006,
+      "step": 12770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1205075274196368e-05,
+      "loss": 1.0009,
+      "step": 12771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1203711143917892e-05,
+      "loss": 1.0286,
+      "step": 12772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1202347028551558e-05,
+      "loss": 1.0189,
+      "step": 12773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1200982928122919e-05,
+      "loss": 1.011,
+      "step": 12774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1199618842657515e-05,
+      "loss": 0.9584,
+      "step": 12775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.119825477218089e-05,
+      "loss": 1.0215,
+      "step": 12776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1196890716718596e-05,
+      "loss": 0.9912,
+      "step": 12777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.119552667629617e-05,
+      "loss": 1.0227,
+      "step": 12778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1194162650939163e-05,
+      "loss": 1.0281,
+      "step": 12779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1192798640673116e-05,
+      "loss": 1.0077,
+      "step": 12780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1191434645523567e-05,
+      "loss": 1.0892,
+      "step": 12781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1190070665516071e-05,
+      "loss": 1.0141,
+      "step": 12782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1188706700676162e-05,
+      "loss": 0.9732,
+      "step": 12783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1187342751029391e-05,
+      "loss": 1.0097,
+      "step": 12784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1185978816601297e-05,
+      "loss": 0.9508,
+      "step": 12785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.118461489741742e-05,
+      "loss": 1.0052,
+      "step": 12786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1183250993503302e-05,
+      "loss": 1.0697,
+      "step": 12787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.118188710488449e-05,
+      "loss": 0.9222,
+      "step": 12788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1180523231586525e-05,
+      "loss": 1.0363,
+      "step": 12789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1179159373634945e-05,
+      "loss": 0.9914,
+      "step": 12790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1177795531055287e-05,
+      "loss": 0.928,
+      "step": 12791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1176431703873104e-05,
+      "loss": 0.9913,
+      "step": 12792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1175067892113922e-05,
+      "loss": 0.9996,
+      "step": 12793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1173704095803294e-05,
+      "loss": 0.946,
+      "step": 12794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1172340314966754e-05,
+      "loss": 0.9923,
+      "step": 12795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1170976549629842e-05,
+      "loss": 1.0023,
+      "step": 12796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1169612799818098e-05,
+      "loss": 1.0209,
+      "step": 12797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1168249065557057e-05,
+      "loss": 1.0058,
+      "step": 12798
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1166885346872264e-05,
+      "loss": 1.0196,
+      "step": 12799
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1165521643789256e-05,
+      "loss": 0.9688,
+      "step": 12800
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1164157956333565e-05,
+      "loss": 1.0145,
+      "step": 12801
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1162794284530738e-05,
+      "loss": 0.9959,
+      "step": 12802
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1161430628406303e-05,
+      "loss": 1.044,
+      "step": 12803
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1160066987985808e-05,
+      "loss": 1.0018,
+      "step": 12804
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1158703363294784e-05,
+      "loss": 0.9832,
+      "step": 12805
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1157339754358765e-05,
+      "loss": 0.9954,
+      "step": 12806
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1155976161203293e-05,
+      "loss": 1.0091,
+      "step": 12807
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1154612583853898e-05,
+      "loss": 0.9875,
+      "step": 12808
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1153249022336123e-05,
+      "loss": 0.9986,
+      "step": 12809
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1151885476675499e-05,
+      "loss": 1.0121,
+      "step": 12810
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.115052194689756e-05,
+      "loss": 0.9379,
+      "step": 12811
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1149158433027844e-05,
+      "loss": 1.0133,
+      "step": 12812
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1147794935091884e-05,
+      "loss": 0.9768,
+      "step": 12813
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1146431453115215e-05,
+      "loss": 1.0124,
+      "step": 12814
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1145067987123371e-05,
+      "loss": 1.0488,
+      "step": 12815
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1143704537141885e-05,
+      "loss": 1.038,
+      "step": 12816
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1142341103196295e-05,
+      "loss": 0.9861,
+      "step": 12817
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1140977685312122e-05,
+      "loss": 1.0254,
+      "step": 12818
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1139614283514913e-05,
+      "loss": 1.0248,
+      "step": 12819
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1138250897830194e-05,
+      "loss": 1.0287,
+      "step": 12820
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1136887528283493e-05,
+      "loss": 0.9859,
+      "step": 12821
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1135524174900351e-05,
+      "loss": 1.0587,
+      "step": 12822
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1134160837706293e-05,
+      "loss": 1.0417,
+      "step": 12823
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1132797516726854e-05,
+      "loss": 1.0052,
+      "step": 12824
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.113143421198756e-05,
+      "loss": 0.9786,
+      "step": 12825
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1130070923513947e-05,
+      "loss": 0.9618,
+      "step": 12826
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1128707651331545e-05,
+      "loss": 1.0078,
+      "step": 12827
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1127344395465876e-05,
+      "loss": 1.0275,
+      "step": 12828
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1125981155942482e-05,
+      "loss": 1.0265,
+      "step": 12829
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1124617932786888e-05,
+      "loss": 1.0602,
+      "step": 12830
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1123254726024614e-05,
+      "loss": 1.0003,
+      "step": 12831
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1121891535681203e-05,
+      "loss": 0.9945,
+      "step": 12832
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1120528361782176e-05,
+      "loss": 0.9967,
+      "step": 12833
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1119165204353063e-05,
+      "loss": 0.9967,
+      "step": 12834
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1117802063419392e-05,
+      "loss": 0.9605,
+      "step": 12835
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1116438939006685e-05,
+      "loss": 1.0164,
+      "step": 12836
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1115075831140478e-05,
+      "loss": 1.0133,
+      "step": 12837
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.111371273984629e-05,
+      "loss": 1.0177,
+      "step": 12838
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1112349665149657e-05,
+      "loss": 1.0423,
+      "step": 12839
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1110986607076102e-05,
+      "loss": 0.99,
+      "step": 12840
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1109623565651142e-05,
+      "loss": 0.9807,
+      "step": 12841
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1108260540900316e-05,
+      "loss": 1.0511,
+      "step": 12842
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1106897532849144e-05,
+      "loss": 1.0215,
+      "step": 12843
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1105534541523146e-05,
+      "loss": 1.0136,
+      "step": 12844
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1104171566947856e-05,
+      "loss": 0.9632,
+      "step": 12845
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1102808609148789e-05,
+      "loss": 0.9793,
+      "step": 12846
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1101445668151478e-05,
+      "loss": 1.0082,
+      "step": 12847
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1100082743981443e-05,
+      "loss": 1.0376,
+      "step": 12848
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1098719836664205e-05,
+      "loss": 1.0374,
+      "step": 12849
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1097356946225292e-05,
+      "loss": 0.9338,
+      "step": 12850
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1095994072690223e-05,
+      "loss": 0.9969,
+      "step": 12851
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1094631216084522e-05,
+      "loss": 1.0169,
+      "step": 12852
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1093268376433714e-05,
+      "loss": 1.0086,
+      "step": 12853
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1091905553763317e-05,
+      "loss": 1.0409,
+      "step": 12854
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1090542748098857e-05,
+      "loss": 1.0129,
+      "step": 12855
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1089179959465847e-05,
+      "loss": 0.9955,
+      "step": 12856
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.108781718788982e-05,
+      "loss": 1.1087,
+      "step": 12857
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1086454433396288e-05,
+      "loss": 1.0227,
+      "step": 12858
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.108509169601077e-05,
+      "loss": 1.0676,
+      "step": 12859
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1083728975758794e-05,
+      "loss": 0.9982,
+      "step": 12860
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1082366272665873e-05,
+      "loss": 1.0226,
+      "step": 12861
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1081003586757533e-05,
+      "loss": 1.0044,
+      "step": 12862
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1079640918059288e-05,
+      "loss": 1.0341,
+      "step": 12863
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1078278266596653e-05,
+      "loss": 1.0577,
+      "step": 12864
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1076915632395156e-05,
+      "loss": 0.9814,
+      "step": 12865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1075553015480304e-05,
+      "loss": 1.0249,
+      "step": 12866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1074190415877628e-05,
+      "loss": 0.9955,
+      "step": 12867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1072827833612637e-05,
+      "loss": 1.0499,
+      "step": 12868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1071465268710845e-05,
+      "loss": 0.9956,
+      "step": 12869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.107010272119778e-05,
+      "loss": 0.9401,
+      "step": 12870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.106874019109895e-05,
+      "loss": 0.992,
+      "step": 12871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1067377678439873e-05,
+      "loss": 1.0293,
+      "step": 12872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1066015183246067e-05,
+      "loss": 0.9663,
+      "step": 12873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.106465270554304e-05,
+      "loss": 1.0152,
+      "step": 12874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.106329024535632e-05,
+      "loss": 0.9832,
+      "step": 12875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1061927802711407e-05,
+      "loss": 0.9932,
+      "step": 12876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1060565377633829e-05,
+      "loss": 1.0054,
+      "step": 12877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1059202970149094e-05,
+      "loss": 0.9784,
+      "step": 12878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1057840580282715e-05,
+      "loss": 0.9798,
+      "step": 12879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1056478208060206e-05,
+      "loss": 1.0707,
+      "step": 12880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1055115853507082e-05,
+      "loss": 1.0342,
+      "step": 12881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1053753516648858e-05,
+      "loss": 0.9975,
+      "step": 12882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.105239119751104e-05,
+      "loss": 0.9416,
+      "step": 12883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1051028896119141e-05,
+      "loss": 0.9698,
+      "step": 12884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1049666612498683e-05,
+      "loss": 1.0409,
+      "step": 12885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1048304346675162e-05,
+      "loss": 1.075,
+      "step": 12886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1046942098674103e-05,
+      "loss": 1.0364,
+      "step": 12887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1045579868521012e-05,
+      "loss": 0.9965,
+      "step": 12888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1044217656241397e-05,
+      "loss": 0.9438,
+      "step": 12889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1042855461860772e-05,
+      "loss": 0.9919,
+      "step": 12890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1041493285404641e-05,
+      "loss": 0.9943,
+      "step": 12891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.104013112689852e-05,
+      "loss": 1.0748,
+      "step": 12892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1038768986367916e-05,
+      "loss": 1.05,
+      "step": 12893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1037406863838335e-05,
+      "loss": 1.0331,
+      "step": 12894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1036044759335292e-05,
+      "loss": 1.0619,
+      "step": 12895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1034682672884287e-05,
+      "loss": 1.0478,
+      "step": 12896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1033320604510838e-05,
+      "loss": 0.9842,
+      "step": 12897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1031958554240444e-05,
+      "loss": 0.9723,
+      "step": 12898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.1030596522098615e-05,
+      "loss": 0.9571,
+      "step": 12899
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.102923450811086e-05,
+      "loss": 1.0132,
+      "step": 12900
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1027872512302676e-05,
+      "loss": 0.9871,
+      "step": 12901
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1026510534699583e-05,
+      "loss": 1.0321,
+      "step": 12902
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1025148575327082e-05,
+      "loss": 0.9832,
+      "step": 12903
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.102378663421067e-05,
+      "loss": 1.0184,
+      "step": 12904
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1022424711375864e-05,
+      "loss": 1.0257,
+      "step": 12905
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1021062806848162e-05,
+      "loss": 1.0377,
+      "step": 12906
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.101970092065307e-05,
+      "loss": 0.9569,
+      "step": 12907
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1018339052816095e-05,
+      "loss": 0.9752,
+      "step": 12908
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1016977203362734e-05,
+      "loss": 1.0116,
+      "step": 12909
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.10156153723185e-05,
+      "loss": 0.9889,
+      "step": 12910
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1014253559708884e-05,
+      "loss": 1.0121,
+      "step": 12911
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.10128917655594e-05,
+      "loss": 1.0612,
+      "step": 12912
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1011529989895546e-05,
+      "loss": 1.0201,
+      "step": 12913
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.101016823274282e-05,
+      "loss": 1.0532,
+      "step": 12914
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1008806494126733e-05,
+      "loss": 1.0308,
+      "step": 12915
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1007444774072778e-05,
+      "loss": 1.0497,
+      "step": 12916
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1006083072606458e-05,
+      "loss": 1.0356,
+      "step": 12917
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1004721389753275e-05,
+      "loss": 0.9795,
+      "step": 12918
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.1003359725538729e-05,
+      "loss": 1.024,
+      "step": 12919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.100199807998832e-05,
+      "loss": 1.0327,
+      "step": 12920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.100063645312755e-05,
+      "loss": 0.9615,
+      "step": 12921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.099927484498191e-05,
+      "loss": 0.9602,
+      "step": 12922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0997913255576909e-05,
+      "loss": 0.9836,
+      "step": 12923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0996551684938038e-05,
+      "loss": 1.0153,
+      "step": 12924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0995190133090801e-05,
+      "loss": 1.0167,
+      "step": 12925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0993828600060694e-05,
+      "loss": 1.017,
+      "step": 12926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0992467085873211e-05,
+      "loss": 0.9482,
+      "step": 12927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0991105590553855e-05,
+      "loss": 1.0205,
+      "step": 12928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0989744114128116e-05,
+      "loss": 1.1246,
+      "step": 12929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0988382656621496e-05,
+      "loss": 0.9675,
+      "step": 12930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0987021218059492e-05,
+      "loss": 0.9681,
+      "step": 12931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.098565979846759e-05,
+      "loss": 1.0565,
+      "step": 12932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0984298397871298e-05,
+      "loss": 1.0575,
+      "step": 12933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0982937016296101e-05,
+      "loss": 0.9884,
+      "step": 12934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0981575653767501e-05,
+      "loss": 0.9678,
+      "step": 12935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0980214310310991e-05,
+      "loss": 0.986,
+      "step": 12936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0978852985952061e-05,
+      "loss": 0.989,
+      "step": 12937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0977491680716209e-05,
+      "loss": 1.0387,
+      "step": 12938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.097613039462892e-05,
+      "loss": 1.0141,
+      "step": 12939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.09747691277157e-05,
+      "loss": 1.0362,
+      "step": 12940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0973407880002032e-05,
+      "loss": 0.994,
+      "step": 12941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0972046651513408e-05,
+      "loss": 1.0167,
+      "step": 12942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0970685442275327e-05,
+      "loss": 1.0238,
+      "step": 12943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0969324252313272e-05,
+      "loss": 0.9963,
+      "step": 12944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.096796308165274e-05,
+      "loss": 1.0597,
+      "step": 12945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.096660193031922e-05,
+      "loss": 1.0218,
+      "step": 12946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0965240798338201e-05,
+      "loss": 0.9856,
+      "step": 12947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0963879685735176e-05,
+      "loss": 0.9978,
+      "step": 12948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0962518592535628e-05,
+      "loss": 1.039,
+      "step": 12949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0961157518765056e-05,
+      "loss": 1.0142,
+      "step": 12950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0959796464448943e-05,
+      "loss": 1.0032,
+      "step": 12951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0958435429612775e-05,
+      "loss": 1.0212,
+      "step": 12952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0957074414282049e-05,
+      "loss": 0.9601,
+      "step": 12953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0955713418482243e-05,
+      "loss": 0.9743,
+      "step": 12954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0954352442238851e-05,
+      "loss": 1.0329,
+      "step": 12955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0952991485577358e-05,
+      "loss": 1.0088,
+      "step": 12956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0951630548523247e-05,
+      "loss": 1.0174,
+      "step": 12957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0950269631102012e-05,
+      "loss": 1.0356,
+      "step": 12958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0948908733339131e-05,
+      "loss": 0.9695,
+      "step": 12959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0947547855260097e-05,
+      "loss": 0.9828,
+      "step": 12960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0946186996890393e-05,
+      "loss": 1.0263,
+      "step": 12961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0944826158255497e-05,
+      "loss": 1.0286,
+      "step": 12962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0943465339380905e-05,
+      "loss": 1.0039,
+      "step": 12963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.094210454029209e-05,
+      "loss": 1.0394,
+      "step": 12964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0940743761014547e-05,
+      "loss": 1.074,
+      "step": 12965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093938300157375e-05,
+      "loss": 1.0142,
+      "step": 12966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0938022261995184e-05,
+      "loss": 1.0093,
+      "step": 12967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0936661542304337e-05,
+      "loss": 0.9541,
+      "step": 12968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0935300842526681e-05,
+      "loss": 0.9738,
+      "step": 12969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.093394016268771e-05,
+      "loss": 0.9546,
+      "step": 12970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.09325795028129e-05,
+      "loss": 1.0212,
+      "step": 12971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0931218862927728e-05,
+      "loss": 1.0586,
+      "step": 12972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.092985824305768e-05,
+      "loss": 0.9889,
+      "step": 12973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0928497643228235e-05,
+      "loss": 1.0046,
+      "step": 12974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0927137063464876e-05,
+      "loss": 1.039,
+      "step": 12975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.092577650379308e-05,
+      "loss": 1.0137,
+      "step": 12976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.092441596423832e-05,
+      "loss": 1.0213,
+      "step": 12977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0923055444826085e-05,
+      "loss": 1.0305,
+      "step": 12978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0921694945581847e-05,
+      "loss": 0.9958,
+      "step": 12979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0920334466531091e-05,
+      "loss": 1.0678,
+      "step": 12980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.091897400769929e-05,
+      "loss": 1.1149,
+      "step": 12981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0917613569111917e-05,
+      "loss": 1.0538,
+      "step": 12982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0916253150794459e-05,
+      "loss": 1.0061,
+      "step": 12983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0914892752772383e-05,
+      "loss": 1.0435,
+      "step": 12984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0913532375071172e-05,
+      "loss": 1.0152,
+      "step": 12985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0912172017716301e-05,
+      "loss": 0.9979,
+      "step": 12986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0910811680733238e-05,
+      "loss": 1.0318,
+      "step": 12987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0909451364147469e-05,
+      "loss": 1.0346,
+      "step": 12988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0908091067984465e-05,
+      "loss": 0.9932,
+      "step": 12989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0906730792269694e-05,
+      "loss": 0.9717,
+      "step": 12990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0905370537028642e-05,
+      "loss": 1.0344,
+      "step": 12991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.090401030228677e-05,
+      "loss": 0.997,
+      "step": 12992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0902650088069561e-05,
+      "loss": 1.0217,
+      "step": 12993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0901289894402482e-05,
+      "loss": 1.0073,
+      "step": 12994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0899929721311004e-05,
+      "loss": 0.9802,
+      "step": 12995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0898569568820606e-05,
+      "loss": 0.9832,
+      "step": 12996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0897209436956753e-05,
+      "loss": 0.9989,
+      "step": 12997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0895849325744921e-05,
+      "loss": 0.9659,
+      "step": 12998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.089448923521058e-05,
+      "loss": 1.0237,
+      "step": 12999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0893129165379195e-05,
+      "loss": 1.084,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "eval_text_loss": 0.9526932239532471,
+      "eval_text_runtime": 1.5994,
+      "eval_text_samples_per_second": 46.269,
+      "eval_text_steps_per_second": 1.251,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "eval_svit_loss": 1.1044390201568604,
+      "eval_svit_runtime": 3.805,
+      "eval_svit_samples_per_second": 178.974,
+      "eval_svit_steps_per_second": 2.891,
+      "step": 13000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0891769116276246e-05,
+      "loss": 0.9819,
+      "step": 13001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0890409087927195e-05,
+      "loss": 1.0102,
+      "step": 13002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0889049080357514e-05,
+      "loss": 1.0453,
+      "step": 13003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0887689093592672e-05,
+      "loss": 1.0052,
+      "step": 13004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0886329127658133e-05,
+      "loss": 1.0905,
+      "step": 13005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0884969182579375e-05,
+      "loss": 1.057,
+      "step": 13006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0883609258381854e-05,
+      "loss": 0.9892,
+      "step": 13007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0882249355091048e-05,
+      "loss": 0.9936,
+      "step": 13008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.088088947273242e-05,
+      "loss": 0.9793,
+      "step": 13009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0879529611331433e-05,
+      "loss": 0.9855,
+      "step": 13010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0878169770913556e-05,
+      "loss": 1.0349,
+      "step": 13011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0876809951504255e-05,
+      "loss": 0.9905,
+      "step": 13012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0875450153128997e-05,
+      "loss": 1.0476,
+      "step": 13013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0874090375813246e-05,
+      "loss": 0.9474,
+      "step": 13014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0872730619582463e-05,
+      "loss": 1.0117,
+      "step": 13015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0871370884462117e-05,
+      "loss": 1.0089,
+      "step": 13016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0870011170477667e-05,
+      "loss": 1.0101,
+      "step": 13017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0868651477654585e-05,
+      "loss": 0.9955,
+      "step": 13018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0867291806018327e-05,
+      "loss": 0.9465,
+      "step": 13019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0865932155594355e-05,
+      "loss": 0.95,
+      "step": 13020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0864572526408137e-05,
+      "loss": 0.9676,
+      "step": 13021
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0863212918485126e-05,
+      "loss": 0.9565,
+      "step": 13022
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0861853331850792e-05,
+      "loss": 1.0575,
+      "step": 13023
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0860493766530593e-05,
+      "loss": 1.0226,
+      "step": 13024
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0859134222549986e-05,
+      "loss": 0.9687,
+      "step": 13025
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.085777469993444e-05,
+      "loss": 0.9417,
+      "step": 13026
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0856415198709404e-05,
+      "loss": 1.0217,
+      "step": 13027
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0855055718900346e-05,
+      "loss": 1.0092,
+      "step": 13028
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0853696260532725e-05,
+      "loss": 1.0089,
+      "step": 13029
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0852336823631993e-05,
+      "loss": 0.9865,
+      "step": 13030
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0850977408223613e-05,
+      "loss": 0.9598,
+      "step": 13031
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0849618014333039e-05,
+      "loss": 0.9573,
+      "step": 13032
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0848258641985735e-05,
+      "loss": 1.0229,
+      "step": 13033
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0846899291207154e-05,
+      "loss": 0.9473,
+      "step": 13034
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0845539962022748e-05,
+      "loss": 1.059,
+      "step": 13035
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0844180654457985e-05,
+      "loss": 0.973,
+      "step": 13036
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0842821368538309e-05,
+      "loss": 0.9833,
+      "step": 13037
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0841462104289182e-05,
+      "loss": 1.0599,
+      "step": 13038
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0840102861736059e-05,
+      "loss": 1.0072,
+      "step": 13039
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.083874364090439e-05,
+      "loss": 0.9881,
+      "step": 13040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0837384441819636e-05,
+      "loss": 1.0186,
+      "step": 13041
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0836025264507243e-05,
+      "loss": 0.9994,
+      "step": 13042
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0834666108992671e-05,
+      "loss": 0.9994,
+      "step": 13043
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0833306975301372e-05,
+      "loss": 0.9809,
+      "step": 13044
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0831947863458792e-05,
+      "loss": 1.0396,
+      "step": 13045
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.083058877349039e-05,
+      "loss": 0.9899,
+      "step": 13046
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0829229705421618e-05,
+      "loss": 0.943,
+      "step": 13047
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0827870659277925e-05,
+      "loss": 1.0082,
+      "step": 13048
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0826511635084762e-05,
+      "loss": 1.023,
+      "step": 13049
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0825152632867576e-05,
+      "loss": 1.0066,
+      "step": 13050
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0823793652651826e-05,
+      "loss": 1.0355,
+      "step": 13051
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0822434694462955e-05,
+      "loss": 1.0484,
+      "step": 13052
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.082107575832641e-05,
+      "loss": 1.0185,
+      "step": 13053
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0819716844267647e-05,
+      "loss": 1.0059,
+      "step": 13054
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0818357952312107e-05,
+      "loss": 1.0452,
+      "step": 13055
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0816999082485246e-05,
+      "loss": 1.0365,
+      "step": 13056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.081564023481251e-05,
+      "loss": 1.036,
+      "step": 13057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.081428140931934e-05,
+      "loss": 1.0033,
+      "step": 13058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0812922606031189e-05,
+      "loss": 0.9492,
+      "step": 13059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0811563824973498e-05,
+      "loss": 1.0272,
+      "step": 13060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0810205066171719e-05,
+      "loss": 0.9729,
+      "step": 13061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0808846329651296e-05,
+      "loss": 1.0758,
+      "step": 13062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0807487615437667e-05,
+      "loss": 1.0333,
+      "step": 13063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0806128923556289e-05,
+      "loss": 0.961,
+      "step": 13064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0804770254032594e-05,
+      "loss": 1.0136,
+      "step": 13065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0803411606892036e-05,
+      "loss": 1.0112,
+      "step": 13066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0802052982160054e-05,
+      "loss": 1.0799,
+      "step": 13067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0800694379862088e-05,
+      "loss": 1.0047,
+      "step": 13068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0799335800023589e-05,
+      "loss": 0.9914,
+      "step": 13069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0797977242669987e-05,
+      "loss": 0.9496,
+      "step": 13070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0796618707826735e-05,
+      "loss": 0.9276,
+      "step": 13071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0795260195519272e-05,
+      "loss": 1.0311,
+      "step": 13072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0793901705773032e-05,
+      "loss": 1.005,
+      "step": 13073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0792543238613465e-05,
+      "loss": 1.0336,
+      "step": 13074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0791184794066006e-05,
+      "loss": 0.9928,
+      "step": 13075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0789826372156095e-05,
+      "loss": 1.0326,
+      "step": 13076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.078846797290917e-05,
+      "loss": 1.0329,
+      "step": 13077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0787109596350672e-05,
+      "loss": 1.0086,
+      "step": 13078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.078575124250604e-05,
+      "loss": 1.0451,
+      "step": 13079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0784392911400707e-05,
+      "loss": 1.0143,
+      "step": 13080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.078303460306012e-05,
+      "loss": 0.9733,
+      "step": 13081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0781676317509709e-05,
+      "loss": 0.9832,
+      "step": 13082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0780318054774907e-05,
+      "loss": 0.9903,
+      "step": 13083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.077895981488116e-05,
+      "loss": 1.087,
+      "step": 13084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0777601597853899e-05,
+      "loss": 1.0078,
+      "step": 13085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.077624340371856e-05,
+      "loss": 1.0292,
+      "step": 13086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0774885232500578e-05,
+      "loss": 1.0074,
+      "step": 13087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0773527084225382e-05,
+      "loss": 1.0321,
+      "step": 13088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0772168958918416e-05,
+      "loss": 1.0581,
+      "step": 13089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0770810856605105e-05,
+      "loss": 1.0462,
+      "step": 13090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0769452777310892e-05,
+      "loss": 0.9626,
+      "step": 13091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0768094721061202e-05,
+      "loss": 0.995,
+      "step": 13092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0766736687881465e-05,
+      "loss": 1.0404,
+      "step": 13093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0765378677797121e-05,
+      "loss": 1.003,
+      "step": 13094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0764020690833597e-05,
+      "loss": 0.9568,
+      "step": 13095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0762662727016328e-05,
+      "loss": 0.9716,
+      "step": 13096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.076130478637074e-05,
+      "loss": 1.0118,
+      "step": 13097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0759946868922262e-05,
+      "loss": 0.9912,
+      "step": 13098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.075858897469633e-05,
+      "loss": 1.0017,
+      "step": 13099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0757231103718367e-05,
+      "loss": 1.0396,
+      "step": 13100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0755873256013807e-05,
+      "loss": 0.9293,
+      "step": 13101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0754515431608077e-05,
+      "loss": 0.9703,
+      "step": 13102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0753157630526604e-05,
+      "loss": 0.9658,
+      "step": 13103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0751799852794816e-05,
+      "loss": 1.0032,
+      "step": 13104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0750442098438138e-05,
+      "loss": 1.0402,
+      "step": 13105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0749084367482003e-05,
+      "loss": 1.011,
+      "step": 13106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0747726659951833e-05,
+      "loss": 1.0337,
+      "step": 13107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0746368975873049e-05,
+      "loss": 0.9597,
+      "step": 13108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0745011315271087e-05,
+      "loss": 0.9903,
+      "step": 13109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0743653678171362e-05,
+      "loss": 0.9989,
+      "step": 13110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0742296064599308e-05,
+      "loss": 1.0231,
+      "step": 13111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0740938474580344e-05,
+      "loss": 0.9717,
+      "step": 13112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.073958090813989e-05,
+      "loss": 1.0498,
+      "step": 13113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0738223365303377e-05,
+      "loss": 1.0765,
+      "step": 13114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0736865846096218e-05,
+      "loss": 1.0439,
+      "step": 13115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0735508350543848e-05,
+      "loss": 1.0332,
+      "step": 13116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0734150878671682e-05,
+      "loss": 1.0787,
+      "step": 13117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0732793430505135e-05,
+      "loss": 1.0572,
+      "step": 13118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.073143600606964e-05,
+      "loss": 1.0333,
+      "step": 13119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0730078605390613e-05,
+      "loss": 1.0268,
+      "step": 13120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0728721228493468e-05,
+      "loss": 1.0393,
+      "step": 13121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0727363875403635e-05,
+      "loss": 1.0716,
+      "step": 13122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0726006546146526e-05,
+      "loss": 0.9837,
+      "step": 13123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0724649240747563e-05,
+      "loss": 0.9681,
+      "step": 13124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0723291959232163e-05,
+      "loss": 1.0161,
+      "step": 13125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0721934701625742e-05,
+      "loss": 1.0482,
+      "step": 13126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0720577467953722e-05,
+      "loss": 1.0344,
+      "step": 13127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0719220258241513e-05,
+      "loss": 1.0023,
+      "step": 13128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0717863072514543e-05,
+      "loss": 1.0512,
+      "step": 13129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.071650591079822e-05,
+      "loss": 1.0428,
+      "step": 13130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0715148773117955e-05,
+      "loss": 1.0256,
+      "step": 13131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0713791659499174e-05,
+      "loss": 1.0909,
+      "step": 13132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0712434569967287e-05,
+      "loss": 0.9915,
+      "step": 13133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0711077504547708e-05,
+      "loss": 1.0371,
+      "step": 13134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0709720463265851e-05,
+      "loss": 1.0339,
+      "step": 13135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.0708363446147125e-05,
+      "loss": 1.0191,
+      "step": 13136
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0707006453216953e-05,
+      "loss": 1.0211,
+      "step": 13137
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0705649484500735e-05,
+      "loss": 0.9733,
+      "step": 13138
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0704292540023895e-05,
+      "loss": 0.9697,
+      "step": 13139
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.070293561981184e-05,
+      "loss": 0.9729,
+      "step": 13140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0701578723889979e-05,
+      "loss": 1.0006,
+      "step": 13141
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0700221852283725e-05,
+      "loss": 0.9724,
+      "step": 13142
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0698865005018483e-05,
+      "loss": 1.0115,
+      "step": 13143
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.069750818211967e-05,
+      "loss": 0.9693,
+      "step": 13144
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0696151383612696e-05,
+      "loss": 1.0158,
+      "step": 13145
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.069479460952296e-05,
+      "loss": 0.9311,
+      "step": 13146
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.069343785987588e-05,
+      "loss": 0.9938,
+      "step": 13147
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0692081134696858e-05,
+      "loss": 1.014,
+      "step": 13148
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0690724434011307e-05,
+      "loss": 1.0074,
+      "step": 13149
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0689367757844631e-05,
+      "loss": 1.0161,
+      "step": 13150
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0688011106222235e-05,
+      "loss": 1.0393,
+      "step": 13151
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0686654479169528e-05,
+      "loss": 0.9996,
+      "step": 13152
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.068529787671191e-05,
+      "loss": 0.9779,
+      "step": 13153
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0683941298874796e-05,
+      "loss": 1.0752,
+      "step": 13154
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0682584745683584e-05,
+      "loss": 0.9591,
+      "step": 13155
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0681228217163674e-05,
+      "loss": 0.9627,
+      "step": 13156
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0679871713340482e-05,
+      "loss": 0.989,
+      "step": 13157
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.06785152342394e-05,
+      "loss": 0.9766,
+      "step": 13158
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0677158779885837e-05,
+      "loss": 1.0281,
+      "step": 13159
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0675802350305196e-05,
+      "loss": 0.9279,
+      "step": 13160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0674445945522873e-05,
+      "loss": 0.9909,
+      "step": 13161
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0673089565564276e-05,
+      "loss": 0.9561,
+      "step": 13162
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0671733210454799e-05,
+      "loss": 0.9853,
+      "step": 13163
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0670376880219852e-05,
+      "loss": 0.9909,
+      "step": 13164
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0669020574884828e-05,
+      "loss": 0.9443,
+      "step": 13165
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0667664294475123e-05,
+      "loss": 1.0189,
+      "step": 13166
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0666308039016149e-05,
+      "loss": 1.0893,
+      "step": 13167
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0664951808533294e-05,
+      "loss": 1.0216,
+      "step": 13168
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.066359560305196e-05,
+      "loss": 1.0597,
+      "step": 13169
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0662239422597544e-05,
+      "loss": 0.9951,
+      "step": 13170
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.066088326719544e-05,
+      "loss": 0.983,
+      "step": 13171
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0659527136871053e-05,
+      "loss": 1.0298,
+      "step": 13172
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.065817103164977e-05,
+      "loss": 0.9491,
+      "step": 13173
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0656814951556995e-05,
+      "loss": 0.9858,
+      "step": 13174
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.065545889661812e-05,
+      "loss": 1.0032,
+      "step": 13175
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0654102866858535e-05,
+      "loss": 1.0231,
+      "step": 13176
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0652746862303645e-05,
+      "loss": 0.978,
+      "step": 13177
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0651390882978834e-05,
+      "loss": 1.0585,
+      "step": 13178
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0650034928909503e-05,
+      "loss": 1.0279,
+      "step": 13179
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.064867900012104e-05,
+      "loss": 1.1184,
+      "step": 13180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0647323096638836e-05,
+      "loss": 1.0264,
+      "step": 13181
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.064596721848829e-05,
+      "loss": 0.9807,
+      "step": 13182
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0644611365694786e-05,
+      "loss": 0.9843,
+      "step": 13183
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0643255538283723e-05,
+      "loss": 1.0258,
+      "step": 13184
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0641899736280487e-05,
+      "loss": 1.0644,
+      "step": 13185
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0640543959710466e-05,
+      "loss": 1.0146,
+      "step": 13186
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0639188208599057e-05,
+      "loss": 0.9838,
+      "step": 13187
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0637832482971642e-05,
+      "loss": 1.0389,
+      "step": 13188
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0636476782853611e-05,
+      "loss": 1.0062,
+      "step": 13189
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0635121108270357e-05,
+      "loss": 1.0619,
+      "step": 13190
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.063376545924726e-05,
+      "loss": 1.0142,
+      "step": 13191
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0632409835809716e-05,
+      "loss": 1.0,
+      "step": 13192
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0631054237983107e-05,
+      "loss": 1.044,
+      "step": 13193
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0629698665792817e-05,
+      "loss": 0.9506,
+      "step": 13194
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0628343119264238e-05,
+      "loss": 1.0217,
+      "step": 13195
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0626987598422751e-05,
+      "loss": 1.015,
+      "step": 13196
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0625632103293741e-05,
+      "loss": 1.0465,
+      "step": 13197
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0624276633902597e-05,
+      "loss": 1.0571,
+      "step": 13198
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0622921190274697e-05,
+      "loss": 1.051,
+      "step": 13199
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.062156577243543e-05,
+      "loss": 1.0822,
+      "step": 13200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.062021038041017e-05,
+      "loss": 1.0544,
+      "step": 13201
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.061885501422431e-05,
+      "loss": 0.9594,
+      "step": 13202
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0617499673903228e-05,
+      "loss": 1.0594,
+      "step": 13203
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0616144359472303e-05,
+      "loss": 0.9894,
+      "step": 13204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.061478907095692e-05,
+      "loss": 1.023,
+      "step": 13205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0613433808382455e-05,
+      "loss": 1.0243,
+      "step": 13206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0612078571774293e-05,
+      "loss": 0.9874,
+      "step": 13207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0610723361157811e-05,
+      "loss": 1.0687,
+      "step": 13208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0609368176558388e-05,
+      "loss": 0.9675,
+      "step": 13209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0608013018001403e-05,
+      "loss": 0.9932,
+      "step": 13210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0606657885512233e-05,
+      "loss": 1.0501,
+      "step": 13211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0605302779116259e-05,
+      "loss": 0.9962,
+      "step": 13212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0603947698838857e-05,
+      "loss": 1.0229,
+      "step": 13213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0602592644705396e-05,
+      "loss": 0.9942,
+      "step": 13214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0601237616741266e-05,
+      "loss": 0.9854,
+      "step": 13215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0599882614971833e-05,
+      "loss": 0.9797,
+      "step": 13216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0598527639422476e-05,
+      "loss": 1.0262,
+      "step": 13217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0597172690118568e-05,
+      "loss": 1.0713,
+      "step": 13218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0595817767085481e-05,
+      "loss": 0.9979,
+      "step": 13219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0594462870348594e-05,
+      "loss": 1.002,
+      "step": 13220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0593107999933275e-05,
+      "loss": 1.0391,
+      "step": 13221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0591753155864901e-05,
+      "loss": 1.0297,
+      "step": 13222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0590398338168845e-05,
+      "loss": 0.9807,
+      "step": 13223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.058904354687047e-05,
+      "loss": 0.9877,
+      "step": 13224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0587688781995158e-05,
+      "loss": 1.0083,
+      "step": 13225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0586334043568275e-05,
+      "loss": 0.978,
+      "step": 13226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0584979331615193e-05,
+      "loss": 1.024,
+      "step": 13227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0583624646161278e-05,
+      "loss": 1.0474,
+      "step": 13228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.05822699872319e-05,
+      "loss": 0.9566,
+      "step": 13229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0580915354852432e-05,
+      "loss": 0.9993,
+      "step": 13230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0579560749048237e-05,
+      "loss": 1.0035,
+      "step": 13231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0578206169844688e-05,
+      "loss": 1.0644,
+      "step": 13232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.057685161726715e-05,
+      "loss": 0.9881,
+      "step": 13233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0575497091340988e-05,
+      "loss": 1.0398,
+      "step": 13234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0574142592091572e-05,
+      "loss": 1.0914,
+      "step": 13235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0572788119544259e-05,
+      "loss": 1.0531,
+      "step": 13236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0571433673724426e-05,
+      "loss": 1.0569,
+      "step": 13237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0570079254657435e-05,
+      "loss": 1.0589,
+      "step": 13238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0568724862368641e-05,
+      "loss": 1.0274,
+      "step": 13239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.056737049688342e-05,
+      "loss": 1.0084,
+      "step": 13240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0566016158227126e-05,
+      "loss": 0.9967,
+      "step": 13241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0564661846425129e-05,
+      "loss": 1.0086,
+      "step": 13242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.056330756150279e-05,
+      "loss": 0.9899,
+      "step": 13243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0561953303485464e-05,
+      "loss": 1.0759,
+      "step": 13244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0560599072398519e-05,
+      "loss": 1.0187,
+      "step": 13245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.055924486826731e-05,
+      "loss": 1.0638,
+      "step": 13246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0557890691117206e-05,
+      "loss": 0.9677,
+      "step": 13247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0556536540973562e-05,
+      "loss": 0.9809,
+      "step": 13248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0555182417861732e-05,
+      "loss": 0.9219,
+      "step": 13249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0553828321807084e-05,
+      "loss": 1.0093,
+      "step": 13250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0552474252834969e-05,
+      "loss": 1.0325,
+      "step": 13251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.055112021097075e-05,
+      "loss": 1.0184,
+      "step": 13252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0549766196239782e-05,
+      "loss": 0.9815,
+      "step": 13253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.054841220866742e-05,
+      "loss": 1.0196,
+      "step": 13254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0547058248279026e-05,
+      "loss": 1.0023,
+      "step": 13255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0545704315099948e-05,
+      "loss": 0.9858,
+      "step": 13256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0544350409155542e-05,
+      "loss": 1.0194,
+      "step": 13257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.054299653047117e-05,
+      "loss": 1.0,
+      "step": 13258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0541642679072176e-05,
+      "loss": 1.0245,
+      "step": 13259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0540288854983924e-05,
+      "loss": 1.0074,
+      "step": 13260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0538935058231763e-05,
+      "loss": 0.9368,
+      "step": 13261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0537581288841042e-05,
+      "loss": 0.9911,
+      "step": 13262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0536227546837117e-05,
+      "loss": 0.9303,
+      "step": 13263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0534873832245337e-05,
+      "loss": 0.9615,
+      "step": 13264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0533520145091058e-05,
+      "loss": 1.0653,
+      "step": 13265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0532166485399628e-05,
+      "loss": 1.0319,
+      "step": 13266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.053081285319639e-05,
+      "loss": 0.9814,
+      "step": 13267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0529459248506706e-05,
+      "loss": 0.9942,
+      "step": 13268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0528105671355912e-05,
+      "loss": 1.0119,
+      "step": 13269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.052675212176937e-05,
+      "loss": 1.0222,
+      "step": 13270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0525398599772421e-05,
+      "loss": 0.9898,
+      "step": 13271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0524045105390412e-05,
+      "loss": 1.0501,
+      "step": 13272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0522691638648692e-05,
+      "loss": 0.999,
+      "step": 13273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0521338199572602e-05,
+      "loss": 1.0695,
+      "step": 13274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0519984788187495e-05,
+      "loss": 1.035,
+      "step": 13275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0518631404518715e-05,
+      "loss": 0.9663,
+      "step": 13276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0517278048591602e-05,
+      "loss": 1.0001,
+      "step": 13277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0515924720431508e-05,
+      "loss": 1.0114,
+      "step": 13278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0514571420063767e-05,
+      "loss": 0.9551,
+      "step": 13279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0513218147513734e-05,
+      "loss": 0.9702,
+      "step": 13280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0511864902806745e-05,
+      "loss": 1.0305,
+      "step": 13281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.051051168596814e-05,
+      "loss": 1.0486,
+      "step": 13282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0509158497023268e-05,
+      "loss": 0.939,
+      "step": 13283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0507805335997462e-05,
+      "loss": 1.0257,
+      "step": 13284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.050645220291607e-05,
+      "loss": 1.0052,
+      "step": 13285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0505099097804429e-05,
+      "loss": 1.0579,
+      "step": 13286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0503746020687875e-05,
+      "loss": 1.0488,
+      "step": 13287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0502392971591755e-05,
+      "loss": 1.0068,
+      "step": 13288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0501039950541403e-05,
+      "loss": 0.9717,
+      "step": 13289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0499686957562157e-05,
+      "loss": 1.1117,
+      "step": 13290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0498333992679356e-05,
+      "loss": 0.9792,
+      "step": 13291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0496981055918335e-05,
+      "loss": 1.0405,
+      "step": 13292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0495628147304433e-05,
+      "loss": 1.026,
+      "step": 13293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0494275266862982e-05,
+      "loss": 1.006,
+      "step": 13294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0492922414619324e-05,
+      "loss": 1.0326,
+      "step": 13295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.049156959059879e-05,
+      "loss": 1.0522,
+      "step": 13296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.049021679482671e-05,
+      "loss": 0.9964,
+      "step": 13297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0488864027328427e-05,
+      "loss": 1.0204,
+      "step": 13298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0487511288129268e-05,
+      "loss": 1.0122,
+      "step": 13299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.048615857725457e-05,
+      "loss": 0.9629,
+      "step": 13300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.048480589472966e-05,
+      "loss": 1.0557,
+      "step": 13301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0483453240579875e-05,
+      "loss": 0.9853,
+      "step": 13302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0482100614830543e-05,
+      "loss": 1.0571,
+      "step": 13303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0480748017506993e-05,
+      "loss": 1.0206,
+      "step": 13304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0479395448634561e-05,
+      "loss": 0.9882,
+      "step": 13305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0478042908238574e-05,
+      "loss": 1.0191,
+      "step": 13306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0476690396344358e-05,
+      "loss": 1.0297,
+      "step": 13307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0475337912977246e-05,
+      "loss": 0.9924,
+      "step": 13308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0473985458162563e-05,
+      "loss": 1.0142,
+      "step": 13309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.047263303192564e-05,
+      "loss": 1.0156,
+      "step": 13310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0471280634291801e-05,
+      "loss": 0.9943,
+      "step": 13311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0469928265286368e-05,
+      "loss": 0.9953,
+      "step": 13312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0468575924934677e-05,
+      "loss": 1.03,
+      "step": 13313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0467223613262045e-05,
+      "loss": 0.991,
+      "step": 13314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0465871330293803e-05,
+      "loss": 1.0813,
+      "step": 13315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0464519076055273e-05,
+      "loss": 0.9886,
+      "step": 13316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0463166850571773e-05,
+      "loss": 1.0494,
+      "step": 13317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0461814653868636e-05,
+      "loss": 1.0448,
+      "step": 13318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0460462485971178e-05,
+      "loss": 0.9874,
+      "step": 13319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0459110346904726e-05,
+      "loss": 0.9786,
+      "step": 13320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0457758236694596e-05,
+      "loss": 1.009,
+      "step": 13321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.045640615536611e-05,
+      "loss": 1.0588,
+      "step": 13322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0455054102944593e-05,
+      "loss": 1.0548,
+      "step": 13323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0453702079455362e-05,
+      "loss": 1.0234,
+      "step": 13324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0452350084923733e-05,
+      "loss": 0.9965,
+      "step": 13325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0450998119375033e-05,
+      "loss": 1.0029,
+      "step": 13326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0449646182834572e-05,
+      "loss": 1.0362,
+      "step": 13327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0448294275327676e-05,
+      "loss": 1.0362,
+      "step": 13328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0446942396879654e-05,
+      "loss": 0.9894,
+      "step": 13329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0445590547515827e-05,
+      "loss": 1.0068,
+      "step": 13330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0444238727261512e-05,
+      "loss": 1.0056,
+      "step": 13331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.044288693614202e-05,
+      "loss": 1.0264,
+      "step": 13332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0441535174182675e-05,
+      "loss": 1.0075,
+      "step": 13333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0440183441408785e-05,
+      "loss": 1.0516,
+      "step": 13334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0438831737845658e-05,
+      "loss": 1.0245,
+      "step": 13335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0437480063518621e-05,
+      "loss": 1.0025,
+      "step": 13336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0436128418452977e-05,
+      "loss": 1.0105,
+      "step": 13337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0434776802674045e-05,
+      "loss": 1.0358,
+      "step": 13338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.043342521620713e-05,
+      "loss": 1.0035,
+      "step": 13339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0432073659077547e-05,
+      "loss": 1.0241,
+      "step": 13340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0430722131310607e-05,
+      "loss": 0.9642,
+      "step": 13341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0429370632931617e-05,
+      "loss": 1.0067,
+      "step": 13342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0428019163965891e-05,
+      "loss": 1.0198,
+      "step": 13343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0426667724438738e-05,
+      "loss": 1.034,
+      "step": 13344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0425316314375459e-05,
+      "loss": 0.9924,
+      "step": 13345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0423964933801371e-05,
+      "loss": 0.9219,
+      "step": 13346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0422613582741776e-05,
+      "loss": 0.9909,
+      "step": 13347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0421262261221985e-05,
+      "loss": 1.0012,
+      "step": 13348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0419910969267302e-05,
+      "loss": 0.943,
+      "step": 13349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0418559706903028e-05,
+      "loss": 0.9757,
+      "step": 13350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0417208474154476e-05,
+      "loss": 1.0725,
+      "step": 13351
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0415857271046945e-05,
+      "loss": 1.0775,
+      "step": 13352
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0414506097605744e-05,
+      "loss": 1.0141,
+      "step": 13353
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0413154953856175e-05,
+      "loss": 1.0003,
+      "step": 13354
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0411803839823537e-05,
+      "loss": 0.9792,
+      "step": 13355
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0410452755533136e-05,
+      "loss": 1.0288,
+      "step": 13356
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040910170101027e-05,
+      "loss": 0.9814,
+      "step": 13357
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0407750676280248e-05,
+      "loss": 0.9727,
+      "step": 13358
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0406399681368368e-05,
+      "loss": 1.0104,
+      "step": 13359
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0405048716299921e-05,
+      "loss": 1.0342,
+      "step": 13360
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0403697781100218e-05,
+      "loss": 1.0228,
+      "step": 13361
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.040234687579455e-05,
+      "loss": 0.9717,
+      "step": 13362
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0400996000408224e-05,
+      "loss": 1.011,
+      "step": 13363
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0399645154966534e-05,
+      "loss": 1.007,
+      "step": 13364
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0398294339494771e-05,
+      "loss": 1.0122,
+      "step": 13365
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0396943554018243e-05,
+      "loss": 0.9989,
+      "step": 13366
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0395592798562235e-05,
+      "loss": 0.9131,
+      "step": 13367
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0394242073152052e-05,
+      "loss": 1.0309,
+      "step": 13368
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0392891377812985e-05,
+      "loss": 1.0282,
+      "step": 13369
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0391540712570326e-05,
+      "loss": 1.0222,
+      "step": 13370
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.0390190077449376e-05,
+      "loss": 1.0218,
+      "step": 13371
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.038883947247542e-05,
+      "loss": 0.9488,
+      "step": 13372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0387488897673758e-05,
+      "loss": 1.0037,
+      "step": 13373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0386138353069681e-05,
+      "loss": 0.9871,
+      "step": 13374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0384787838688475e-05,
+      "loss": 0.9838,
+      "step": 13375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.038343735455544e-05,
+      "loss": 0.9629,
+      "step": 13376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0382086900695857e-05,
+      "loss": 0.9801,
+      "step": 13377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0380736477135025e-05,
+      "loss": 0.9909,
+      "step": 13378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0379386083898228e-05,
+      "loss": 0.9868,
+      "step": 13379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0378035721010752e-05,
+      "loss": 1.0488,
+      "step": 13380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0376685388497896e-05,
+      "loss": 1.0152,
+      "step": 13381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.037533508638494e-05,
+      "loss": 0.9505,
+      "step": 13382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0373984814697169e-05,
+      "loss": 1.0198,
+      "step": 13383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0372634573459877e-05,
+      "loss": 1.018,
+      "step": 13384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0371284362698344e-05,
+      "loss": 0.9128,
+      "step": 13385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036993418243786e-05,
+      "loss": 0.9603,
+      "step": 13386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0368584032703702e-05,
+      "loss": 1.046,
+      "step": 13387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0367233913521164e-05,
+      "loss": 1.0185,
+      "step": 13388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0365883824915528e-05,
+      "loss": 1.0138,
+      "step": 13389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036453376691207e-05,
+      "loss": 1.0243,
+      "step": 13390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.036318373953608e-05,
+      "loss": 0.9765,
+      "step": 13391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0361833742812836e-05,
+      "loss": 1.0092,
+      "step": 13392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0360483776767624e-05,
+      "loss": 1.0079,
+      "step": 13393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0359133841425719e-05,
+      "loss": 1.118,
+      "step": 13394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0357783936812403e-05,
+      "loss": 1.0567,
+      "step": 13395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.035643406295296e-05,
+      "loss": 1.0218,
+      "step": 13396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0355084219872666e-05,
+      "loss": 1.0169,
+      "step": 13397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0353734407596796e-05,
+      "loss": 1.086,
+      "step": 13398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0352384626150635e-05,
+      "loss": 0.9643,
+      "step": 13399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0351034875559453e-05,
+      "loss": 0.9625,
+      "step": 13400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0349685155848534e-05,
+      "loss": 0.9861,
+      "step": 13401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0348335467043153e-05,
+      "loss": 1.0429,
+      "step": 13402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.034698580916858e-05,
+      "loss": 0.9867,
+      "step": 13403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0345636182250097e-05,
+      "loss": 1.0617,
+      "step": 13404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0344286586312972e-05,
+      "loss": 1.0986,
+      "step": 13405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0342937021382486e-05,
+      "loss": 1.0299,
+      "step": 13406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0341587487483911e-05,
+      "loss": 1.0271,
+      "step": 13407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.034023798464251e-05,
+      "loss": 0.9605,
+      "step": 13408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.033888851288357e-05,
+      "loss": 1.0424,
+      "step": 13409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0337539072232348e-05,
+      "loss": 0.963,
+      "step": 13410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0336189662714129e-05,
+      "loss": 0.9876,
+      "step": 13411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0334840284354176e-05,
+      "loss": 0.9798,
+      "step": 13412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0333490937177757e-05,
+      "loss": 0.9746,
+      "step": 13413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0332141621210148e-05,
+      "loss": 1.0541,
+      "step": 13414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0330792336476609e-05,
+      "loss": 1.0114,
+      "step": 13415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0329443083002418e-05,
+      "loss": 0.9781,
+      "step": 13416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0328093860812838e-05,
+      "loss": 1.0369,
+      "step": 13417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.032674466993313e-05,
+      "loss": 0.9663,
+      "step": 13418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.032539551038857e-05,
+      "loss": 1.0375,
+      "step": 13419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0324046382204418e-05,
+      "loss": 1.0127,
+      "step": 13420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0322697285405943e-05,
+      "loss": 0.9887,
+      "step": 13421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0321348220018406e-05,
+      "loss": 0.9998,
+      "step": 13422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.031999918606707e-05,
+      "loss": 0.9914,
+      "step": 13423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0318650183577205e-05,
+      "loss": 0.975,
+      "step": 13424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0317301212574066e-05,
+      "loss": 1.0129,
+      "step": 13425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0315952273082924e-05,
+      "loss": 0.965,
+      "step": 13426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0314603365129035e-05,
+      "loss": 0.9828,
+      "step": 13427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0313254488737654e-05,
+      "loss": 0.9922,
+      "step": 13428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0311905643934054e-05,
+      "loss": 0.9896,
+      "step": 13429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.031055683074349e-05,
+      "loss": 1.0332,
+      "step": 13430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0309208049191218e-05,
+      "loss": 1.0463,
+      "step": 13431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0307859299302502e-05,
+      "loss": 0.9655,
+      "step": 13432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0306510581102592e-05,
+      "loss": 1.1014,
+      "step": 13433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0305161894616755e-05,
+      "loss": 1.0281,
+      "step": 13434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0303813239870239e-05,
+      "loss": 1.0332,
+      "step": 13435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.030246461688831e-05,
+      "loss": 1.0145,
+      "step": 13436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0301116025696219e-05,
+      "loss": 1.0434,
+      "step": 13437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0299767466319217e-05,
+      "loss": 1.0119,
+      "step": 13438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0298418938782566e-05,
+      "loss": 0.9987,
+      "step": 13439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0297070443111513e-05,
+      "loss": 1.0627,
+      "step": 13440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.029572197933132e-05,
+      "loss": 0.9862,
+      "step": 13441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0294373547467232e-05,
+      "loss": 1.0576,
+      "step": 13442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0293025147544501e-05,
+      "loss": 1.052,
+      "step": 13443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0291676779588384e-05,
+      "loss": 1.0825,
+      "step": 13444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0290328443624125e-05,
+      "loss": 0.992,
+      "step": 13445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0288980139676984e-05,
+      "loss": 0.9758,
+      "step": 13446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0287631867772204e-05,
+      "loss": 0.9799,
+      "step": 13447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0286283627935035e-05,
+      "loss": 1.0408,
+      "step": 13448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0284935420190725e-05,
+      "loss": 0.9464,
+      "step": 13449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0283587244564523e-05,
+      "loss": 1.0169,
+      "step": 13450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.028223910108168e-05,
+      "loss": 1.008,
+      "step": 13451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0280890989767436e-05,
+      "loss": 1.0328,
+      "step": 13452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0279542910647038e-05,
+      "loss": 0.9905,
+      "step": 13453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0278194863745739e-05,
+      "loss": 1.0532,
+      "step": 13454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0276846849088773e-05,
+      "loss": 0.936,
+      "step": 13455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0275498866701394e-05,
+      "loss": 1.0193,
+      "step": 13456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0274150916608842e-05,
+      "loss": 0.9765,
+      "step": 13457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0272802998836359e-05,
+      "loss": 0.9413,
+      "step": 13458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.027145511340919e-05,
+      "loss": 0.959,
+      "step": 13459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.027010726035257e-05,
+      "loss": 1.024,
+      "step": 13460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0268759439691752e-05,
+      "loss": 0.9988,
+      "step": 13461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026741165145197e-05,
+      "loss": 1.088,
+      "step": 13462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026606389565846e-05,
+      "loss": 0.9862,
+      "step": 13463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026471617233647e-05,
+      "loss": 0.9862,
+      "step": 13464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0263368481511235e-05,
+      "loss": 0.9866,
+      "step": 13465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.026202082320799e-05,
+      "loss": 1.015,
+      "step": 13466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0260673197451978e-05,
+      "loss": 1.0021,
+      "step": 13467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0259325604268434e-05,
+      "loss": 0.9616,
+      "step": 13468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0257978043682596e-05,
+      "loss": 1.0093,
+      "step": 13469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.02566305157197e-05,
+      "loss": 0.9642,
+      "step": 13470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0255283020404971e-05,
+      "loss": 0.9958,
+      "step": 13471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0253935557763658e-05,
+      "loss": 1.0452,
+      "step": 13472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0252588127820986e-05,
+      "loss": 0.9881,
+      "step": 13473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0251240730602196e-05,
+      "loss": 1.0277,
+      "step": 13474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0249893366132515e-05,
+      "loss": 1.0158,
+      "step": 13475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0248546034437172e-05,
+      "loss": 1.0677,
+      "step": 13476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0247198735541409e-05,
+      "loss": 1.0033,
+      "step": 13477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0245851469470448e-05,
+      "loss": 0.9612,
+      "step": 13478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0244504236249525e-05,
+      "loss": 1.034,
+      "step": 13479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0243157035903866e-05,
+      "loss": 1.0457,
+      "step": 13480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0241809868458697e-05,
+      "loss": 1.0317,
+      "step": 13481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0240462733939254e-05,
+      "loss": 0.9545,
+      "step": 13482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0239115632370759e-05,
+      "loss": 0.9625,
+      "step": 13483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0237768563778444e-05,
+      "loss": 0.9885,
+      "step": 13484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0236421528187535e-05,
+      "loss": 0.8913,
+      "step": 13485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0235074525623253e-05,
+      "loss": 0.981,
+      "step": 13486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0233727556110825e-05,
+      "loss": 0.9929,
+      "step": 13487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0232380619675481e-05,
+      "loss": 1.0331,
+      "step": 13488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0231033716342443e-05,
+      "loss": 1.0461,
+      "step": 13489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0229686846136932e-05,
+      "loss": 1.0253,
+      "step": 13490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0228340009084167e-05,
+      "loss": 0.953,
+      "step": 13491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0226993205209383e-05,
+      "loss": 1.0392,
+      "step": 13492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0225646434537787e-05,
+      "loss": 0.9563,
+      "step": 13493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.022429969709461e-05,
+      "loss": 0.9974,
+      "step": 13494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0222952992905072e-05,
+      "loss": 1.1007,
+      "step": 13495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0221606321994388e-05,
+      "loss": 1.0042,
+      "step": 13496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0220259684387782e-05,
+      "loss": 1.0117,
+      "step": 13497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0218913080110464e-05,
+      "loss": 1.0082,
+      "step": 13498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0217566509187663e-05,
+      "loss": 0.941,
+      "step": 13499
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0216219971644591e-05,
+      "loss": 0.9738,
+      "step": 13500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0214873467506461e-05,
+      "loss": 1.0004,
+      "step": 13501
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0213526996798499e-05,
+      "loss": 1.0106,
+      "step": 13502
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0212180559545908e-05,
+      "loss": 1.0069,
+      "step": 13503
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0210834155773915e-05,
+      "loss": 1.0143,
+      "step": 13504
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0209487785507727e-05,
+      "loss": 1.0548,
+      "step": 13505
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0208141448772557e-05,
+      "loss": 1.0554,
+      "step": 13506
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0206795145593622e-05,
+      "loss": 0.8837,
+      "step": 13507
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.020544887599613e-05,
+      "loss": 1.024,
+      "step": 13508
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0204102640005296e-05,
+      "loss": 0.9789,
+      "step": 13509
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0202756437646331e-05,
+      "loss": 0.9781,
+      "step": 13510
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.020141026894444e-05,
+      "loss": 0.9908,
+      "step": 13511
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0200064133924842e-05,
+      "loss": 1.0006,
+      "step": 13512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0198718032612739e-05,
+      "loss": 0.9846,
+      "step": 13513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0197371965033343e-05,
+      "loss": 0.9702,
+      "step": 13514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.019602593121186e-05,
+      "loss": 1.0021,
+      "step": 13515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0194679931173497e-05,
+      "loss": 0.9961,
+      "step": 13516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0193333964943464e-05,
+      "loss": 0.9989,
+      "step": 13517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0191988032546959e-05,
+      "loss": 0.9952,
+      "step": 13518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0190642134009198e-05,
+      "loss": 1.013,
+      "step": 13519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018929626935538e-05,
+      "loss": 1.1044,
+      "step": 13520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0187950438610705e-05,
+      "loss": 0.9988,
+      "step": 13521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0186604641800385e-05,
+      "loss": 1.0787,
+      "step": 13522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0185258878949619e-05,
+      "loss": 1.0369,
+      "step": 13523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018391315008361e-05,
+      "loss": 0.9444,
+      "step": 13524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0182567455227557e-05,
+      "loss": 1.0663,
+      "step": 13525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.018122179440666e-05,
+      "loss": 1.0393,
+      "step": 13526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0179876167646125e-05,
+      "loss": 0.9546,
+      "step": 13527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0178530574971145e-05,
+      "loss": 0.9817,
+      "step": 13528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0177185016406927e-05,
+      "loss": 0.9788,
+      "step": 13529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0175839491978662e-05,
+      "loss": 0.9909,
+      "step": 13530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.017449400171155e-05,
+      "loss": 0.9813,
+      "step": 13531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.017314854563079e-05,
+      "loss": 0.9558,
+      "step": 13532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0171803123761578e-05,
+      "loss": 0.9876,
+      "step": 13533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0170457736129105e-05,
+      "loss": 1.028,
+      "step": 13534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0169112382758573e-05,
+      "loss": 0.9993,
+      "step": 13535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.016776706367517e-05,
+      "loss": 1.0211,
+      "step": 13536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0166421778904096e-05,
+      "loss": 0.9763,
+      "step": 13537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0165076528470544e-05,
+      "loss": 0.999,
+      "step": 13538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0163731312399695e-05,
+      "loss": 1.0492,
+      "step": 13539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.016238613071676e-05,
+      "loss": 1.0391,
+      "step": 13540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0161040983446913e-05,
+      "loss": 1.0181,
+      "step": 13541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0159695870615354e-05,
+      "loss": 1.0253,
+      "step": 13542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0158350792247274e-05,
+      "loss": 0.9892,
+      "step": 13543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0157005748367856e-05,
+      "loss": 0.9684,
+      "step": 13544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0155660739002295e-05,
+      "loss": 1.0447,
+      "step": 13545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015431576417577e-05,
+      "loss": 0.9873,
+      "step": 13546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015297082391348e-05,
+      "loss": 1.0222,
+      "step": 13547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0151625918240606e-05,
+      "loss": 1.0063,
+      "step": 13548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.015028104718233e-05,
+      "loss": 1.0222,
+      "step": 13549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0148936210763846e-05,
+      "loss": 0.916,
+      "step": 13550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0147591409010333e-05,
+      "loss": 1.0526,
+      "step": 13551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0146246641946977e-05,
+      "loss": 0.9814,
+      "step": 13552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.014490190959896e-05,
+      "loss": 1.0355,
+      "step": 13553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0143557211991467e-05,
+      "loss": 0.987,
+      "step": 13554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0142212549149681e-05,
+      "loss": 0.9962,
+      "step": 13555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0140867921098778e-05,
+      "loss": 1.0081,
+      "step": 13556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0139523327863948e-05,
+      "loss": 0.9907,
+      "step": 13557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0138178769470364e-05,
+      "loss": 1.0247,
+      "step": 13558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0136834245943205e-05,
+      "loss": 1.0557,
+      "step": 13559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0135489757307658e-05,
+      "loss": 1.0037,
+      "step": 13560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0134145303588892e-05,
+      "loss": 1.0881,
+      "step": 13561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0132800884812093e-05,
+      "loss": 1.061,
+      "step": 13562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0131456501002432e-05,
+      "loss": 1.016,
+      "step": 13563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0130112152185082e-05,
+      "loss": 0.9943,
+      "step": 13564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.012876783838523e-05,
+      "loss": 1.0417,
+      "step": 13565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.012742355962804e-05,
+      "loss": 1.0574,
+      "step": 13566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0126079315938695e-05,
+      "loss": 1.0049,
+      "step": 13567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0124735107342367e-05,
+      "loss": 1.0791,
+      "step": 13568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.012339093386422e-05,
+      "loss": 0.9989,
+      "step": 13569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0122046795529439e-05,
+      "loss": 1.0182,
+      "step": 13570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0120702692363188e-05,
+      "loss": 1.0002,
+      "step": 13571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0119358624390642e-05,
+      "loss": 0.9512,
+      "step": 13572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.011801459163697e-05,
+      "loss": 0.9915,
+      "step": 13573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0116670594127336e-05,
+      "loss": 0.9901,
+      "step": 13574
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0115326631886922e-05,
+      "loss": 0.9667,
+      "step": 13575
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.011398270494088e-05,
+      "loss": 1.0011,
+      "step": 13576
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0112638813314395e-05,
+      "loss": 1.0443,
+      "step": 13577
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0111294957032625e-05,
+      "loss": 0.9404,
+      "step": 13578
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0109951136120735e-05,
+      "loss": 1.0039,
+      "step": 13579
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0108607350603892e-05,
+      "loss": 1.0031,
+      "step": 13580
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0107263600507263e-05,
+      "loss": 1.017,
+      "step": 13581
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0105919885856018e-05,
+      "loss": 0.9928,
+      "step": 13582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.010457620667531e-05,
+      "loss": 0.9332,
+      "step": 13583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0103232562990304e-05,
+      "loss": 1.0229,
+      "step": 13584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0101888954826168e-05,
+      "loss": 1.0286,
+      "step": 13585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0100545382208058e-05,
+      "loss": 0.9988,
+      "step": 13586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0099201845161144e-05,
+      "loss": 1.0303,
+      "step": 13587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.009785834371058e-05,
+      "loss": 0.9803,
+      "step": 13588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0096514877881524e-05,
+      "loss": 1.0114,
+      "step": 13589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.009517144769914e-05,
+      "loss": 1.0113,
+      "step": 13590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.009382805318858e-05,
+      "loss": 1.0142,
+      "step": 13591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0092484694375011e-05,
+      "loss": 1.002,
+      "step": 13592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0091141371283583e-05,
+      "loss": 1.0262,
+      "step": 13593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0089798083939453e-05,
+      "loss": 0.9566,
+      "step": 13594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0088454832367782e-05,
+      "loss": 0.9566,
+      "step": 13595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0087111616593718e-05,
+      "loss": 0.9554,
+      "step": 13596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0085768436642421e-05,
+      "loss": 1.0169,
+      "step": 13597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0084425292539045e-05,
+      "loss": 1.0297,
+      "step": 13598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0083082184308739e-05,
+      "loss": 1.012,
+      "step": 13599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.008173911197666e-05,
+      "loss": 0.9744,
+      "step": 13600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0080396075567958e-05,
+      "loss": 1.0076,
+      "step": 13601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0079053075107776e-05,
+      "loss": 0.9985,
+      "step": 13602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0077710110621279e-05,
+      "loss": 0.996,
+      "step": 13603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0076367182133604e-05,
+      "loss": 0.9257,
+      "step": 13604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0075024289669911e-05,
+      "loss": 1.0159,
+      "step": 13605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0073681433255343e-05,
+      "loss": 0.9255,
+      "step": 13606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0072338612915045e-05,
+      "loss": 1.0426,
+      "step": 13607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0070995828674167e-05,
+      "loss": 0.973,
+      "step": 13608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.0069653080557857e-05,
+      "loss": 1.0339,
+      "step": 13609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0068310368591261e-05,
+      "loss": 1.0279,
+      "step": 13610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.006696769279952e-05,
+      "loss": 1.0307,
+      "step": 13611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0065625053207778e-05,
+      "loss": 1.0152,
+      "step": 13612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0064282449841186e-05,
+      "loss": 0.9653,
+      "step": 13613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0062939882724878e-05,
+      "loss": 1.0683,
+      "step": 13614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0061597351884002e-05,
+      "loss": 1.0834,
+      "step": 13615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0060254857343701e-05,
+      "loss": 1.0289,
+      "step": 13616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.005891239912911e-05,
+      "loss": 1.0156,
+      "step": 13617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0057569977265374e-05,
+      "loss": 1.0177,
+      "step": 13618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0056227591777628e-05,
+      "loss": 1.0132,
+      "step": 13619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0054885242691018e-05,
+      "loss": 0.9389,
+      "step": 13620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0053542930030678e-05,
+      "loss": 1.0514,
+      "step": 13621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.005220065382174e-05,
+      "loss": 0.957,
+      "step": 13622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0050858414089353e-05,
+      "loss": 0.9843,
+      "step": 13623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0049516210858643e-05,
+      "loss": 1.0044,
+      "step": 13624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0048174044154752e-05,
+      "loss": 1.0803,
+      "step": 13625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0046831914002813e-05,
+      "loss": 0.9825,
+      "step": 13626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0045489820427956e-05,
+      "loss": 0.9932,
+      "step": 13627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0044147763455323e-05,
+      "loss": 1.0591,
+      "step": 13628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0042805743110035e-05,
+      "loss": 0.9962,
+      "step": 13629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0041463759417234e-05,
+      "loss": 1.0365,
+      "step": 13630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.004012181240205e-05,
+      "loss": 0.954,
+      "step": 13631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0038779902089606e-05,
+      "loss": 1.0097,
+      "step": 13632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0037438028505044e-05,
+      "loss": 1.0716,
+      "step": 13633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0036096191673484e-05,
+      "loss": 0.9967,
+      "step": 13634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0034754391620059e-05,
+      "loss": 1.0288,
+      "step": 13635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0033412628369897e-05,
+      "loss": 1.0122,
+      "step": 13636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.003207090194812e-05,
+      "loss": 0.9925,
+      "step": 13637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0030729212379862e-05,
+      "loss": 0.9782,
+      "step": 13638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0029387559690242e-05,
+      "loss": 0.9814,
+      "step": 13639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0028045943904393e-05,
+      "loss": 1.0279,
+      "step": 13640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0026704365047435e-05,
+      "loss": 1.0075,
+      "step": 13641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0025362823144487e-05,
+      "loss": 0.978,
+      "step": 13642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0024021318220681e-05,
+      "loss": 0.9656,
+      "step": 13643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0022679850301136e-05,
+      "loss": 1.0109,
+      "step": 13644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0021338419410975e-05,
+      "loss": 0.969,
+      "step": 13645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0019997025575314e-05,
+      "loss": 1.0593,
+      "step": 13646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0018655668819277e-05,
+      "loss": 0.9695,
+      "step": 13647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0017314349167985e-05,
+      "loss": 0.9285,
+      "step": 13648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0015973066646553e-05,
+      "loss": 1.025,
+      "step": 13649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0014631821280105e-05,
+      "loss": 0.9908,
+      "step": 13650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0013290613093756e-05,
+      "loss": 1.0464,
+      "step": 13651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0011949442112617e-05,
+      "loss": 1.0317,
+      "step": 13652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0010608308361811e-05,
+      "loss": 0.9895,
+      "step": 13653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0009267211866453e-05,
+      "loss": 0.989,
+      "step": 13654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0007926152651658e-05,
+      "loss": 1.0085,
+      "step": 13655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0006585130742538e-05,
+      "loss": 0.9862,
+      "step": 13656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0005244146164204e-05,
+      "loss": 1.0026,
+      "step": 13657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0003903198941775e-05,
+      "loss": 0.983,
+      "step": 13658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0002562289100357e-05,
+      "loss": 1.0623,
+      "step": 13659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.0001221416665067e-05,
+      "loss": 0.9647,
+      "step": 13660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.999880581661012e-06,
+      "loss": 0.9723,
+      "step": 13661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.9985397841133e-06,
+      "loss": 0.9721,
+      "step": 13662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.997199024047046e-06,
+      "loss": 0.9256,
+      "step": 13663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.995858301487353e-06,
+      "loss": 0.9581,
+      "step": 13664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.994517616459333e-06,
+      "loss": 0.9523,
+      "step": 13665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.99317696898809e-06,
+      "loss": 1.0251,
+      "step": 13666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.99183635909873e-06,
+      "loss": 0.999,
+      "step": 13667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.990495786816362e-06,
+      "loss": 1.026,
+      "step": 13668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.989155252166089e-06,
+      "loss": 1.016,
+      "step": 13669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.987814755173013e-06,
+      "loss": 1.0012,
+      "step": 13670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.986474295862243e-06,
+      "loss": 1.0119,
+      "step": 13671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.985133874258877e-06,
+      "loss": 1.064,
+      "step": 13672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.983793490388019e-06,
+      "loss": 0.9517,
+      "step": 13673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.98245314427477e-06,
+      "loss": 1.0402,
+      "step": 13674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.981112835944232e-06,
+      "loss": 0.9772,
+      "step": 13675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.979772565421507e-06,
+      "loss": 0.9567,
+      "step": 13676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.978432332731684e-06,
+      "loss": 0.9725,
+      "step": 13677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.977092137899875e-06,
+      "loss": 0.9421,
+      "step": 13678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.975751980951172e-06,
+      "loss": 1.0376,
+      "step": 13679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.974411861910667e-06,
+      "loss": 1.0397,
+      "step": 13680
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.973071780803467e-06,
+      "loss": 1.0688,
+      "step": 13681
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.971731737654659e-06,
+      "loss": 0.9824,
+      "step": 13682
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.970391732489345e-06,
+      "loss": 0.992,
+      "step": 13683
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.969051765332615e-06,
+      "loss": 0.9512,
+      "step": 13684
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.96771183620956e-06,
+      "loss": 1.0992,
+      "step": 13685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.966371945145279e-06,
+      "loss": 1.0248,
+      "step": 13686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.965032092164858e-06,
+      "loss": 0.925,
+      "step": 13687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.963692277293397e-06,
+      "loss": 1.0072,
+      "step": 13688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.962352500555982e-06,
+      "loss": 0.954,
+      "step": 13689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.961012761977698e-06,
+      "loss": 0.9723,
+      "step": 13690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.959673061583644e-06,
+      "loss": 0.9864,
+      "step": 13691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.9583333993989e-06,
+      "loss": 0.9635,
+      "step": 13692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.956993775448562e-06,
+      "loss": 1.0355,
+      "step": 13693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.955654189757712e-06,
+      "loss": 0.9736,
+      "step": 13694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.954314642351434e-06,
+      "loss": 0.9359,
+      "step": 13695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.952975133254823e-06,
+      "loss": 1.0012,
+      "step": 13696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.951635662492953e-06,
+      "loss": 0.96,
+      "step": 13697
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.950296230090916e-06,
+      "loss": 0.9917,
+      "step": 13698
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.948956836073797e-06,
+      "loss": 0.9463,
+      "step": 13699
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.94761748046667e-06,
+      "loss": 1.0327,
+      "step": 13700
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.946278163294623e-06,
+      "loss": 0.9657,
+      "step": 13701
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.944938884582739e-06,
+      "loss": 0.9618,
+      "step": 13702
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.943599644356096e-06,
+      "loss": 0.9744,
+      "step": 13703
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.942260442639777e-06,
+      "loss": 1.0255,
+      "step": 13704
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.940921279458854e-06,
+      "loss": 0.9905,
+      "step": 13705
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.939582154838413e-06,
+      "loss": 0.919,
+      "step": 13706
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.938243068803528e-06,
+      "loss": 1.0022,
+      "step": 13707
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.936904021379282e-06,
+      "loss": 0.9875,
+      "step": 13708
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.935565012590744e-06,
+      "loss": 0.9634,
+      "step": 13709
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.934226042462993e-06,
+      "loss": 1.0088,
+      "step": 13710
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.932887111021105e-06,
+      "loss": 0.9699,
+      "step": 13711
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.93154821829015e-06,
+      "loss": 1.0065,
+      "step": 13712
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.930209364295206e-06,
+      "loss": 0.9743,
+      "step": 13713
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.928870549061346e-06,
+      "loss": 0.9592,
+      "step": 13714
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.927531772613635e-06,
+      "loss": 1.0158,
+      "step": 13715
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.926193034977156e-06,
+      "loss": 0.9164,
+      "step": 13716
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.924854336176965e-06,
+      "loss": 1.0314,
+      "step": 13717
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.923515676238147e-06,
+      "loss": 0.9817,
+      "step": 13718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.922177055185765e-06,
+      "loss": 1.0587,
+      "step": 13719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.920838473044883e-06,
+      "loss": 1.038,
+      "step": 13720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.919499929840575e-06,
+      "loss": 0.9642,
+      "step": 13721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.918161425597902e-06,
+      "loss": 0.9396,
+      "step": 13722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.916822960341937e-06,
+      "loss": 0.9909,
+      "step": 13723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.915484534097743e-06,
+      "loss": 1.0562,
+      "step": 13724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.914146146890378e-06,
+      "loss": 1.0051,
+      "step": 13725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.912807798744917e-06,
+      "loss": 1.0203,
+      "step": 13726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.911469489686417e-06,
+      "loss": 1.0217,
+      "step": 13727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.910131219739942e-06,
+      "loss": 0.9898,
+      "step": 13728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.908792988930553e-06,
+      "loss": 0.9753,
+      "step": 13729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.907454797283313e-06,
+      "loss": 0.9749,
+      "step": 13730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.906116644823281e-06,
+      "loss": 1.0057,
+      "step": 13731
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.904778531575515e-06,
+      "loss": 0.9904,
+      "step": 13732
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.903440457565077e-06,
+      "loss": 0.9634,
+      "step": 13733
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.902102422817026e-06,
+      "loss": 0.9763,
+      "step": 13734
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.900764427356413e-06,
+      "loss": 1.0453,
+      "step": 13735
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.899426471208304e-06,
+      "loss": 0.995,
+      "step": 13736
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.898088554397748e-06,
+      "loss": 1.0403,
+      "step": 13737
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.896750676949802e-06,
+      "loss": 0.9515,
+      "step": 13738
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.895412838889521e-06,
+      "loss": 0.9641,
+      "step": 13739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.894075040241958e-06,
+      "loss": 1.0199,
+      "step": 13740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.89273728103217e-06,
+      "loss": 1.038,
+      "step": 13741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.891399561285204e-06,
+      "loss": 0.9559,
+      "step": 13742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.890061881026111e-06,
+      "loss": 1.0142,
+      "step": 13743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.888724240279947e-06,
+      "loss": 1.0286,
+      "step": 13744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.887386639071757e-06,
+      "loss": 0.9881,
+      "step": 13745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.886049077426595e-06,
+      "loss": 0.9654,
+      "step": 13746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.884711555369509e-06,
+      "loss": 0.9517,
+      "step": 13747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.88337407292554e-06,
+      "loss": 0.9974,
+      "step": 13748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.882036630119746e-06,
+      "loss": 1.0494,
+      "step": 13749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.880699226977161e-06,
+      "loss": 0.9715,
+      "step": 13750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.879361863522843e-06,
+      "loss": 1.0491,
+      "step": 13751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.87802453978183e-06,
+      "loss": 0.9738,
+      "step": 13752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.876687255779163e-06,
+      "loss": 1.0299,
+      "step": 13753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.875350011539896e-06,
+      "loss": 1.0073,
+      "step": 13754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.874012807089059e-06,
+      "loss": 0.9877,
+      "step": 13755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.872675642451705e-06,
+      "loss": 1.0977,
+      "step": 13756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.87133851765287e-06,
+      "loss": 1.0409,
+      "step": 13757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.870001432717592e-06,
+      "loss": 0.9837,
+      "step": 13758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.868664387670918e-06,
+      "loss": 1.0268,
+      "step": 13759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.867327382537877e-06,
+      "loss": 0.9595,
+      "step": 13760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.865990417343515e-06,
+      "loss": 0.9969,
+      "step": 13761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.864653492112868e-06,
+      "loss": 1.0276,
+      "step": 13762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.863316606870968e-06,
+      "loss": 1.0486,
+      "step": 13763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.861979761642859e-06,
+      "loss": 0.9445,
+      "step": 13764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.860642956453568e-06,
+      "loss": 0.9917,
+      "step": 13765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.859306191328135e-06,
+      "loss": 0.9737,
+      "step": 13766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.857969466291593e-06,
+      "loss": 0.9477,
+      "step": 13767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.856632781368972e-06,
+      "loss": 1.0111,
+      "step": 13768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.855296136585307e-06,
+      "loss": 0.9703,
+      "step": 13769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.853959531965626e-06,
+      "loss": 0.9916,
+      "step": 13770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.852622967534964e-06,
+      "loss": 1.0158,
+      "step": 13771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.851286443318349e-06,
+      "loss": 1.0119,
+      "step": 13772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.849949959340808e-06,
+      "loss": 1.0365,
+      "step": 13773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.848613515627374e-06,
+      "loss": 1.0104,
+      "step": 13774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.84727711220307e-06,
+      "loss": 0.9325,
+      "step": 13775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.845940749092928e-06,
+      "loss": 1.0081,
+      "step": 13776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.84460442632197e-06,
+      "loss": 0.9839,
+      "step": 13777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.843268143915218e-06,
+      "loss": 1.0539,
+      "step": 13778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.841931901897707e-06,
+      "loss": 0.9846,
+      "step": 13779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.84059570029445e-06,
+      "loss": 1.0041,
+      "step": 13780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.839259539130481e-06,
+      "loss": 0.9462,
+      "step": 13781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.837923418430814e-06,
+      "loss": 1.0416,
+      "step": 13782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.83658733822047e-06,
+      "loss": 0.9945,
+      "step": 13783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.835251298524478e-06,
+      "loss": 0.9623,
+      "step": 13784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.83391529936785e-06,
+      "loss": 0.9765,
+      "step": 13785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.83257934077561e-06,
+      "loss": 1.0541,
+      "step": 13786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.831243422772776e-06,
+      "loss": 1.0354,
+      "step": 13787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.829907545384362e-06,
+      "loss": 1.0063,
+      "step": 13788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.82857170863539e-06,
+      "loss": 1.037,
+      "step": 13789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.82723591255087e-06,
+      "loss": 1.025,
+      "step": 13790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.825900157155827e-06,
+      "loss": 0.9772,
+      "step": 13791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.824564442475269e-06,
+      "loss": 1.1058,
+      "step": 13792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.823228768534209e-06,
+      "loss": 1.003,
+      "step": 13793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.821893135357665e-06,
+      "loss": 0.9848,
+      "step": 13794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.820557542970645e-06,
+      "loss": 0.9759,
+      "step": 13795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.819221991398166e-06,
+      "loss": 1.0098,
+      "step": 13796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.817886480665236e-06,
+      "loss": 0.9868,
+      "step": 13797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.816551010796859e-06,
+      "loss": 0.9078,
+      "step": 13798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.815215581818058e-06,
+      "loss": 0.9148,
+      "step": 13799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.813880193753826e-06,
+      "loss": 0.9795,
+      "step": 13800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.812544846629185e-06,
+      "loss": 1.0599,
+      "step": 13801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.811209540469136e-06,
+      "loss": 0.8944,
+      "step": 13802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.809874275298682e-06,
+      "loss": 1.001,
+      "step": 13803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.808539051142837e-06,
+      "loss": 1.0019,
+      "step": 13804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.807203868026596e-06,
+      "loss": 1.012,
+      "step": 13805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.80586872597497e-06,
+      "loss": 1.023,
+      "step": 13806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.80453362501296e-06,
+      "loss": 0.9875,
+      "step": 13807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.803198565165565e-06,
+      "loss": 0.9897,
+      "step": 13808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.801863546457796e-06,
+      "loss": 0.9904,
+      "step": 13809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.800528568914648e-06,
+      "loss": 0.9194,
+      "step": 13810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.799193632561117e-06,
+      "loss": 0.9873,
+      "step": 13811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.79785873742221e-06,
+      "loss": 1.0063,
+      "step": 13812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.796523883522921e-06,
+      "loss": 0.9901,
+      "step": 13813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.795189070888254e-06,
+      "loss": 1.0356,
+      "step": 13814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.793854299543199e-06,
+      "loss": 1.0117,
+      "step": 13815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.79251956951275e-06,
+      "loss": 1.0065,
+      "step": 13816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.791184880821915e-06,
+      "loss": 1.0025,
+      "step": 13817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.789850233495675e-06,
+      "loss": 0.9926,
+      "step": 13818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.788515627559034e-06,
+      "loss": 0.9825,
+      "step": 13819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.787181063036983e-06,
+      "loss": 0.9537,
+      "step": 13820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.785846539954508e-06,
+      "loss": 0.9858,
+      "step": 13821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.78451205833661e-06,
+      "loss": 1.0058,
+      "step": 13822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.783177618208272e-06,
+      "loss": 0.9862,
+      "step": 13823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.781843219594492e-06,
+      "loss": 0.9902,
+      "step": 13824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.780508862520254e-06,
+      "loss": 0.9941,
+      "step": 13825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.779174547010545e-06,
+      "loss": 1.0206,
+      "step": 13826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.777840273090357e-06,
+      "loss": 0.9811,
+      "step": 13827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.776506040784674e-06,
+      "loss": 1.0432,
+      "step": 13828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.775171850118487e-06,
+      "loss": 1.0402,
+      "step": 13829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.773837701116778e-06,
+      "loss": 0.9587,
+      "step": 13830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.77250359380453e-06,
+      "loss": 1.046,
+      "step": 13831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.771169528206728e-06,
+      "loss": 1.0141,
+      "step": 13832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.769835504348355e-06,
+      "loss": 1.0594,
+      "step": 13833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.768501522254399e-06,
+      "loss": 1.0384,
+      "step": 13834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.767167581949836e-06,
+      "loss": 0.9749,
+      "step": 13835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.765833683459642e-06,
+      "loss": 1.0302,
+      "step": 13836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.76449982680881e-06,
+      "loss": 1.0775,
+      "step": 13837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.763166012022303e-06,
+      "loss": 1.005,
+      "step": 13838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.761832239125115e-06,
+      "loss": 1.0136,
+      "step": 13839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.760498508142216e-06,
+      "loss": 1.0213,
+      "step": 13840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.759164819098583e-06,
+      "loss": 1.0382,
+      "step": 13841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.757831172019195e-06,
+      "loss": 1.0213,
+      "step": 13842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.756497566929018e-06,
+      "loss": 1.0067,
+      "step": 13843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.755164003853039e-06,
+      "loss": 1.0324,
+      "step": 13844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.753830482816227e-06,
+      "loss": 1.0456,
+      "step": 13845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.75249700384355e-06,
+      "loss": 0.9765,
+      "step": 13846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.751163566959988e-06,
+      "loss": 0.9704,
+      "step": 13847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.749830172190504e-06,
+      "loss": 0.9273,
+      "step": 13848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.748496819560078e-06,
+      "loss": 0.9617,
+      "step": 13849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.747163509093675e-06,
+      "loss": 1.0062,
+      "step": 13850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.745830240816262e-06,
+      "loss": 0.9991,
+      "step": 13851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.74449701475281e-06,
+      "loss": 1.0433,
+      "step": 13852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.743163830928283e-06,
+      "loss": 1.0509,
+      "step": 13853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.741830689367653e-06,
+      "loss": 1.079,
+      "step": 13854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.740497590095883e-06,
+      "loss": 1.0106,
+      "step": 13855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.739164533137934e-06,
+      "loss": 0.9721,
+      "step": 13856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.73783151851878e-06,
+      "loss": 0.9986,
+      "step": 13857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.736498546263374e-06,
+      "loss": 1.0072,
+      "step": 13858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.735165616396684e-06,
+      "loss": 0.9743,
+      "step": 13859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.733832728943673e-06,
+      "loss": 0.9947,
+      "step": 13860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.732499883929299e-06,
+      "loss": 1.017,
+      "step": 13861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.731167081378525e-06,
+      "loss": 1.0319,
+      "step": 13862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.729834321316307e-06,
+      "loss": 1.005,
+      "step": 13863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.728501603767605e-06,
+      "loss": 1.0491,
+      "step": 13864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.727168928757381e-06,
+      "loss": 0.9774,
+      "step": 13865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.725836296310583e-06,
+      "loss": 0.9708,
+      "step": 13866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.724503706452178e-06,
+      "loss": 1.0379,
+      "step": 13867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.723171159207114e-06,
+      "loss": 1.0485,
+      "step": 13868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.72183865460035e-06,
+      "loss": 0.968,
+      "step": 13869
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.720506192656838e-06,
+      "loss": 1.0224,
+      "step": 13870
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.719173773401525e-06,
+      "loss": 1.0272,
+      "step": 13871
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.717841396859375e-06,
+      "loss": 1.0052,
+      "step": 13872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.716509063055329e-06,
+      "loss": 1.0007,
+      "step": 13873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.715176772014347e-06,
+      "loss": 0.961,
+      "step": 13874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.713844523761373e-06,
+      "loss": 0.9547,
+      "step": 13875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.712512318321355e-06,
+      "loss": 0.9411,
+      "step": 13876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.711180155719247e-06,
+      "loss": 0.9638,
+      "step": 13877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.709848035979994e-06,
+      "loss": 0.981,
+      "step": 13878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.708515959128538e-06,
+      "loss": 1.0143,
+      "step": 13879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.707183925189834e-06,
+      "loss": 1.0197,
+      "step": 13880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.705851934188815e-06,
+      "loss": 1.0164,
+      "step": 13881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.704519986150435e-06,
+      "loss": 1.0333,
+      "step": 13882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.703188081099636e-06,
+      "loss": 1.0338,
+      "step": 13883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.701856219061356e-06,
+      "loss": 1.0051,
+      "step": 13884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.700524400060544e-06,
+      "loss": 0.994,
+      "step": 13885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.699192624122134e-06,
+      "loss": 0.9928,
+      "step": 13886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.69786089127107e-06,
+      "loss": 0.9575,
+      "step": 13887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.696529201532293e-06,
+      "loss": 1.0299,
+      "step": 13888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.695197554930737e-06,
+      "loss": 0.9803,
+      "step": 13889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.693865951491344e-06,
+      "loss": 0.9695,
+      "step": 13890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.692534391239044e-06,
+      "loss": 1.0223,
+      "step": 13891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.691202874198783e-06,
+      "loss": 0.9355,
+      "step": 13892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.689871400395494e-06,
+      "loss": 1.0473,
+      "step": 13893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.688539969854104e-06,
+      "loss": 1.0338,
+      "step": 13894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.687208582599556e-06,
+      "loss": 0.9643,
+      "step": 13895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.685877238656776e-06,
+      "loss": 0.9935,
+      "step": 13896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.684545938050701e-06,
+      "loss": 1.05,
+      "step": 13897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.683214680806262e-06,
+      "loss": 1.0212,
+      "step": 13898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.681883466948384e-06,
+      "loss": 0.9495,
+      "step": 13899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.680552296502005e-06,
+      "loss": 0.9784,
+      "step": 13900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.679221169492046e-06,
+      "loss": 0.9592,
+      "step": 13901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.677890085943443e-06,
+      "loss": 1.048,
+      "step": 13902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.67655904588112e-06,
+      "loss": 0.9696,
+      "step": 13903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.675228049329996e-06,
+      "loss": 0.9711,
+      "step": 13904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.673897096315008e-06,
+      "loss": 0.9988,
+      "step": 13905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.672566186861074e-06,
+      "loss": 0.9835,
+      "step": 13906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.671235320993122e-06,
+      "loss": 1.0672,
+      "step": 13907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.669904498736074e-06,
+      "loss": 0.9895,
+      "step": 13908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.668573720114847e-06,
+      "loss": 0.9826,
+      "step": 13909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.667242985154373e-06,
+      "loss": 1.0187,
+      "step": 13910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.66591229387956e-06,
+      "loss": 1.0034,
+      "step": 13911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.664581646315341e-06,
+      "loss": 1.0157,
+      "step": 13912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.66325104248663e-06,
+      "loss": 0.8839,
+      "step": 13913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.661920482418336e-06,
+      "loss": 1.0018,
+      "step": 13914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.660589966135392e-06,
+      "loss": 0.9578,
+      "step": 13915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.659259493662705e-06,
+      "loss": 0.9968,
+      "step": 13916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.657929065025193e-06,
+      "loss": 1.0119,
+      "step": 13917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.656598680247775e-06,
+      "loss": 1.0139,
+      "step": 13918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.655268339355354e-06,
+      "loss": 0.9689,
+      "step": 13919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.653938042372857e-06,
+      "loss": 0.9838,
+      "step": 13920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.652607789325186e-06,
+      "loss": 1.0102,
+      "step": 13921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.651277580237261e-06,
+      "loss": 1.0063,
+      "step": 13922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.649947415133989e-06,
+      "loss": 0.9986,
+      "step": 13923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.648617294040277e-06,
+      "loss": 0.9676,
+      "step": 13924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.647287216981038e-06,
+      "loss": 0.9121,
+      "step": 13925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.64595718398118e-06,
+      "loss": 0.9761,
+      "step": 13926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.644627195065612e-06,
+      "loss": 0.9985,
+      "step": 13927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.643297250259238e-06,
+      "loss": 0.9752,
+      "step": 13928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.641967349586962e-06,
+      "loss": 1.0239,
+      "step": 13929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.640637493073696e-06,
+      "loss": 0.9626,
+      "step": 13930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.639307680744337e-06,
+      "loss": 1.046,
+      "step": 13931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.637977912623796e-06,
+      "loss": 1.0016,
+      "step": 13932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.63664818873697e-06,
+      "loss": 1.0108,
+      "step": 13933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.635318509108763e-06,
+      "loss": 1.0278,
+      "step": 13934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.633988873764076e-06,
+      "loss": 1.0003,
+      "step": 13935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.632659282727803e-06,
+      "loss": 1.0609,
+      "step": 13936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.631329736024854e-06,
+      "loss": 1.0944,
+      "step": 13937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.630000233680123e-06,
+      "loss": 1.0184,
+      "step": 13938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.628670775718504e-06,
+      "loss": 0.9795,
+      "step": 13939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.627341362164899e-06,
+      "loss": 1.0357,
+      "step": 13940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.626011993044198e-06,
+      "loss": 1.0386,
+      "step": 13941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.624682668381303e-06,
+      "loss": 1.001,
+      "step": 13942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.623353388201107e-06,
+      "loss": 1.01,
+      "step": 13943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.6220241525285e-06,
+      "loss": 0.9268,
+      "step": 13944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.620694961388378e-06,
+      "loss": 0.9712,
+      "step": 13945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.61936581480563e-06,
+      "loss": 0.971,
+      "step": 13946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.618036712805146e-06,
+      "loss": 0.9731,
+      "step": 13947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.61670765541182e-06,
+      "loss": 1.008,
+      "step": 13948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.615378642650539e-06,
+      "loss": 1.0128,
+      "step": 13949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.614049674546194e-06,
+      "loss": 1.0166,
+      "step": 13950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.612720751123673e-06,
+      "loss": 1.0392,
+      "step": 13951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.611391872407855e-06,
+      "loss": 1.0204,
+      "step": 13952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.610063038423636e-06,
+      "loss": 0.9876,
+      "step": 13953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.608734249195895e-06,
+      "loss": 1.0512,
+      "step": 13954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.607405504749523e-06,
+      "loss": 0.9796,
+      "step": 13955
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.606076805109394e-06,
+      "loss": 1.0494,
+      "step": 13956
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.604748150300395e-06,
+      "loss": 0.9667,
+      "step": 13957
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.60341954034741e-06,
+      "loss": 1.0285,
+      "step": 13958
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.602090975275316e-06,
+      "loss": 0.9671,
+      "step": 13959
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.600762455108997e-06,
+      "loss": 1.1002,
+      "step": 13960
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.599433979873331e-06,
+      "loss": 0.9848,
+      "step": 13961
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.598105549593195e-06,
+      "loss": 1.0136,
+      "step": 13962
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.59677716429347e-06,
+      "loss": 0.9618,
+      "step": 13963
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.595448823999024e-06,
+      "loss": 0.9843,
+      "step": 13964
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.594120528734745e-06,
+      "loss": 0.9479,
+      "step": 13965
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.592792278525501e-06,
+      "loss": 1.0658,
+      "step": 13966
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.591464073396165e-06,
+      "loss": 1.0279,
+      "step": 13967
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.590135913371615e-06,
+      "loss": 1.0355,
+      "step": 13968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.588807798476717e-06,
+      "loss": 1.0583,
+      "step": 13969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.587479728736351e-06,
+      "loss": 0.9874,
+      "step": 13970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.586151704175385e-06,
+      "loss": 0.9139,
+      "step": 13971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.584823724818686e-06,
+      "loss": 1.0189,
+      "step": 13972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.583495790691126e-06,
+      "loss": 0.9864,
+      "step": 13973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.582167901817569e-06,
+      "loss": 1.0401,
+      "step": 13974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.58084005822289e-06,
+      "loss": 1.0445,
+      "step": 13975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.57951225993195e-06,
+      "loss": 0.9826,
+      "step": 13976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.578184506969611e-06,
+      "loss": 1.1022,
+      "step": 13977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.576856799360748e-06,
+      "loss": 0.9441,
+      "step": 13978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.57552913713022e-06,
+      "loss": 0.9585,
+      "step": 13979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.574201520302886e-06,
+      "loss": 1.0051,
+      "step": 13980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.572873948903618e-06,
+      "loss": 1.0319,
+      "step": 13981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.57154642295727e-06,
+      "loss": 0.9403,
+      "step": 13982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.570218942488705e-06,
+      "loss": 0.9852,
+      "step": 13983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.56889150752278e-06,
+      "loss": 0.9884,
+      "step": 13984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.56756411808436e-06,
+      "loss": 1.0034,
+      "step": 13985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.566236774198299e-06,
+      "loss": 1.0948,
+      "step": 13986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.564909475889451e-06,
+      "loss": 0.9768,
+      "step": 13987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.563582223182679e-06,
+      "loss": 1.0247,
+      "step": 13988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.562255016102835e-06,
+      "loss": 0.9421,
+      "step": 13989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.560927854674778e-06,
+      "loss": 0.9551,
+      "step": 13990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.559600738923353e-06,
+      "loss": 1.0438,
+      "step": 13991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55827366887342e-06,
+      "loss": 0.9729,
+      "step": 13992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55694664454983e-06,
+      "loss": 1.0705,
+      "step": 13993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.555619665977431e-06,
+      "loss": 1.0029,
+      "step": 13994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.55429273318108e-06,
+      "loss": 0.9905,
+      "step": 13995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.552965846185622e-06,
+      "loss": 0.9908,
+      "step": 13996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.5516390050159e-06,
+      "loss": 1.0069,
+      "step": 13997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.550312209696776e-06,
+      "loss": 0.9478,
+      "step": 13998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.548985460253084e-06,
+      "loss": 1.0324,
+      "step": 13999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.547658756709679e-06,
+      "loss": 1.0485,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "eval_text_loss": 0.949796736240387,
+      "eval_text_runtime": 1.5515,
+      "eval_text_samples_per_second": 47.695,
+      "eval_text_steps_per_second": 1.289,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "eval_svit_loss": 1.0999064445495605,
+      "eval_svit_runtime": 5.6015,
+      "eval_svit_samples_per_second": 121.574,
+      "eval_svit_steps_per_second": 1.964,
+      "step": 14000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.5463320990914e-06,
+      "loss": 0.9845,
+      "step": 14001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.545005487423091e-06,
+      "loss": 1.0049,
+      "step": 14002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.543678921729603e-06,
+      "loss": 1.0295,
+      "step": 14003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.542352402035768e-06,
+      "loss": 0.9878,
+      "step": 14004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.541025928366439e-06,
+      "loss": 1.0291,
+      "step": 14005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.539699500746451e-06,
+      "loss": 1.0599,
+      "step": 14006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.538373119200642e-06,
+      "loss": 0.9638,
+      "step": 14007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.537046783753856e-06,
+      "loss": 1.0013,
+      "step": 14008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.53572049443093e-06,
+      "loss": 1.0178,
+      "step": 14009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.5343942512567e-06,
+      "loss": 0.9251,
+      "step": 14010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.533068054256005e-06,
+      "loss": 1.0078,
+      "step": 14011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.531741903453676e-06,
+      "loss": 1.0231,
+      "step": 14012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.530415798874552e-06,
+      "loss": 1.0073,
+      "step": 14013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.529089740543469e-06,
+      "loss": 1.0486,
+      "step": 14014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.527763728485253e-06,
+      "loss": 0.984,
+      "step": 14015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.526437762724745e-06,
+      "loss": 1.0596,
+      "step": 14016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.525111843286768e-06,
+      "loss": 0.9891,
+      "step": 14017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.523785970196158e-06,
+      "loss": 0.9314,
+      "step": 14018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.522460143477745e-06,
+      "loss": 1.039,
+      "step": 14019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.521134363156353e-06,
+      "loss": 1.0544,
+      "step": 14020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.519808629256816e-06,
+      "loss": 1.0446,
+      "step": 14021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.518482941803955e-06,
+      "loss": 1.0427,
+      "step": 14022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.517157300822601e-06,
+      "loss": 1.0247,
+      "step": 14023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.51583170633758e-06,
+      "loss": 1.0024,
+      "step": 14024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.51450615837371e-06,
+      "loss": 0.9427,
+      "step": 14025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.513180656955822e-06,
+      "loss": 1.0654,
+      "step": 14026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.511855202108736e-06,
+      "loss": 0.9705,
+      "step": 14027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.510529793857273e-06,
+      "loss": 0.9486,
+      "step": 14028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.509204432226254e-06,
+      "loss": 1.0221,
+      "step": 14029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.507879117240497e-06,
+      "loss": 0.9634,
+      "step": 14030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.506553848924827e-06,
+      "loss": 1.0145,
+      "step": 14031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.505228627304056e-06,
+      "loss": 0.9992,
+      "step": 14032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.503903452403008e-06,
+      "loss": 0.9832,
+      "step": 14033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.502578324246495e-06,
+      "loss": 1.0051,
+      "step": 14034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.501253242859331e-06,
+      "loss": 1.0153,
+      "step": 14035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.499928208266339e-06,
+      "loss": 1.0173,
+      "step": 14036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.498603220492325e-06,
+      "loss": 0.9909,
+      "step": 14037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.497278279562107e-06,
+      "loss": 0.9855,
+      "step": 14038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.495953385500497e-06,
+      "loss": 0.9815,
+      "step": 14039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.494628538332298e-06,
+      "loss": 1.0047,
+      "step": 14040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.493303738082332e-06,
+      "loss": 1.0098,
+      "step": 14041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.4919789847754e-06,
+      "loss": 0.9551,
+      "step": 14042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.49065427843632e-06,
+      "loss": 1.0296,
+      "step": 14043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.489329619089891e-06,
+      "loss": 1.0365,
+      "step": 14044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.48800500676092e-06,
+      "loss": 0.9921,
+      "step": 14045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.486680441474223e-06,
+      "loss": 0.9857,
+      "step": 14046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.485355923254595e-06,
+      "loss": 1.0288,
+      "step": 14047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.484031452126845e-06,
+      "loss": 0.9287,
+      "step": 14048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.482707028115777e-06,
+      "loss": 1.0636,
+      "step": 14049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.481382651246187e-06,
+      "loss": 0.9669,
+      "step": 14050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.480058321542885e-06,
+      "loss": 1.0481,
+      "step": 14051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.478734039030664e-06,
+      "loss": 0.9618,
+      "step": 14052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.477409803734334e-06,
+      "loss": 1.0184,
+      "step": 14053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.476085615678686e-06,
+      "loss": 0.9727,
+      "step": 14054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.474761474888522e-06,
+      "loss": 1.0106,
+      "step": 14055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.473437381388635e-06,
+      "loss": 0.9724,
+      "step": 14056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.472113335203823e-06,
+      "loss": 1.0295,
+      "step": 14057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.470789336358886e-06,
+      "loss": 1.0001,
+      "step": 14058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.469465384878616e-06,
+      "loss": 1.0023,
+      "step": 14059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.4681414807878e-06,
+      "loss": 0.922,
+      "step": 14060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.466817624111242e-06,
+      "loss": 1.0457,
+      "step": 14061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.465493814873726e-06,
+      "loss": 0.9644,
+      "step": 14062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.464170053100048e-06,
+      "loss": 1.0332,
+      "step": 14063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.462846338814994e-06,
+      "loss": 1.0528,
+      "step": 14064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.461522672043355e-06,
+      "loss": 1.0162,
+      "step": 14065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.460199052809923e-06,
+      "loss": 1.0153,
+      "step": 14066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.458875481139476e-06,
+      "loss": 1.0304,
+      "step": 14067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45755195705681e-06,
+      "loss": 0.9805,
+      "step": 14068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.456228480586709e-06,
+      "loss": 0.9884,
+      "step": 14069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45490505175395e-06,
+      "loss": 0.9862,
+      "step": 14070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45358167058333e-06,
+      "loss": 0.9572,
+      "step": 14071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.45225833709962e-06,
+      "loss": 1.0451,
+      "step": 14072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.450935051327608e-06,
+      "loss": 1.0181,
+      "step": 14073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.44961181329208e-06,
+      "loss": 0.9576,
+      "step": 14074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.448288623017804e-06,
+      "loss": 0.93,
+      "step": 14075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.44696548052957e-06,
+      "loss": 1.0183,
+      "step": 14076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.445642385852149e-06,
+      "loss": 1.0301,
+      "step": 14077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.444319339010326e-06,
+      "loss": 1.0003,
+      "step": 14078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.442996340028874e-06,
+      "loss": 1.0127,
+      "step": 14079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.441673388932567e-06,
+      "loss": 0.9317,
+      "step": 14080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.440350485746184e-06,
+      "loss": 1.0284,
+      "step": 14081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.439027630494498e-06,
+      "loss": 1.024,
+      "step": 14082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.437704823202279e-06,
+      "loss": 1.0234,
+      "step": 14083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.436382063894301e-06,
+      "loss": 0.9787,
+      "step": 14084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.435059352595337e-06,
+      "loss": 1.0195,
+      "step": 14085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.433736689330159e-06,
+      "loss": 1.0258,
+      "step": 14086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.432414074123533e-06,
+      "loss": 0.9619,
+      "step": 14087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.431091507000226e-06,
+      "loss": 1.0191,
+      "step": 14088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.429768987985011e-06,
+      "loss": 0.9564,
+      "step": 14089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.428446517102649e-06,
+      "loss": 1.0028,
+      "step": 14090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.427124094377913e-06,
+      "loss": 0.9736,
+      "step": 14091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.425801719835566e-06,
+      "loss": 1.0485,
+      "step": 14092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.424479393500367e-06,
+      "loss": 0.9727,
+      "step": 14093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.423157115397086e-06,
+      "loss": 0.9971,
+      "step": 14094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.421834885550478e-06,
+      "loss": 1.0195,
+      "step": 14095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.420512703985313e-06,
+      "loss": 1.0355,
+      "step": 14096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.419190570726348e-06,
+      "loss": 1.032,
+      "step": 14097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.417868485798336e-06,
+      "loss": 1.0172,
+      "step": 14098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.416546449226047e-06,
+      "loss": 0.9765,
+      "step": 14099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.415224461034231e-06,
+      "loss": 0.9909,
+      "step": 14100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.41390252124765e-06,
+      "loss": 1.0098,
+      "step": 14101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.412580629891058e-06,
+      "loss": 0.9999,
+      "step": 14102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.411258786989208e-06,
+      "loss": 0.9929,
+      "step": 14103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.409936992566858e-06,
+      "loss": 1.0674,
+      "step": 14104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.408615246648755e-06,
+      "loss": 0.9952,
+      "step": 14105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.40729354925966e-06,
+      "loss": 1.0452,
+      "step": 14106
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.40597190042432e-06,
+      "loss": 0.9857,
+      "step": 14107
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.404650300167482e-06,
+      "loss": 0.9516,
+      "step": 14108
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.403328748513903e-06,
+      "loss": 1.002,
+      "step": 14109
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.402007245488327e-06,
+      "loss": 1.0438,
+      "step": 14110
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.400685791115503e-06,
+      "loss": 0.9407,
+      "step": 14111
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.39936438542018e-06,
+      "loss": 1.0108,
+      "step": 14112
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.3980430284271e-06,
+      "loss": 1.0042,
+      "step": 14113
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.39672172016101e-06,
+      "loss": 1.0713,
+      "step": 14114
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.395400460646654e-06,
+      "loss": 1.0083,
+      "step": 14115
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.394079249908778e-06,
+      "loss": 1.0037,
+      "step": 14116
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.392758087972123e-06,
+      "loss": 0.9702,
+      "step": 14117
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.391436974861425e-06,
+      "loss": 0.9194,
+      "step": 14118
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.390115910601431e-06,
+      "loss": 1.0141,
+      "step": 14119
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38879489521688e-06,
+      "loss": 1.011,
+      "step": 14120
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.387473928732511e-06,
+      "loss": 0.9517,
+      "step": 14121
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38615301117306e-06,
+      "loss": 1.0365,
+      "step": 14122
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.38483214256326e-06,
+      "loss": 1.0476,
+      "step": 14123
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.383511322927855e-06,
+      "loss": 1.0502,
+      "step": 14124
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.382190552291573e-06,
+      "loss": 0.9477,
+      "step": 14125
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.380869830679154e-06,
+      "loss": 0.9915,
+      "step": 14126
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.37954915811533e-06,
+      "loss": 1.0371,
+      "step": 14127
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.378228534624827e-06,
+      "loss": 0.9866,
+      "step": 14128
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.376907960232386e-06,
+      "loss": 0.9419,
+      "step": 14129
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.37558743496273e-06,
+      "loss": 0.9448,
+      "step": 14130
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.374266958840595e-06,
+      "loss": 0.967,
+      "step": 14131
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.372946531890704e-06,
+      "loss": 1.0055,
+      "step": 14132
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.371626154137783e-06,
+      "loss": 0.9994,
+      "step": 14133
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.370305825606569e-06,
+      "loss": 1.0038,
+      "step": 14134
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.368985546321774e-06,
+      "loss": 0.9763,
+      "step": 14135
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.367665316308136e-06,
+      "loss": 0.9809,
+      "step": 14136
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.366345135590373e-06,
+      "loss": 0.9981,
+      "step": 14137
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.365025004193205e-06,
+      "loss": 0.9883,
+      "step": 14138
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.36370492214136e-06,
+      "loss": 1.0369,
+      "step": 14139
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.362384889459557e-06,
+      "loss": 0.9878,
+      "step": 14140
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.361064906172517e-06,
+      "loss": 1.0336,
+      "step": 14141
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.359744972304958e-06,
+      "loss": 1.0169,
+      "step": 14142
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.358425087881596e-06,
+      "loss": 0.9788,
+      "step": 14143
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.357105252927154e-06,
+      "loss": 0.9518,
+      "step": 14144
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.355785467466343e-06,
+      "loss": 1.003,
+      "step": 14145
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.354465731523887e-06,
+      "loss": 1.0215,
+      "step": 14146
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.353146045124494e-06,
+      "loss": 0.9982,
+      "step": 14147
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.351826408292878e-06,
+      "loss": 1.0124,
+      "step": 14148
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.350506821053755e-06,
+      "loss": 1.0473,
+      "step": 14149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.349187283431835e-06,
+      "loss": 1.0204,
+      "step": 14150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.347867795451828e-06,
+      "loss": 1.0176,
+      "step": 14151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.346548357138448e-06,
+      "loss": 1.0081,
+      "step": 14152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.345228968516398e-06,
+      "loss": 0.9931,
+      "step": 14153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.343909629610393e-06,
+      "loss": 0.9157,
+      "step": 14154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.342590340445139e-06,
+      "loss": 0.978,
+      "step": 14155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.341271101045336e-06,
+      "loss": 0.9837,
+      "step": 14156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.339951911435696e-06,
+      "loss": 1.0435,
+      "step": 14157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.338632771640923e-06,
+      "loss": 1.1108,
+      "step": 14158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.337313681685721e-06,
+      "loss": 0.9199,
+      "step": 14159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.33599464159479e-06,
+      "loss": 1.0457,
+      "step": 14160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.334675651392828e-06,
+      "loss": 0.9912,
+      "step": 14161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.333356711104545e-06,
+      "loss": 0.9465,
+      "step": 14162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.332037820754632e-06,
+      "loss": 1.0147,
+      "step": 14163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.330718980367798e-06,
+      "loss": 1.0592,
+      "step": 14164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.329400189968734e-06,
+      "loss": 0.9649,
+      "step": 14165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.328081449582135e-06,
+      "loss": 0.9452,
+      "step": 14166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.326762759232703e-06,
+      "loss": 0.9601,
+      "step": 14167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.325444118945131e-06,
+      "loss": 1.0034,
+      "step": 14168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.324125528744115e-06,
+      "loss": 1.013,
+      "step": 14169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.322806988654345e-06,
+      "loss": 0.9987,
+      "step": 14170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.321488498700512e-06,
+      "loss": 1.052,
+      "step": 14171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.320170058907312e-06,
+      "loss": 0.9627,
+      "step": 14172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.318851669299432e-06,
+      "loss": 0.9365,
+      "step": 14173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.317533329901564e-06,
+      "loss": 0.9513,
+      "step": 14174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.316215040738399e-06,
+      "loss": 0.9926,
+      "step": 14175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.314896801834617e-06,
+      "loss": 0.9496,
+      "step": 14176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.313578613214911e-06,
+      "loss": 1.0036,
+      "step": 14177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.312260474903963e-06,
+      "loss": 0.9881,
+      "step": 14178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.310942386926462e-06,
+      "loss": 1.0331,
+      "step": 14179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.309624349307091e-06,
+      "loss": 1.0548,
+      "step": 14180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.308306362070527e-06,
+      "loss": 1.016,
+      "step": 14181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.306988425241461e-06,
+      "loss": 0.9481,
+      "step": 14182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.305670538844567e-06,
+      "loss": 1.0052,
+      "step": 14183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.304352702904529e-06,
+      "loss": 0.9744,
+      "step": 14184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.303034917446027e-06,
+      "loss": 0.9622,
+      "step": 14185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.301717182493736e-06,
+      "loss": 0.9609,
+      "step": 14186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.300399498072335e-06,
+      "loss": 0.9887,
+      "step": 14187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.299081864206496e-06,
+      "loss": 0.9663,
+      "step": 14188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.297764280920902e-06,
+      "loss": 1.0054,
+      "step": 14189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.296446748240226e-06,
+      "loss": 1.0244,
+      "step": 14190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.295129266189134e-06,
+      "loss": 1.0436,
+      "step": 14191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.293811834792307e-06,
+      "loss": 1.0691,
+      "step": 14192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.29249445407441e-06,
+      "loss": 0.971,
+      "step": 14193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.291177124060122e-06,
+      "loss": 0.9715,
+      "step": 14194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.289859844774107e-06,
+      "loss": 1.0232,
+      "step": 14195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.288542616241032e-06,
+      "loss": 1.0465,
+      "step": 14196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.28722543848557e-06,
+      "loss": 0.9825,
+      "step": 14197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.285908311532381e-06,
+      "loss": 0.9839,
+      "step": 14198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.284591235406141e-06,
+      "loss": 0.9809,
+      "step": 14199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.283274210131509e-06,
+      "loss": 1.0464,
+      "step": 14200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.281957235733143e-06,
+      "loss": 0.9666,
+      "step": 14201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.280640312235716e-06,
+      "loss": 1.0098,
+      "step": 14202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.279323439663885e-06,
+      "loss": 1.0507,
+      "step": 14203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.278006618042314e-06,
+      "loss": 0.9899,
+      "step": 14204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.276689847395662e-06,
+      "loss": 1.0086,
+      "step": 14205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.275373127748586e-06,
+      "loss": 0.987,
+      "step": 14206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.274056459125748e-06,
+      "loss": 0.9694,
+      "step": 14207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.2727398415518e-06,
+      "loss": 0.9936,
+      "step": 14208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.271423275051409e-06,
+      "loss": 0.9519,
+      "step": 14209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.270106759649218e-06,
+      "loss": 0.9791,
+      "step": 14210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.268790295369885e-06,
+      "loss": 1.0209,
+      "step": 14211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.26747388223807e-06,
+      "loss": 1.0153,
+      "step": 14212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.266157520278418e-06,
+      "loss": 1.0164,
+      "step": 14213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.264841209515585e-06,
+      "loss": 1.0029,
+      "step": 14214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.26352494997422e-06,
+      "loss": 0.972,
+      "step": 14215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.262208741678968e-06,
+      "loss": 1.0038,
+      "step": 14216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.260892584654486e-06,
+      "loss": 1.0202,
+      "step": 14217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.259576478925421e-06,
+      "loss": 1.0132,
+      "step": 14218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.25826042451641e-06,
+      "loss": 0.9946,
+      "step": 14219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.256944421452112e-06,
+      "loss": 0.9819,
+      "step": 14220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.255628469757159e-06,
+      "loss": 0.9969,
+      "step": 14221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.254312569456207e-06,
+      "loss": 1.0619,
+      "step": 14222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.252996720573893e-06,
+      "loss": 0.9986,
+      "step": 14223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.251680923134856e-06,
+      "loss": 0.9819,
+      "step": 14224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.250365177163746e-06,
+      "loss": 1.0222,
+      "step": 14225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.249049482685189e-06,
+      "loss": 0.9466,
+      "step": 14226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.247733839723839e-06,
+      "loss": 1.0097,
+      "step": 14227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.246418248304328e-06,
+      "loss": 1.0483,
+      "step": 14228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.245102708451289e-06,
+      "loss": 1.0599,
+      "step": 14229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.243787220189363e-06,
+      "loss": 0.9999,
+      "step": 14230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.242471783543182e-06,
+      "loss": 0.9871,
+      "step": 14231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.241156398537386e-06,
+      "loss": 0.9466,
+      "step": 14232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.239841065196607e-06,
+      "loss": 1.0205,
+      "step": 14233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.23852578354547e-06,
+      "loss": 0.9716,
+      "step": 14234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.237210553608616e-06,
+      "loss": 0.9825,
+      "step": 14235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.235895375410664e-06,
+      "loss": 1.0116,
+      "step": 14236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.234580248976255e-06,
+      "loss": 0.9566,
+      "step": 14237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.233265174330014e-06,
+      "loss": 1.0024,
+      "step": 14238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.231950151496561e-06,
+      "loss": 0.9858,
+      "step": 14239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.230635180500534e-06,
+      "loss": 1.0626,
+      "step": 14240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.229320261366552e-06,
+      "loss": 1.0001,
+      "step": 14241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.228005394119242e-06,
+      "loss": 0.9353,
+      "step": 14242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.226690578783224e-06,
+      "loss": 1.0827,
+      "step": 14243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.225375815383122e-06,
+      "loss": 1.0425,
+      "step": 14244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.224061103943562e-06,
+      "loss": 0.9587,
+      "step": 14245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.222746444489157e-06,
+      "loss": 1.0343,
+      "step": 14246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.221431837044534e-06,
+      "loss": 0.9545,
+      "step": 14247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.22011728163431e-06,
+      "loss": 0.9865,
+      "step": 14248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.218802778283096e-06,
+      "loss": 1.0027,
+      "step": 14249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.217488327015518e-06,
+      "loss": 1.0146,
+      "step": 14250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.216173927856187e-06,
+      "loss": 0.9913,
+      "step": 14251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.214859580829721e-06,
+      "loss": 1.026,
+      "step": 14252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.21354528596073e-06,
+      "loss": 0.9675,
+      "step": 14253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.212231043273825e-06,
+      "loss": 0.9967,
+      "step": 14254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.210916852793627e-06,
+      "loss": 0.9677,
+      "step": 14255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.209602714544737e-06,
+      "loss": 1.0421,
+      "step": 14256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.208288628551771e-06,
+      "loss": 1.0312,
+      "step": 14257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.206974594839339e-06,
+      "loss": 0.9541,
+      "step": 14258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.205660613432041e-06,
+      "loss": 0.9566,
+      "step": 14259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.204346684354493e-06,
+      "loss": 1.0171,
+      "step": 14260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.203032807631296e-06,
+      "loss": 1.0233,
+      "step": 14261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.201718983287058e-06,
+      "loss": 0.9844,
+      "step": 14262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.20040521134638e-06,
+      "loss": 0.9809,
+      "step": 14263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.199091491833865e-06,
+      "loss": 0.9576,
+      "step": 14264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.19777782477412e-06,
+      "loss": 1.0506,
+      "step": 14265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.196464210191736e-06,
+      "loss": 0.9907,
+      "step": 14266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.195150648111325e-06,
+      "loss": 1.004,
+      "step": 14267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.193837138557483e-06,
+      "loss": 1.0005,
+      "step": 14268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.192523681554802e-06,
+      "loss": 0.9607,
+      "step": 14269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.191210277127883e-06,
+      "loss": 0.9677,
+      "step": 14270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.189896925301323e-06,
+      "loss": 0.9797,
+      "step": 14271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.188583626099718e-06,
+      "loss": 0.966,
+      "step": 14272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.187270379547661e-06,
+      "loss": 0.9348,
+      "step": 14273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.185957185669743e-06,
+      "loss": 1.0103,
+      "step": 14274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.18464404449056e-06,
+      "loss": 1.0609,
+      "step": 14275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.1833309560347e-06,
+      "loss": 0.9992,
+      "step": 14276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.182017920326758e-06,
+      "loss": 0.9976,
+      "step": 14277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.18070493739132e-06,
+      "loss": 0.9756,
+      "step": 14278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.179392007252972e-06,
+      "loss": 0.9529,
+      "step": 14279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.178079129936306e-06,
+      "loss": 0.986,
+      "step": 14280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.176766305465904e-06,
+      "loss": 1.0162,
+      "step": 14281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.175453533866354e-06,
+      "loss": 1.0357,
+      "step": 14282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.174140815162242e-06,
+      "loss": 0.9826,
+      "step": 14283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.172828149378146e-06,
+      "loss": 1.0053,
+      "step": 14284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.171515536538653e-06,
+      "loss": 1.0099,
+      "step": 14285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.170202976668346e-06,
+      "loss": 0.9803,
+      "step": 14286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.168890469791797e-06,
+      "loss": 0.9727,
+      "step": 14287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.167578015933594e-06,
+      "loss": 0.9729,
+      "step": 14288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.166265615118311e-06,
+      "loss": 0.9202,
+      "step": 14289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.164953267370526e-06,
+      "loss": 0.9628,
+      "step": 14290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.16364097271482e-06,
+      "loss": 0.9563,
+      "step": 14291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.162328731175757e-06,
+      "loss": 0.9852,
+      "step": 14292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.161016542777924e-06,
+      "loss": 1.0103,
+      "step": 14293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.159704407545885e-06,
+      "loss": 1.0055,
+      "step": 14294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.158392325504221e-06,
+      "loss": 1.0099,
+      "step": 14295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.157080296677499e-06,
+      "loss": 1.0169,
+      "step": 14296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.155768321090285e-06,
+      "loss": 0.9971,
+      "step": 14297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.154456398767158e-06,
+      "loss": 1.0089,
+      "step": 14298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.153144529732678e-06,
+      "loss": 0.9854,
+      "step": 14299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.151832714011417e-06,
+      "loss": 1.0423,
+      "step": 14300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.150520951627941e-06,
+      "loss": 0.9653,
+      "step": 14301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.149209242606813e-06,
+      "loss": 0.945,
+      "step": 14302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.1478975869726e-06,
+      "loss": 0.965,
+      "step": 14303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.146585984749863e-06,
+      "loss": 0.9826,
+      "step": 14304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.145274435963168e-06,
+      "loss": 0.9374,
+      "step": 14305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.143962940637075e-06,
+      "loss": 1.0328,
+      "step": 14306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.142651498796143e-06,
+      "loss": 1.0105,
+      "step": 14307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.141340110464934e-06,
+      "loss": 0.9996,
+      "step": 14308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.140028775668e-06,
+      "loss": 0.9974,
+      "step": 14309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.138717494429907e-06,
+      "loss": 1.015,
+      "step": 14310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.137406266775208e-06,
+      "loss": 1.0058,
+      "step": 14311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.136095092728454e-06,
+      "loss": 0.958,
+      "step": 14312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.134783972314207e-06,
+      "loss": 0.9666,
+      "step": 14313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.133472905557013e-06,
+      "loss": 0.9453,
+      "step": 14314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.132161892481434e-06,
+      "loss": 1.0417,
+      "step": 14315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.130850933112014e-06,
+      "loss": 0.9885,
+      "step": 14316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.129540027473304e-06,
+      "loss": 1.0013,
+      "step": 14317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.128229175589858e-06,
+      "loss": 0.9662,
+      "step": 14318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.126918377486215e-06,
+      "loss": 1.0494,
+      "step": 14319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.125607633186932e-06,
+      "loss": 0.9692,
+      "step": 14320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.124296942716553e-06,
+      "loss": 1.0038,
+      "step": 14321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.122986306099617e-06,
+      "loss": 1.0509,
+      "step": 14322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.12167572336068e-06,
+      "loss": 0.9369,
+      "step": 14323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.120365194524272e-06,
+      "loss": 0.9652,
+      "step": 14324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.119054719614948e-06,
+      "loss": 1.0162,
+      "step": 14325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.117744298657242e-06,
+      "loss": 1.0022,
+      "step": 14326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.116433931675694e-06,
+      "loss": 1.032,
+      "step": 14327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.115123618694848e-06,
+      "loss": 1.0562,
+      "step": 14328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.113813359739236e-06,
+      "loss": 0.9352,
+      "step": 14329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.1125031548334e-06,
+      "loss": 1.0366,
+      "step": 14330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.111193004001878e-06,
+      "loss": 1.0522,
+      "step": 14331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.109882907269197e-06,
+      "loss": 1.0182,
+      "step": 14332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.108572864659902e-06,
+      "loss": 0.9815,
+      "step": 14333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.107262876198517e-06,
+      "loss": 1.0332,
+      "step": 14334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10595294190958e-06,
+      "loss": 1.0309,
+      "step": 14335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.104643061817618e-06,
+      "loss": 0.9759,
+      "step": 14336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.103333235947165e-06,
+      "loss": 0.9438,
+      "step": 14337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10202346432275e-06,
+      "loss": 1.0343,
+      "step": 14338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.100713746968894e-06,
+      "loss": 1.0638,
+      "step": 14339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.099404083910137e-06,
+      "loss": 1.0849,
+      "step": 14340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.098094475170996e-06,
+      "loss": 0.9933,
+      "step": 14341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.096784920775993e-06,
+      "loss": 0.9844,
+      "step": 14342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.095475420749664e-06,
+      "loss": 0.9513,
+      "step": 14343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.094165975116522e-06,
+      "loss": 1.0524,
+      "step": 14344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.092856583901095e-06,
+      "loss": 0.9889,
+      "step": 14345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.091547247127901e-06,
+      "loss": 1.0009,
+      "step": 14346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.090237964821455e-06,
+      "loss": 0.9718,
+      "step": 14347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.088928737006288e-06,
+      "loss": 0.9837,
+      "step": 14348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.087619563706907e-06,
+      "loss": 1.0057,
+      "step": 14349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.086310444947837e-06,
+      "loss": 0.959,
+      "step": 14350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.085001380753588e-06,
+      "loss": 1.0055,
+      "step": 14351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.083692371148676e-06,
+      "loss": 0.9547,
+      "step": 14352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.08238341615762e-06,
+      "loss": 0.9268,
+      "step": 14353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.081074515804927e-06,
+      "loss": 1.0,
+      "step": 14354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.079765670115112e-06,
+      "loss": 1.0294,
+      "step": 14355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.078456879112685e-06,
+      "loss": 0.9866,
+      "step": 14356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.077148142822152e-06,
+      "loss": 1.0743,
+      "step": 14357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.075839461268028e-06,
+      "loss": 0.9622,
+      "step": 14358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.074530834474818e-06,
+      "loss": 0.9865,
+      "step": 14359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.073222262467025e-06,
+      "loss": 1.0106,
+      "step": 14360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.071913745269163e-06,
+      "loss": 0.9116,
+      "step": 14361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.070605282905731e-06,
+      "loss": 0.9318,
+      "step": 14362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.069296875401233e-06,
+      "loss": 1.0192,
+      "step": 14363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.067988522780174e-06,
+      "loss": 0.9792,
+      "step": 14364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.066680225067052e-06,
+      "loss": 1.0359,
+      "step": 14365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.065371982286372e-06,
+      "loss": 1.0177,
+      "step": 14366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.064063794462626e-06,
+      "loss": 0.9888,
+      "step": 14367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.062755661620323e-06,
+      "loss": 1.0493,
+      "step": 14368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.061447583783953e-06,
+      "loss": 0.9855,
+      "step": 14369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.060139560978012e-06,
+      "loss": 1.0079,
+      "step": 14370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.058831593227002e-06,
+      "loss": 0.9675,
+      "step": 14371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.057523680555412e-06,
+      "loss": 1.0344,
+      "step": 14372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.056215822987738e-06,
+      "loss": 0.9985,
+      "step": 14373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.054908020548468e-06,
+      "loss": 0.9473,
+      "step": 14374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.053600273262096e-06,
+      "loss": 0.9341,
+      "step": 14375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.052292581153116e-06,
+      "loss": 0.8844,
+      "step": 14376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.050984944246009e-06,
+      "loss": 1.0533,
+      "step": 14377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.049677362565273e-06,
+      "loss": 0.9965,
+      "step": 14378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.048369836135389e-06,
+      "loss": 0.9876,
+      "step": 14379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.047062364980838e-06,
+      "loss": 1.0098,
+      "step": 14380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.045754949126118e-06,
+      "loss": 0.9667,
+      "step": 14381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.044447588595703e-06,
+      "loss": 1.0031,
+      "step": 14382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.043140283414083e-06,
+      "loss": 0.9678,
+      "step": 14383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.041833033605736e-06,
+      "loss": 0.9967,
+      "step": 14384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.040525839195137e-06,
+      "loss": 0.9949,
+      "step": 14385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.03921870020678e-06,
+      "loss": 0.9802,
+      "step": 14386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.03791161666513e-06,
+      "loss": 0.99,
+      "step": 14387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.036604588594672e-06,
+      "loss": 0.9893,
+      "step": 14388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.035297616019886e-06,
+      "loss": 1.0601,
+      "step": 14389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.033990698965237e-06,
+      "loss": 1.0463,
+      "step": 14390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.032683837455211e-06,
+      "loss": 1.0384,
+      "step": 14391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.031377031514274e-06,
+      "loss": 0.9404,
+      "step": 14392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.030070281166905e-06,
+      "loss": 1.0,
+      "step": 14393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.02876358643757e-06,
+      "loss": 0.929,
+      "step": 14394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.027456947350738e-06,
+      "loss": 0.9951,
+      "step": 14395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.026150363930886e-06,
+      "loss": 0.9337,
+      "step": 14396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.024843836202475e-06,
+      "loss": 0.9423,
+      "step": 14397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.02353736418998e-06,
+      "loss": 1.0158,
+      "step": 14398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.022230947917863e-06,
+      "loss": 0.9965,
+      "step": 14399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.020924587410587e-06,
+      "loss": 1.0042,
+      "step": 14400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.019618282692621e-06,
+      "loss": 0.927,
+      "step": 14401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.018312033788421e-06,
+      "loss": 0.9377,
+      "step": 14402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.017005840722461e-06,
+      "loss": 0.96,
+      "step": 14403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.015699703519194e-06,
+      "loss": 1.0191,
+      "step": 14404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.014393622203077e-06,
+      "loss": 1.0542,
+      "step": 14405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.013087596798577e-06,
+      "loss": 0.8939,
+      "step": 14406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.011781627330147e-06,
+      "loss": 1.0817,
+      "step": 14407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.010475713822249e-06,
+      "loss": 1.0426,
+      "step": 14408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.009169856299334e-06,
+      "loss": 1.016,
+      "step": 14409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.007864054785856e-06,
+      "loss": 0.9619,
+      "step": 14410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.006558309306274e-06,
+      "loss": 1.0347,
+      "step": 14411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.005252619885035e-06,
+      "loss": 1.0698,
+      "step": 14412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.003946986546597e-06,
+      "loss": 0.9575,
+      "step": 14413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.002641409315407e-06,
+      "loss": 0.9863,
+      "step": 14414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.001335888215912e-06,
+      "loss": 1.0124,
+      "step": 14415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.000030423272566e-06,
+      "loss": 0.9936,
+      "step": 14416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.99872501450981e-06,
+      "loss": 0.9467,
+      "step": 14417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.9974196619521e-06,
+      "loss": 1.0017,
+      "step": 14418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.996114365623874e-06,
+      "loss": 1.0251,
+      "step": 14419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.994809125549577e-06,
+      "loss": 0.9664,
+      "step": 14420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.993503941753656e-06,
+      "loss": 1.0086,
+      "step": 14421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.992198814260548e-06,
+      "loss": 1.0557,
+      "step": 14422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.990893743094698e-06,
+      "loss": 0.9701,
+      "step": 14423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.989588728280548e-06,
+      "loss": 0.9997,
+      "step": 14424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.988283769842528e-06,
+      "loss": 0.9844,
+      "step": 14425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.986978867805087e-06,
+      "loss": 0.9845,
+      "step": 14426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.985674022192656e-06,
+      "loss": 1.0063,
+      "step": 14427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.984369233029672e-06,
+      "loss": 0.9576,
+      "step": 14428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.983064500340568e-06,
+      "loss": 0.9685,
+      "step": 14429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.981759824149782e-06,
+      "loss": 0.9905,
+      "step": 14430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.980455204481744e-06,
+      "loss": 1.0344,
+      "step": 14431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.97915064136089e-06,
+      "loss": 1.0277,
+      "step": 14432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.97784613481164e-06,
+      "loss": 0.991,
+      "step": 14433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.976541684858434e-06,
+      "loss": 0.9992,
+      "step": 14434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.975237291525694e-06,
+      "loss": 0.9723,
+      "step": 14435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.973932954837854e-06,
+      "loss": 1.0487,
+      "step": 14436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.972628674819337e-06,
+      "loss": 0.976,
+      "step": 14437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.971324451494565e-06,
+      "loss": 0.9579,
+      "step": 14438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.97002028488797e-06,
+      "loss": 1.0189,
+      "step": 14439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.968716175023964e-06,
+      "loss": 0.9892,
+      "step": 14440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.96741212192698e-06,
+      "loss": 1.0296,
+      "step": 14441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.966108125621436e-06,
+      "loss": 0.9898,
+      "step": 14442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.964804186131745e-06,
+      "loss": 1.0413,
+      "step": 14443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.963500303482337e-06,
+      "loss": 1.0435,
+      "step": 14444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.96219647769762e-06,
+      "loss": 0.9678,
+      "step": 14445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.960892708802018e-06,
+      "loss": 0.9667,
+      "step": 14446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.959588996819947e-06,
+      "loss": 0.9563,
+      "step": 14447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.958285341775815e-06,
+      "loss": 0.9633,
+      "step": 14448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.956981743694042e-06,
+      "loss": 0.976,
+      "step": 14449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.955678202599034e-06,
+      "loss": 0.9711,
+      "step": 14450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.954374718515209e-06,
+      "loss": 0.995,
+      "step": 14451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.953071291466978e-06,
+      "loss": 0.9604,
+      "step": 14452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.95176792147874e-06,
+      "loss": 1.0311,
+      "step": 14453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.950464608574916e-06,
+      "loss": 0.9946,
+      "step": 14454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.949161352779906e-06,
+      "loss": 0.9357,
+      "step": 14455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.947858154118117e-06,
+      "loss": 0.9691,
+      "step": 14456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.946555012613957e-06,
+      "loss": 1.0143,
+      "step": 14457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.945251928291826e-06,
+      "loss": 0.9981,
+      "step": 14458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.943948901176132e-06,
+      "loss": 0.9924,
+      "step": 14459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.942645931291268e-06,
+      "loss": 1.0082,
+      "step": 14460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.941343018661646e-06,
+      "loss": 1.0386,
+      "step": 14461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.940040163311659e-06,
+      "loss": 1.0125,
+      "step": 14462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.938737365265704e-06,
+      "loss": 0.9405,
+      "step": 14463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.937434624548185e-06,
+      "loss": 0.992,
+      "step": 14464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.936131941183492e-06,
+      "loss": 0.9178,
+      "step": 14465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.934829315196027e-06,
+      "loss": 0.9308,
+      "step": 14466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.933526746610177e-06,
+      "loss": 0.9379,
+      "step": 14467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.93222423545034e-06,
+      "loss": 0.9561,
+      "step": 14468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.930921781740909e-06,
+      "loss": 1.0416,
+      "step": 14469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.92961938550627e-06,
+      "loss": 0.9739,
+      "step": 14470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.928317046770819e-06,
+      "loss": 0.9535,
+      "step": 14471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.927014765558943e-06,
+      "loss": 0.9762,
+      "step": 14472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.925712541895026e-06,
+      "loss": 1.0085,
+      "step": 14473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.924410375803461e-06,
+      "loss": 0.9672,
+      "step": 14474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.923108267308629e-06,
+      "loss": 1.0003,
+      "step": 14475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.921806216434919e-06,
+      "loss": 0.9382,
+      "step": 14476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.920504223206712e-06,
+      "loss": 0.9869,
+      "step": 14477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.919202287648386e-06,
+      "loss": 1.0238,
+      "step": 14478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.917900409784331e-06,
+      "loss": 0.9923,
+      "step": 14479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.916598589638918e-06,
+      "loss": 0.9537,
+      "step": 14480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.915296827236537e-06,
+      "loss": 1.0404,
+      "step": 14481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.91399512260156e-06,
+      "loss": 0.9779,
+      "step": 14482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.91269347575836e-06,
+      "loss": 0.917,
+      "step": 14483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.911391886731324e-06,
+      "loss": 0.9941,
+      "step": 14484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.910090355544818e-06,
+      "loss": 0.9614,
+      "step": 14485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.908788882223221e-06,
+      "loss": 0.9257,
+      "step": 14486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.907487466790903e-06,
+      "loss": 1.0432,
+      "step": 14487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.90618610927223e-06,
+      "loss": 0.9962,
+      "step": 14488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.904884809691584e-06,
+      "loss": 1.0062,
+      "step": 14489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.903583568073325e-06,
+      "loss": 1.0107,
+      "step": 14490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.902282384441831e-06,
+      "loss": 0.9438,
+      "step": 14491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.900981258821462e-06,
+      "loss": 0.9973,
+      "step": 14492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.899680191236584e-06,
+      "loss": 0.9963,
+      "step": 14493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.898379181711568e-06,
+      "loss": 1.0224,
+      "step": 14494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.89707823027077e-06,
+      "loss": 0.99,
+      "step": 14495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.89577733693856e-06,
+      "loss": 1.0061,
+      "step": 14496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.894476501739297e-06,
+      "loss": 0.9881,
+      "step": 14497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.893175724697339e-06,
+      "loss": 0.9807,
+      "step": 14498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.891875005837053e-06,
+      "loss": 0.9629,
+      "step": 14499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.890574345182794e-06,
+      "loss": 0.9429,
+      "step": 14500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.889273742758913e-06,
+      "loss": 1.006,
+      "step": 14501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.88797319858978e-06,
+      "loss": 0.9891,
+      "step": 14502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.886672712699737e-06,
+      "loss": 1.0538,
+      "step": 14503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.885372285113148e-06,
+      "loss": 0.9797,
+      "step": 14504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.88407191585436e-06,
+      "loss": 0.9999,
+      "step": 14505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.882771604947725e-06,
+      "loss": 0.9897,
+      "step": 14506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.8814713524176e-06,
+      "loss": 0.9926,
+      "step": 14507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.880171158288327e-06,
+      "loss": 0.9146,
+      "step": 14508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.878871022584263e-06,
+      "loss": 1.0129,
+      "step": 14509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.877570945329753e-06,
+      "loss": 0.9654,
+      "step": 14510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.876270926549137e-06,
+      "loss": 0.9635,
+      "step": 14511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.874970966266772e-06,
+      "loss": 1.0165,
+      "step": 14512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.873671064506992e-06,
+      "loss": 0.9991,
+      "step": 14513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.872371221294148e-06,
+      "loss": 1.0244,
+      "step": 14514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.871071436652577e-06,
+      "loss": 1.0077,
+      "step": 14515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.869771710606619e-06,
+      "loss": 1.0127,
+      "step": 14516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.86847204318062e-06,
+      "loss": 0.9184,
+      "step": 14517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.867172434398915e-06,
+      "loss": 0.9547,
+      "step": 14518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.865872884285844e-06,
+      "loss": 0.9777,
+      "step": 14519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.864573392865744e-06,
+      "loss": 0.9679,
+      "step": 14520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.863273960162945e-06,
+      "loss": 1.007,
+      "step": 14521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.861974586201787e-06,
+      "loss": 0.9236,
+      "step": 14522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.860675271006602e-06,
+      "loss": 0.9889,
+      "step": 14523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.859376014601726e-06,
+      "loss": 0.9686,
+      "step": 14524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.858076817011486e-06,
+      "loss": 0.97,
+      "step": 14525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.85677767826021e-06,
+      "loss": 0.9321,
+      "step": 14526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.855478598372231e-06,
+      "loss": 0.9941,
+      "step": 14527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.854179577371875e-06,
+      "loss": 1.0707,
+      "step": 14528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.852880615283472e-06,
+      "loss": 0.9167,
+      "step": 14529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.851581712131346e-06,
+      "loss": 0.9844,
+      "step": 14530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.850282867939819e-06,
+      "loss": 0.9934,
+      "step": 14531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.848984082733219e-06,
+      "loss": 1.0416,
+      "step": 14532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.847685356535862e-06,
+      "loss": 1.0393,
+      "step": 14533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.846386689372076e-06,
+      "loss": 1.0936,
+      "step": 14534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.84508808126618e-06,
+      "loss": 1.0059,
+      "step": 14535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.843789532242487e-06,
+      "loss": 0.9584,
+      "step": 14536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.842491042325326e-06,
+      "loss": 1.0091,
+      "step": 14537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.841192611539e-06,
+      "loss": 0.945,
+      "step": 14538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.839894239907839e-06,
+      "loss": 0.9714,
+      "step": 14539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.838595927456149e-06,
+      "loss": 0.9811,
+      "step": 14540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.837297674208245e-06,
+      "loss": 1.0085,
+      "step": 14541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.835999480188442e-06,
+      "loss": 1.0127,
+      "step": 14542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.834701345421045e-06,
+      "loss": 0.9689,
+      "step": 14543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.833403269930375e-06,
+      "loss": 0.9942,
+      "step": 14544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.832105253740732e-06,
+      "loss": 0.9762,
+      "step": 14545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.830807296876423e-06,
+      "loss": 1.0023,
+      "step": 14546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.829509399361762e-06,
+      "loss": 0.9467,
+      "step": 14547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.828211561221052e-06,
+      "loss": 1.0174,
+      "step": 14548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.826913782478597e-06,
+      "loss": 1.0048,
+      "step": 14549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.825616063158701e-06,
+      "loss": 0.8959,
+      "step": 14550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.824318403285666e-06,
+      "loss": 1.0137,
+      "step": 14551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.823020802883792e-06,
+      "loss": 1.021,
+      "step": 14552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.82172326197738e-06,
+      "loss": 0.9748,
+      "step": 14553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.820425780590735e-06,
+      "loss": 0.9931,
+      "step": 14554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.819128358748148e-06,
+      "loss": 1.0164,
+      "step": 14555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.817830996473914e-06,
+      "loss": 1.0793,
+      "step": 14556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.816533693792337e-06,
+      "loss": 0.9264,
+      "step": 14557
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.815236450727705e-06,
+      "loss": 0.9733,
+      "step": 14558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.813939267304317e-06,
+      "loss": 1.0556,
+      "step": 14559
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.812642143546458e-06,
+      "loss": 0.9922,
+      "step": 14560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.811345079478426e-06,
+      "loss": 1.0415,
+      "step": 14561
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.810048075124509e-06,
+      "loss": 0.997,
+      "step": 14562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.808751130508995e-06,
+      "loss": 0.9471,
+      "step": 14563
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.807454245656171e-06,
+      "loss": 0.9795,
+      "step": 14564
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.806157420590328e-06,
+      "loss": 1.0189,
+      "step": 14565
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.804860655335747e-06,
+      "loss": 0.9713,
+      "step": 14566
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.803563949916717e-06,
+      "loss": 1.025,
+      "step": 14567
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.80226730435752e-06,
+      "loss": 1.0225,
+      "step": 14568
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.800970718682435e-06,
+      "loss": 0.9371,
+      "step": 14569
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.79967419291575e-06,
+      "loss": 1.005,
+      "step": 14570
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.798377727081736e-06,
+      "loss": 0.9991,
+      "step": 14571
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.79708132120468e-06,
+      "loss": 0.9836,
+      "step": 14572
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.79578497530886e-06,
+      "loss": 0.9683,
+      "step": 14573
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.794488689418543e-06,
+      "loss": 1.0191,
+      "step": 14574
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.793192463558015e-06,
+      "loss": 0.9142,
+      "step": 14575
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.791896297751543e-06,
+      "loss": 1.0295,
+      "step": 14576
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.79060019202341e-06,
+      "loss": 0.9803,
+      "step": 14577
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.789304146397881e-06,
+      "loss": 0.9315,
+      "step": 14578
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.788008160899226e-06,
+      "loss": 0.9936,
+      "step": 14579
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.78671223555172e-06,
+      "loss": 0.969,
+      "step": 14580
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.785416370379626e-06,
+      "loss": 0.963,
+      "step": 14581
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.784120565407219e-06,
+      "loss": 0.9661,
+      "step": 14582
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.782824820658761e-06,
+      "loss": 1.0199,
+      "step": 14583
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.781529136158515e-06,
+      "loss": 0.9821,
+      "step": 14584
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.780233511930751e-06,
+      "loss": 0.9813,
+      "step": 14585
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.778937947999728e-06,
+      "loss": 0.9319,
+      "step": 14586
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.777642444389714e-06,
+      "loss": 1.0398,
+      "step": 14587
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.776347001124963e-06,
+      "loss": 0.9776,
+      "step": 14588
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.775051618229735e-06,
+      "loss": 1.0264,
+      "step": 14589
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.773756295728296e-06,
+      "loss": 0.9841,
+      "step": 14590
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.772461033644895e-06,
+      "loss": 0.9994,
+      "step": 14591
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.771165832003797e-06,
+      "loss": 0.964,
+      "step": 14592
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76987069082925e-06,
+      "loss": 0.9527,
+      "step": 14593
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.768575610145509e-06,
+      "loss": 1.0176,
+      "step": 14594
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.767280589976832e-06,
+      "loss": 0.9661,
+      "step": 14595
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.765985630347465e-06,
+      "loss": 1.0125,
+      "step": 14596
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.764690731281665e-06,
+      "loss": 0.9237,
+      "step": 14597
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76339589280368e-06,
+      "loss": 1.0076,
+      "step": 14598
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.76210111493775e-06,
+      "loss": 1.051,
+      "step": 14599
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.760806397708133e-06,
+      "loss": 0.9908,
+      "step": 14600
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.759511741139068e-06,
+      "loss": 1.0088,
+      "step": 14601
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.758217145254809e-06,
+      "loss": 0.9265,
+      "step": 14602
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.756922610079593e-06,
+      "loss": 0.9533,
+      "step": 14603
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.755628135637659e-06,
+      "loss": 0.9538,
+      "step": 14604
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.75433372195326e-06,
+      "loss": 0.9931,
+      "step": 14605
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.75303936905063e-06,
+      "loss": 1.0345,
+      "step": 14606
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.751745076954006e-06,
+      "loss": 0.9734,
+      "step": 14607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.750450845687634e-06,
+      "loss": 0.9434,
+      "step": 14608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.74915667527574e-06,
+      "loss": 1.0029,
+      "step": 14609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.747862565742571e-06,
+      "loss": 0.9925,
+      "step": 14610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.746568517112353e-06,
+      "loss": 0.9852,
+      "step": 14611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.745274529409329e-06,
+      "loss": 0.9213,
+      "step": 14612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.743980602657725e-06,
+      "loss": 0.9965,
+      "step": 14613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.742686736881773e-06,
+      "loss": 0.9693,
+      "step": 14614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.741392932105703e-06,
+      "loss": 0.9953,
+      "step": 14615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.740099188353748e-06,
+      "loss": 0.9984,
+      "step": 14616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.738805505650132e-06,
+      "loss": 0.9698,
+      "step": 14617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.737511884019085e-06,
+      "loss": 1.0716,
+      "step": 14618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.736218323484829e-06,
+      "loss": 1.0227,
+      "step": 14619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.73492482407159e-06,
+      "loss": 0.9396,
+      "step": 14620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.733631385803592e-06,
+      "loss": 0.9597,
+      "step": 14621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.732338008705061e-06,
+      "loss": 0.9832,
+      "step": 14622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.731044692800213e-06,
+      "loss": 1.0329,
+      "step": 14623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.729751438113267e-06,
+      "loss": 1.0037,
+      "step": 14624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.728458244668448e-06,
+      "loss": 0.9358,
+      "step": 14625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.727165112489966e-06,
+      "loss": 1.0563,
+      "step": 14626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.725872041602044e-06,
+      "loss": 0.9807,
+      "step": 14627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.724579032028898e-06,
+      "loss": 0.9813,
+      "step": 14628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.723286083794732e-06,
+      "loss": 0.9195,
+      "step": 14629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.72199319692377e-06,
+      "loss": 0.9841,
+      "step": 14630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.720700371440223e-06,
+      "loss": 0.9751,
+      "step": 14631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.719407607368293e-06,
+      "loss": 0.9541,
+      "step": 14632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.718114904732202e-06,
+      "loss": 1.0321,
+      "step": 14633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.71682226355615e-06,
+      "loss": 0.9815,
+      "step": 14634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.715529683864346e-06,
+      "loss": 1.0448,
+      "step": 14635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.714237165681e-06,
+      "loss": 0.9905,
+      "step": 14636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.712944709030307e-06,
+      "loss": 0.9936,
+      "step": 14637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.711652313936482e-06,
+      "loss": 1.0466,
+      "step": 14638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.710359980423721e-06,
+      "loss": 1.0241,
+      "step": 14639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.709067708516234e-06,
+      "loss": 0.9516,
+      "step": 14640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.707775498238213e-06,
+      "loss": 0.9475,
+      "step": 14641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.706483349613858e-06,
+      "loss": 0.9678,
+      "step": 14642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.705191262667371e-06,
+      "loss": 0.9568,
+      "step": 14643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.703899237422946e-06,
+      "loss": 0.9616,
+      "step": 14644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.702607273904783e-06,
+      "loss": 1.0366,
+      "step": 14645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.70131537213707e-06,
+      "loss": 0.9589,
+      "step": 14646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.700023532144003e-06,
+      "loss": 1.0057,
+      "step": 14647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.69873175394978e-06,
+      "loss": 0.9612,
+      "step": 14648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.697440037578581e-06,
+      "loss": 0.9516,
+      "step": 14649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.696148383054609e-06,
+      "loss": 0.9208,
+      "step": 14650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.694856790402044e-06,
+      "loss": 0.9633,
+      "step": 14651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.693565259645077e-06,
+      "loss": 1.0147,
+      "step": 14652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.692273790807892e-06,
+      "loss": 0.9805,
+      "step": 14653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.690982383914677e-06,
+      "loss": 0.9329,
+      "step": 14654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.689691038989619e-06,
+      "loss": 1.0041,
+      "step": 14655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.688399756056896e-06,
+      "loss": 0.9669,
+      "step": 14656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.687108535140687e-06,
+      "loss": 0.9714,
+      "step": 14657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.685817376265182e-06,
+      "loss": 1.0457,
+      "step": 14658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.684526279454552e-06,
+      "loss": 0.998,
+      "step": 14659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.683235244732983e-06,
+      "loss": 0.9916,
+      "step": 14660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.681944272124648e-06,
+      "loss": 0.9701,
+      "step": 14661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.680653361653723e-06,
+      "loss": 1.0157,
+      "step": 14662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.679362513344385e-06,
+      "loss": 1.0112,
+      "step": 14663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.678071727220802e-06,
+      "loss": 1.035,
+      "step": 14664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.676781003307156e-06,
+      "loss": 0.9434,
+      "step": 14665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.675490341627613e-06,
+      "loss": 0.9406,
+      "step": 14666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.674199742206341e-06,
+      "loss": 1.0024,
+      "step": 14667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.672909205067514e-06,
+      "loss": 1.0144,
+      "step": 14668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.671618730235296e-06,
+      "loss": 1.0109,
+      "step": 14669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.670328317733858e-06,
+      "loss": 0.9591,
+      "step": 14670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.669037967587365e-06,
+      "loss": 1.042,
+      "step": 14671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.667747679819975e-06,
+      "loss": 0.9956,
+      "step": 14672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.666457454455861e-06,
+      "loss": 0.9715,
+      "step": 14673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.665167291519175e-06,
+      "loss": 0.9591,
+      "step": 14674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.663877191034087e-06,
+      "loss": 0.949,
+      "step": 14675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.662587153024754e-06,
+      "loss": 1.0017,
+      "step": 14676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.661297177515327e-06,
+      "loss": 0.9826,
+      "step": 14677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.660007264529978e-06,
+      "loss": 0.9983,
+      "step": 14678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.65871741409285e-06,
+      "loss": 0.9606,
+      "step": 14679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.657427626228106e-06,
+      "loss": 1.0446,
+      "step": 14680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.656137900959896e-06,
+      "loss": 0.9691,
+      "step": 14681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.654848238312373e-06,
+      "loss": 1.0134,
+      "step": 14682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.653558638309692e-06,
+      "loss": 0.9524,
+      "step": 14683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.652269100975996e-06,
+      "loss": 0.9948,
+      "step": 14684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.650979626335445e-06,
+      "loss": 0.9884,
+      "step": 14685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.649690214412181e-06,
+      "loss": 1.0079,
+      "step": 14686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.648400865230347e-06,
+      "loss": 1.0328,
+      "step": 14687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.647111578814099e-06,
+      "loss": 0.9912,
+      "step": 14688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.64582235518757e-06,
+      "loss": 0.9545,
+      "step": 14689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.644533194374912e-06,
+      "loss": 0.9708,
+      "step": 14690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.643244096400265e-06,
+      "loss": 1.0013,
+      "step": 14691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.641955061287763e-06,
+      "loss": 1.053,
+      "step": 14692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.640666089061558e-06,
+      "loss": 0.9748,
+      "step": 14693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.639377179745778e-06,
+      "loss": 0.9731,
+      "step": 14694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.638088333364567e-06,
+      "loss": 1.0067,
+      "step": 14695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.636799549942062e-06,
+      "loss": 0.9991,
+      "step": 14696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.635510829502389e-06,
+      "loss": 0.8982,
+      "step": 14697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.634222172069695e-06,
+      "loss": 1.0157,
+      "step": 14698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.632933577668105e-06,
+      "loss": 0.9349,
+      "step": 14699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.631645046321749e-06,
+      "loss": 1.0125,
+      "step": 14700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.630356578054763e-06,
+      "loss": 0.9389,
+      "step": 14701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.629068172891269e-06,
+      "loss": 1.0008,
+      "step": 14702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.627779830855404e-06,
+      "loss": 1.0656,
+      "step": 14703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.626491551971288e-06,
+      "loss": 0.9393,
+      "step": 14704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.625203336263045e-06,
+      "loss": 0.9342,
+      "step": 14705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.623915183754809e-06,
+      "loss": 1.005,
+      "step": 14706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.622627094470694e-06,
+      "loss": 1.0434,
+      "step": 14707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.621339068434828e-06,
+      "loss": 0.9879,
+      "step": 14708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.62005110567133e-06,
+      "loss": 0.9913,
+      "step": 14709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.618763206204318e-06,
+      "loss": 1.0041,
+      "step": 14710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.617475370057914e-06,
+      "loss": 1.0445,
+      "step": 14711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.61618759725623e-06,
+      "loss": 1.0313,
+      "step": 14712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.614899887823388e-06,
+      "loss": 0.9946,
+      "step": 14713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.613612241783501e-06,
+      "loss": 0.934,
+      "step": 14714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.612324659160678e-06,
+      "loss": 0.972,
+      "step": 14715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.611037139979043e-06,
+      "loss": 0.9526,
+      "step": 14716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.609749684262696e-06,
+      "loss": 0.9847,
+      "step": 14717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.608462292035754e-06,
+      "loss": 0.9934,
+      "step": 14718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.607174963322321e-06,
+      "loss": 1.0043,
+      "step": 14719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.605887698146506e-06,
+      "loss": 0.9731,
+      "step": 14720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.604600496532424e-06,
+      "loss": 0.9995,
+      "step": 14721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.603313358504166e-06,
+      "loss": 0.9737,
+      "step": 14722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.60202628408585e-06,
+      "loss": 0.975,
+      "step": 14723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.600739273301573e-06,
+      "loss": 0.9079,
+      "step": 14724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.599452326175432e-06,
+      "loss": 1.0439,
+      "step": 14725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.598165442731539e-06,
+      "loss": 0.9956,
+      "step": 14726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.596878622993983e-06,
+      "loss": 0.9873,
+      "step": 14727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.595591866986872e-06,
+      "loss": 1.028,
+      "step": 14728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.594305174734295e-06,
+      "loss": 1.0518,
+      "step": 14729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.593018546260348e-06,
+      "loss": 0.9848,
+      "step": 14730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.591731981589133e-06,
+      "loss": 0.9595,
+      "step": 14731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.590445480744733e-06,
+      "loss": 0.9427,
+      "step": 14732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.589159043751254e-06,
+      "loss": 1.0345,
+      "step": 14733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.587872670632777e-06,
+      "loss": 0.9811,
+      "step": 14734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.586586361413394e-06,
+      "loss": 1.0252,
+      "step": 14735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.585300116117194e-06,
+      "loss": 0.9743,
+      "step": 14736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.584013934768266e-06,
+      "loss": 0.9526,
+      "step": 14737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.582727817390698e-06,
+      "loss": 0.9612,
+      "step": 14738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.581441764008571e-06,
+      "loss": 0.9817,
+      "step": 14739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.580155774645967e-06,
+      "loss": 0.9847,
+      "step": 14740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.578869849326977e-06,
+      "loss": 0.9981,
+      "step": 14741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.577583988075674e-06,
+      "loss": 0.9768,
+      "step": 14742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.576298190916147e-06,
+      "loss": 0.9634,
+      "step": 14743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.57501245787247e-06,
+      "loss": 1.0129,
+      "step": 14744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.573726788968718e-06,
+      "loss": 1.0184,
+      "step": 14745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.572441184228973e-06,
+      "loss": 0.9371,
+      "step": 14746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.57115564367731e-06,
+      "loss": 1.0059,
+      "step": 14747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.569870167337803e-06,
+      "loss": 1.009,
+      "step": 14748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.568584755234524e-06,
+      "loss": 0.9986,
+      "step": 14749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.567299407391544e-06,
+      "loss": 0.9618,
+      "step": 14750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.566014123832938e-06,
+      "loss": 0.9276,
+      "step": 14751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.564728904582767e-06,
+      "loss": 1.0553,
+      "step": 14752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.563443749665113e-06,
+      "loss": 1.0203,
+      "step": 14753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.562158659104033e-06,
+      "loss": 1.0049,
+      "step": 14754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.560873632923595e-06,
+      "loss": 0.9685,
+      "step": 14755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.559588671147864e-06,
+      "loss": 0.9946,
+      "step": 14756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.558303773800902e-06,
+      "loss": 0.9507,
+      "step": 14757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.557018940906775e-06,
+      "loss": 1.0178,
+      "step": 14758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.555734172489544e-06,
+      "loss": 1.044,
+      "step": 14759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.554449468573261e-06,
+      "loss": 0.9744,
+      "step": 14760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.553164829181997e-06,
+      "loss": 1.006,
+      "step": 14761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.5518802543398e-06,
+      "loss": 1.0188,
+      "step": 14762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.550595744070731e-06,
+      "loss": 0.9784,
+      "step": 14763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.549311298398846e-06,
+      "loss": 1.005,
+      "step": 14764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.548026917348194e-06,
+      "loss": 1.0093,
+      "step": 14765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.546742600942831e-06,
+      "loss": 0.9776,
+      "step": 14766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.545458349206805e-06,
+      "loss": 1.0045,
+      "step": 14767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.544174162164172e-06,
+      "loss": 1.0324,
+      "step": 14768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54289003983898e-06,
+      "loss": 0.9979,
+      "step": 14769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.54160598225527e-06,
+      "loss": 0.9543,
+      "step": 14770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.540321989437096e-06,
+      "loss": 0.9434,
+      "step": 14771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.539038061408504e-06,
+      "loss": 0.9765,
+      "step": 14772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.537754198193532e-06,
+      "loss": 1.0326,
+      "step": 14773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.536470399816226e-06,
+      "loss": 1.0063,
+      "step": 14774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.535186666300627e-06,
+      "loss": 0.9714,
+      "step": 14775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.53390299767078e-06,
+      "loss": 1.0285,
+      "step": 14776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.532619393950718e-06,
+      "loss": 0.9408,
+      "step": 14777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.53133585516448e-06,
+      "loss": 0.9942,
+      "step": 14778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.530052381336108e-06,
+      "loss": 0.9584,
+      "step": 14779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.528768972489631e-06,
+      "loss": 0.941,
+      "step": 14780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.52748562864909e-06,
+      "loss": 0.9593,
+      "step": 14781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.526202349838517e-06,
+      "loss": 1.0284,
+      "step": 14782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.524919136081939e-06,
+      "loss": 0.9431,
+      "step": 14783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.523635987403391e-06,
+      "loss": 0.9771,
+      "step": 14784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.522352903826898e-06,
+      "loss": 1.1174,
+      "step": 14785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.521069885376498e-06,
+      "loss": 1.0008,
+      "step": 14786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.519786932076212e-06,
+      "loss": 0.9705,
+      "step": 14787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.51850404395006e-06,
+      "loss": 0.9164,
+      "step": 14788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.517221221022076e-06,
+      "loss": 0.9784,
+      "step": 14789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.51593846331628e-06,
+      "loss": 1.0486,
+      "step": 14790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.514655770856694e-06,
+      "loss": 0.9875,
+      "step": 14791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 8.513373143667344e-06,
+      "loss": 0.9776,
+      "step": 14792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.51209058177224e-06,
+      "loss": 0.9296,
+      "step": 14793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.51080808519541e-06,
+      "loss": 0.9569,
+      "step": 14794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.509525653960864e-06,
+      "loss": 0.9584,
+      "step": 14795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.508243288092624e-06,
+      "loss": 0.9858,
+      "step": 14796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.506960987614703e-06,
+      "loss": 0.9875,
+      "step": 14797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.505678752551112e-06,
+      "loss": 0.9543,
+      "step": 14798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.504396582925867e-06,
+      "loss": 1.0602,
+      "step": 14799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.503114478762978e-06,
+      "loss": 0.9793,
+      "step": 14800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.501832440086454e-06,
+      "loss": 0.9763,
+      "step": 14801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.500550466920305e-06,
+      "loss": 0.9567,
+      "step": 14802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.49926855928854e-06,
+      "loss": 0.9041,
+      "step": 14803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.497986717215164e-06,
+      "loss": 1.0324,
+      "step": 14804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.496704940724178e-06,
+      "loss": 1.0422,
+      "step": 14805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.495423229839593e-06,
+      "loss": 1.0262,
+      "step": 14806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.494141584585409e-06,
+      "loss": 0.9959,
+      "step": 14807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.492860004985624e-06,
+      "loss": 0.9682,
+      "step": 14808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.491578491064244e-06,
+      "loss": 0.9674,
+      "step": 14809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.490297042845263e-06,
+      "loss": 0.9848,
+      "step": 14810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.489015660352685e-06,
+      "loss": 1.0319,
+      "step": 14811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.487734343610497e-06,
+      "loss": 0.9803,
+      "step": 14812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.486453092642701e-06,
+      "loss": 1.0213,
+      "step": 14813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.485171907473292e-06,
+      "loss": 1.0829,
+      "step": 14814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.483890788126257e-06,
+      "loss": 0.9623,
+      "step": 14815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.482609734625594e-06,
+      "loss": 0.9861,
+      "step": 14816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.481328746995291e-06,
+      "loss": 1.0038,
+      "step": 14817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.480047825259332e-06,
+      "loss": 0.9595,
+      "step": 14818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.478766969441715e-06,
+      "loss": 0.9185,
+      "step": 14819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.477486179566419e-06,
+      "loss": 0.9626,
+      "step": 14820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.47620545565743e-06,
+      "loss": 1.0245,
+      "step": 14821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.47492479773874e-06,
+      "loss": 0.9334,
+      "step": 14822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.473644205834317e-06,
+      "loss": 0.9794,
+      "step": 14823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.472363679968156e-06,
+      "loss": 1.0245,
+      "step": 14824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.47108322016423e-06,
+      "loss": 0.9905,
+      "step": 14825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.469802826446525e-06,
+      "loss": 0.9896,
+      "step": 14826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.468522498839016e-06,
+      "loss": 1.041,
+      "step": 14827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.467242237365673e-06,
+      "loss": 0.9656,
+      "step": 14828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.465962042050482e-06,
+      "loss": 0.9907,
+      "step": 14829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.464681912917411e-06,
+      "loss": 1.0612,
+      "step": 14830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.463401849990438e-06,
+      "loss": 0.9644,
+      "step": 14831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.462121853293531e-06,
+      "loss": 0.9209,
+      "step": 14832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.460841922850655e-06,
+      "loss": 0.9979,
+      "step": 14833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.459562058685792e-06,
+      "loss": 1.0497,
+      "step": 14834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.458282260822899e-06,
+      "loss": 1.0369,
+      "step": 14835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.45700252928595e-06,
+      "loss": 0.9705,
+      "step": 14836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.45572286409891e-06,
+      "loss": 0.9833,
+      "step": 14837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.454443265285738e-06,
+      "loss": 1.0542,
+      "step": 14838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.453163732870402e-06,
+      "loss": 1.0047,
+      "step": 14839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.451884266876862e-06,
+      "loss": 1.0339,
+      "step": 14840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.450604867329079e-06,
+      "loss": 0.93,
+      "step": 14841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.449325534251014e-06,
+      "loss": 0.9367,
+      "step": 14842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44804626766662e-06,
+      "loss": 1.0439,
+      "step": 14843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.44676706759986e-06,
+      "loss": 0.9226,
+      "step": 14844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.445487934074687e-06,
+      "loss": 0.9492,
+      "step": 14845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.444208867115051e-06,
+      "loss": 1.0341,
+      "step": 14846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.442929866744916e-06,
+      "loss": 1.0133,
+      "step": 14847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.441650932988223e-06,
+      "loss": 1.0244,
+      "step": 14848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.440372065868932e-06,
+      "loss": 1.008,
+      "step": 14849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.439093265410985e-06,
+      "loss": 0.9959,
+      "step": 14850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.437814531638331e-06,
+      "loss": 1.0325,
+      "step": 14851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.43653586457492e-06,
+      "loss": 1.0024,
+      "step": 14852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.435257264244694e-06,
+      "loss": 0.9779,
+      "step": 14853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.433978730671604e-06,
+      "loss": 0.9262,
+      "step": 14854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.432700263879589e-06,
+      "loss": 0.9611,
+      "step": 14855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.431421863892587e-06,
+      "loss": 1.0136,
+      "step": 14856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.430143530734546e-06,
+      "loss": 0.934,
+      "step": 14857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.428865264429401e-06,
+      "loss": 1.0081,
+      "step": 14858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.427587065001091e-06,
+      "loss": 0.9865,
+      "step": 14859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.426308932473555e-06,
+      "loss": 0.9886,
+      "step": 14860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.42503086687072e-06,
+      "loss": 0.9932,
+      "step": 14861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.423752868216534e-06,
+      "loss": 0.9312,
+      "step": 14862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.422474936534917e-06,
+      "loss": 0.9626,
+      "step": 14863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.421197071849814e-06,
+      "loss": 0.9901,
+      "step": 14864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.419919274185147e-06,
+      "loss": 1.1199,
+      "step": 14865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.418641543564845e-06,
+      "loss": 0.9521,
+      "step": 14866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.417363880012838e-06,
+      "loss": 0.9774,
+      "step": 14867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.416086283553053e-06,
+      "loss": 0.9432,
+      "step": 14868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.414808754209418e-06,
+      "loss": 1.0244,
+      "step": 14869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.413531292005857e-06,
+      "loss": 0.9661,
+      "step": 14870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.412253896966285e-06,
+      "loss": 0.9674,
+      "step": 14871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.410976569114635e-06,
+      "loss": 1.0168,
+      "step": 14872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.40969930847482e-06,
+      "loss": 0.9884,
+      "step": 14873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.408422115070763e-06,
+      "loss": 1.025,
+      "step": 14874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.407144988926383e-06,
+      "loss": 0.9862,
+      "step": 14875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.405867930065591e-06,
+      "loss": 1.0077,
+      "step": 14876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.404590938512311e-06,
+      "loss": 1.0043,
+      "step": 14877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.403314014290448e-06,
+      "loss": 0.9789,
+      "step": 14878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.402037157423924e-06,
+      "loss": 1.0285,
+      "step": 14879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.400760367936647e-06,
+      "loss": 0.9814,
+      "step": 14880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.39948364585252e-06,
+      "loss": 0.9159,
+      "step": 14881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.398206991195466e-06,
+      "loss": 1.044,
+      "step": 14882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.396930403989382e-06,
+      "loss": 1.0547,
+      "step": 14883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.395653884258183e-06,
+      "loss": 1.0901,
+      "step": 14884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.39437743202577e-06,
+      "loss": 1.0593,
+      "step": 14885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.393101047316045e-06,
+      "loss": 1.0338,
+      "step": 14886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.391824730152916e-06,
+      "loss": 0.9641,
+      "step": 14887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.390548480560278e-06,
+      "loss": 0.9984,
+      "step": 14888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.38927229856204e-06,
+      "loss": 0.9083,
+      "step": 14889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.387996184182097e-06,
+      "loss": 0.9963,
+      "step": 14890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.386720137444344e-06,
+      "loss": 0.925,
+      "step": 14891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.385444158372682e-06,
+      "loss": 0.9462,
+      "step": 14892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.384168246991002e-06,
+      "loss": 0.9683,
+      "step": 14893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.382892403323202e-06,
+      "loss": 1.0096,
+      "step": 14894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.381616627393177e-06,
+      "loss": 0.9451,
+      "step": 14895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.38034091922481e-06,
+      "loss": 1.0024,
+      "step": 14896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.379065278842e-06,
+      "loss": 0.963,
+      "step": 14897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.377789706268627e-06,
+      "loss": 0.9816,
+      "step": 14898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.376514201528588e-06,
+      "loss": 1.0101,
+      "step": 14899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.375238764645766e-06,
+      "loss": 0.9984,
+      "step": 14900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.37396339564404e-06,
+      "loss": 0.9704,
+      "step": 14901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.372688094547306e-06,
+      "loss": 0.985,
+      "step": 14902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.371412861379436e-06,
+      "loss": 1.0388,
+      "step": 14903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.370137696164319e-06,
+      "loss": 1.0375,
+      "step": 14904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.368862598925826e-06,
+      "loss": 1.0834,
+      "step": 14905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.367587569687844e-06,
+      "loss": 0.9698,
+      "step": 14906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.36631260847425e-06,
+      "loss": 0.9937,
+      "step": 14907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.365037715308917e-06,
+      "loss": 1.0472,
+      "step": 14908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.363762890215719e-06,
+      "loss": 1.0341,
+      "step": 14909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.362488133218534e-06,
+      "loss": 1.0382,
+      "step": 14910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.361213444341229e-06,
+      "loss": 0.9794,
+      "step": 14911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.359938823607682e-06,
+      "loss": 0.9358,
+      "step": 14912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.35866427104176e-06,
+      "loss": 1.0299,
+      "step": 14913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.357389786667327e-06,
+      "loss": 0.9594,
+      "step": 14914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.35611537050826e-06,
+      "loss": 0.9797,
+      "step": 14915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.354841022588411e-06,
+      "loss": 0.9652,
+      "step": 14916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.35356674293166e-06,
+      "loss": 0.9847,
+      "step": 14917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.352292531561863e-06,
+      "loss": 1.0017,
+      "step": 14918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.351018388502879e-06,
+      "loss": 0.9185,
+      "step": 14919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.349744313778576e-06,
+      "loss": 0.9894,
+      "step": 14920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.348470307412808e-06,
+      "loss": 0.9274,
+      "step": 14921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.347196369429438e-06,
+      "loss": 0.9009,
+      "step": 14922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.345922499852322e-06,
+      "loss": 0.9781,
+      "step": 14923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.344648698705314e-06,
+      "loss": 1.0618,
+      "step": 14924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.34337496601227e-06,
+      "loss": 0.987,
+      "step": 14925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.342101301797038e-06,
+      "loss": 0.9907,
+      "step": 14926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.340827706083479e-06,
+      "loss": 1.038,
+      "step": 14927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.33955417889544e-06,
+      "loss": 1.0121,
+      "step": 14928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.338280720256763e-06,
+      "loss": 0.9927,
+      "step": 14929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.337007330191311e-06,
+      "loss": 0.9792,
+      "step": 14930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.335734008722918e-06,
+      "loss": 0.9733,
+      "step": 14931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.334460755875436e-06,
+      "loss": 1.0625,
+      "step": 14932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.333187571672707e-06,
+      "loss": 1.0334,
+      "step": 14933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.331914456138574e-06,
+      "loss": 1.0055,
+      "step": 14934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.330641409296881e-06,
+      "loss": 0.9587,
+      "step": 14935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.329368431171464e-06,
+      "loss": 1.0286,
+      "step": 14936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.328095521786165e-06,
+      "loss": 0.9975,
+      "step": 14937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.326822681164824e-06,
+      "loss": 1.0095,
+      "step": 14938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.325549909331273e-06,
+      "loss": 0.932,
+      "step": 14939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.32427720630935e-06,
+      "loss": 1.0335,
+      "step": 14940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.323004572122889e-06,
+      "loss": 0.9976,
+      "step": 14941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.321732006795723e-06,
+      "loss": 1.025,
+      "step": 14942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.320459510351681e-06,
+      "loss": 0.9068,
+      "step": 14943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.319187082814592e-06,
+      "loss": 0.9329,
+      "step": 14944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.317914724208291e-06,
+      "loss": 0.9169,
+      "step": 14945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.316642434556599e-06,
+      "loss": 1.0284,
+      "step": 14946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.315370213883349e-06,
+      "loss": 1.0624,
+      "step": 14947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.314098062212361e-06,
+      "loss": 1.0522,
+      "step": 14948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.312825979567457e-06,
+      "loss": 0.9737,
+      "step": 14949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.311553965972465e-06,
+      "loss": 1.0461,
+      "step": 14950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.310282021451202e-06,
+      "loss": 0.9954,
+      "step": 14951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.30901014602749e-06,
+      "loss": 0.9747,
+      "step": 14952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.307738339725146e-06,
+      "loss": 0.9424,
+      "step": 14953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.306466602567986e-06,
+      "loss": 1.0185,
+      "step": 14954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.30519493457983e-06,
+      "loss": 0.9592,
+      "step": 14955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.303923335784485e-06,
+      "loss": 1.0068,
+      "step": 14956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.302651806205775e-06,
+      "loss": 1.007,
+      "step": 14957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.301380345867507e-06,
+      "loss": 1.035,
+      "step": 14958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.300108954793489e-06,
+      "loss": 0.965,
+      "step": 14959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.29883763300753e-06,
+      "loss": 1.012,
+      "step": 14960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.297566380533445e-06,
+      "loss": 0.9734,
+      "step": 14961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.296295197395035e-06,
+      "loss": 1.0224,
+      "step": 14962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.295024083616108e-06,
+      "loss": 0.9675,
+      "step": 14963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.293753039220463e-06,
+      "loss": 0.9635,
+      "step": 14964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.292482064231912e-06,
+      "loss": 0.9303,
+      "step": 14965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.291211158674246e-06,
+      "loss": 1.0071,
+      "step": 14966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.289940322571278e-06,
+      "loss": 0.9668,
+      "step": 14967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.288669555946796e-06,
+      "loss": 0.9516,
+      "step": 14968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.287398858824602e-06,
+      "loss": 0.936,
+      "step": 14969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.286128231228495e-06,
+      "loss": 0.963,
+      "step": 14970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.284857673182262e-06,
+      "loss": 0.9455,
+      "step": 14971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.283587184709708e-06,
+      "loss": 0.9883,
+      "step": 14972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.282316765834618e-06,
+      "loss": 0.9532,
+      "step": 14973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.28104641658078e-06,
+      "loss": 0.9982,
+      "step": 14974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.279776136971996e-06,
+      "loss": 0.9149,
+      "step": 14975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.278505927032045e-06,
+      "loss": 1.0357,
+      "step": 14976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.277235786784712e-06,
+      "loss": 0.9549,
+      "step": 14977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.275965716253794e-06,
+      "loss": 0.959,
+      "step": 14978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.274695715463064e-06,
+      "loss": 1.0707,
+      "step": 14979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.273425784436316e-06,
+      "loss": 0.9796,
+      "step": 14980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.272155923197325e-06,
+      "loss": 0.9768,
+      "step": 14981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.270886131769869e-06,
+      "loss": 0.9034,
+      "step": 14982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.269616410177736e-06,
+      "loss": 0.9737,
+      "step": 14983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.268346758444696e-06,
+      "loss": 0.9607,
+      "step": 14984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.267077176594536e-06,
+      "loss": 1.0084,
+      "step": 14985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.265807664651025e-06,
+      "loss": 0.9754,
+      "step": 14986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.264538222637932e-06,
+      "loss": 1.0161,
+      "step": 14987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.26326885057904e-06,
+      "loss": 0.9232,
+      "step": 14988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.261999548498115e-06,
+      "loss": 0.9713,
+      "step": 14989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.26073031641893e-06,
+      "loss": 0.9197,
+      "step": 14990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.259461154365252e-06,
+      "loss": 1.0004,
+      "step": 14991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.258192062360846e-06,
+      "loss": 0.9673,
+      "step": 14992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.256923040429486e-06,
+      "loss": 0.949,
+      "step": 14993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.255654088594927e-06,
+      "loss": 1.0252,
+      "step": 14994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.254385206880942e-06,
+      "loss": 1.0263,
+      "step": 14995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.253116395311292e-06,
+      "loss": 1.0154,
+      "step": 14996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.251847653909731e-06,
+      "loss": 0.9945,
+      "step": 14997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.250578982700026e-06,
+      "loss": 1.0212,
+      "step": 14998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.249310381705931e-06,
+      "loss": 0.9766,
+      "step": 14999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.248041850951208e-06,
+      "loss": 0.9848,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "eval_text_loss": 0.94095778465271,
+      "eval_text_runtime": 1.5648,
+      "eval_text_samples_per_second": 47.291,
+      "eval_text_steps_per_second": 1.278,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "eval_svit_loss": 1.0947773456573486,
+      "eval_svit_runtime": 3.7359,
+      "eval_svit_samples_per_second": 182.287,
+      "eval_svit_steps_per_second": 2.944,
+      "step": 15000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.24677339045961e-06,
+      "loss": 0.994,
+      "step": 15001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.245505000254887e-06,
+      "loss": 0.9965,
+      "step": 15002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.244236680360804e-06,
+      "loss": 0.8694,
+      "step": 15003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.242968430801099e-06,
+      "loss": 1.0435,
+      "step": 15004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.241700251599532e-06,
+      "loss": 1.0131,
+      "step": 15005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.240432142779852e-06,
+      "loss": 0.9698,
+      "step": 15006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.239164104365801e-06,
+      "loss": 0.9779,
+      "step": 15007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.237896136381133e-06,
+      "loss": 0.9979,
+      "step": 15008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.236628238849583e-06,
+      "loss": 1.0297,
+      "step": 15009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.235360411794906e-06,
+      "loss": 0.9359,
+      "step": 15010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.23409265524084e-06,
+      "loss": 0.9716,
+      "step": 15011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.232824969211123e-06,
+      "loss": 0.9274,
+      "step": 15012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.231557353729502e-06,
+      "loss": 1.0607,
+      "step": 15013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.23028980881971e-06,
+      "loss": 0.9835,
+      "step": 15014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.229022334505488e-06,
+      "loss": 1.0382,
+      "step": 15015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.22775493081057e-06,
+      "loss": 0.9564,
+      "step": 15016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.22648759775869e-06,
+      "loss": 0.9349,
+      "step": 15017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.225220335373587e-06,
+      "loss": 0.9708,
+      "step": 15018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.223953143678983e-06,
+      "loss": 0.9728,
+      "step": 15019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.222686022698618e-06,
+      "loss": 1.0338,
+      "step": 15020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.221418972456219e-06,
+      "loss": 0.9726,
+      "step": 15021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.22015199297551e-06,
+      "loss": 0.9987,
+      "step": 15022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.218885084280225e-06,
+      "loss": 1.0225,
+      "step": 15023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.217618246394081e-06,
+      "loss": 0.9905,
+      "step": 15024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.216351479340808e-06,
+      "loss": 1.0674,
+      "step": 15025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.21508478314413e-06,
+      "loss": 0.9524,
+      "step": 15026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.213818157827763e-06,
+      "loss": 0.9475,
+      "step": 15027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.21255160341543e-06,
+      "loss": 0.9758,
+      "step": 15028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 8.211285119930849e-06,
+      "loss": 0.9646,
+      "step": 15029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.21001870739774e-06,
+      "loss": 1.0039,
+      "step": 15030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.208752365839819e-06,
+      "loss": 0.946,
+      "step": 15031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.207486095280795e-06,
+      "loss": 1.0255,
+      "step": 15032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.20621989574439e-06,
+      "loss": 1.0017,
+      "step": 15033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.20495376725431e-06,
+      "loss": 1.0789,
+      "step": 15034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.203687709834273e-06,
+      "loss": 0.9521,
+      "step": 15035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.202421723507978e-06,
+      "loss": 1.0435,
+      "step": 15036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.201155808299137e-06,
+      "loss": 0.9878,
+      "step": 15037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.199889964231464e-06,
+      "loss": 0.9248,
+      "step": 15038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.198624191328653e-06,
+      "loss": 1.0434,
+      "step": 15039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.19735848961442e-06,
+      "loss": 0.9996,
+      "step": 15040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.196092859112461e-06,
+      "loss": 1.0489,
+      "step": 15041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.194827299846474e-06,
+      "loss": 0.9929,
+      "step": 15042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.193561811840169e-06,
+      "loss": 0.9616,
+      "step": 15043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.192296395117238e-06,
+      "loss": 0.9774,
+      "step": 15044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.191031049701381e-06,
+      "loss": 0.997,
+      "step": 15045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.189765775616294e-06,
+      "loss": 0.9272,
+      "step": 15046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.188500572885667e-06,
+      "loss": 1.0552,
+      "step": 15047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.1872354415332e-06,
+      "loss": 0.9898,
+      "step": 15048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.185970381582587e-06,
+      "loss": 0.9537,
+      "step": 15049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.184705393057508e-06,
+      "loss": 1.029,
+      "step": 15050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.183440475981665e-06,
+      "loss": 1.0041,
+      "step": 15051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.182175630378739e-06,
+      "loss": 0.9696,
+      "step": 15052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.180910856272415e-06,
+      "loss": 0.9396,
+      "step": 15053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.179646153686388e-06,
+      "loss": 0.972,
+      "step": 15054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.17838152264433e-06,
+      "loss": 1.0022,
+      "step": 15055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.177116963169935e-06,
+      "loss": 1.0058,
+      "step": 15056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.175852475286874e-06,
+      "loss": 0.9611,
+      "step": 15057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.174588059018837e-06,
+      "loss": 1.01,
+      "step": 15058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.173323714389497e-06,
+      "loss": 0.9614,
+      "step": 15059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.17205944142253e-06,
+      "loss": 1.0369,
+      "step": 15060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.170795240141617e-06,
+      "loss": 0.9526,
+      "step": 15061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.16953111057043e-06,
+      "loss": 0.9744,
+      "step": 15062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.168267052732647e-06,
+      "loss": 1.0117,
+      "step": 15063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.16700306665193e-06,
+      "loss": 1.0396,
+      "step": 15064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.165739152351958e-06,
+      "loss": 0.9849,
+      "step": 15065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.1644753098564e-06,
+      "loss": 1.0136,
+      "step": 15066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.163211539188918e-06,
+      "loss": 0.9772,
+      "step": 15067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.161947840373185e-06,
+      "loss": 1.008,
+      "step": 15068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.160684213432868e-06,
+      "loss": 1.0381,
+      "step": 15069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.15942065839162e-06,
+      "loss": 0.9468,
+      "step": 15070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.158157175273116e-06,
+      "loss": 1.0348,
+      "step": 15071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.15689376410101e-06,
+      "loss": 0.9802,
+      "step": 15072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.155630424898968e-06,
+      "loss": 0.9765,
+      "step": 15073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.154367157690644e-06,
+      "loss": 1.0262,
+      "step": 15074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.15310396249969e-06,
+      "loss": 1.0046,
+      "step": 15075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.151840839349776e-06,
+      "loss": 0.9631,
+      "step": 15076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.150577788264543e-06,
+      "loss": 0.9996,
+      "step": 15077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.149314809267653e-06,
+      "loss": 0.9851,
+      "step": 15078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.148051902382754e-06,
+      "loss": 1.0208,
+      "step": 15079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.146789067633493e-06,
+      "loss": 0.9663,
+      "step": 15080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.14552630504353e-06,
+      "loss": 0.9727,
+      "step": 15081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.144263614636503e-06,
+      "loss": 0.9876,
+      "step": 15082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.143000996436064e-06,
+      "loss": 1.0078,
+      "step": 15083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.141738450465856e-06,
+      "loss": 0.992,
+      "step": 15084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.140475976749519e-06,
+      "loss": 1.0189,
+      "step": 15085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.139213575310702e-06,
+      "loss": 0.9565,
+      "step": 15086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.13795124617304e-06,
+      "loss": 0.9291,
+      "step": 15087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.136688989360178e-06,
+      "loss": 1.015,
+      "step": 15088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.135426804895755e-06,
+      "loss": 1.0509,
+      "step": 15089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.1341646928034e-06,
+      "loss": 0.9612,
+      "step": 15090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.132902653106757e-06,
+      "loss": 0.9716,
+      "step": 15091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.131640685829455e-06,
+      "loss": 0.9447,
+      "step": 15092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.13037879099513e-06,
+      "loss": 1.0026,
+      "step": 15093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.129116968627415e-06,
+      "loss": 0.9871,
+      "step": 15094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.127855218749932e-06,
+      "loss": 0.953,
+      "step": 15095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.126593541386321e-06,
+      "loss": 0.9878,
+      "step": 15096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.1253319365602e-06,
+      "loss": 0.9861,
+      "step": 15097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.124070404295204e-06,
+      "loss": 0.9673,
+      "step": 15098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.12280894461495e-06,
+      "loss": 0.9316,
+      "step": 15099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.121547557543066e-06,
+      "loss": 0.9727,
+      "step": 15100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.120286243103173e-06,
+      "loss": 1.0667,
+      "step": 15101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.119025001318888e-06,
+      "loss": 0.9725,
+      "step": 15102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.117763832213836e-06,
+      "loss": 0.9521,
+      "step": 15103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.116502735811634e-06,
+      "loss": 1.0112,
+      "step": 15104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.115241712135893e-06,
+      "loss": 0.9673,
+      "step": 15105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.113980761210237e-06,
+      "loss": 0.9937,
+      "step": 15106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.11271988305827e-06,
+      "loss": 0.989,
+      "step": 15107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.111459077703614e-06,
+      "loss": 0.9878,
+      "step": 15108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.110198345169877e-06,
+      "loss": 0.9798,
+      "step": 15109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.108937685480666e-06,
+      "loss": 0.9728,
+      "step": 15110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.107677098659595e-06,
+      "loss": 0.9349,
+      "step": 15111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.106416584730264e-06,
+      "loss": 0.9391,
+      "step": 15112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.10515614371628e-06,
+      "loss": 0.9912,
+      "step": 15113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.103895775641253e-06,
+      "loss": 0.9474,
+      "step": 15114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.102635480528777e-06,
+      "loss": 0.9057,
+      "step": 15115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.101375258402466e-06,
+      "loss": 1.0401,
+      "step": 15116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.100115109285912e-06,
+      "loss": 0.9595,
+      "step": 15117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.09885503320271e-06,
+      "loss": 0.9334,
+      "step": 15118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.097595030176468e-06,
+      "loss": 0.9239,
+      "step": 15119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.096335100230776e-06,
+      "loss": 0.9808,
+      "step": 15120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.09507524338923e-06,
+      "loss": 1.0054,
+      "step": 15121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.093815459675424e-06,
+      "loss": 1.0044,
+      "step": 15122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.092555749112945e-06,
+      "loss": 0.94,
+      "step": 15123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.091296111725391e-06,
+      "loss": 1.002,
+      "step": 15124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.090036547536345e-06,
+      "loss": 1.0197,
+      "step": 15125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.0887770565694e-06,
+      "loss": 1.0117,
+      "step": 15126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.087517638848142e-06,
+      "loss": 0.9419,
+      "step": 15127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.086258294396153e-06,
+      "loss": 1.0193,
+      "step": 15128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.084999023237021e-06,
+      "loss": 0.9942,
+      "step": 15129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.08373982539432e-06,
+      "loss": 0.9504,
+      "step": 15130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.082480700891642e-06,
+      "loss": 0.9805,
+      "step": 15131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.081221649752562e-06,
+      "loss": 1.011,
+      "step": 15132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.079962672000655e-06,
+      "loss": 1.0044,
+      "step": 15133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.078703767659503e-06,
+      "loss": 1.0743,
+      "step": 15134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.077444936752676e-06,
+      "loss": 1.0198,
+      "step": 15135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.076186179303759e-06,
+      "loss": 0.9962,
+      "step": 15136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.074927495336314e-06,
+      "loss": 0.9254,
+      "step": 15137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.073668884873913e-06,
+      "loss": 0.9886,
+      "step": 15138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.072410347940133e-06,
+      "loss": 0.9898,
+      "step": 15139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.071151884558535e-06,
+      "loss": 1.0415,
+      "step": 15140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.069893494752695e-06,
+      "loss": 0.989,
+      "step": 15141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.068635178546173e-06,
+      "loss": 1.0127,
+      "step": 15142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.06737693596253e-06,
+      "loss": 0.8773,
+      "step": 15143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.066118767025339e-06,
+      "loss": 0.9668,
+      "step": 15144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.064860671758153e-06,
+      "loss": 0.9482,
+      "step": 15145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.063602650184537e-06,
+      "loss": 1.0558,
+      "step": 15146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.06234470232805e-06,
+      "loss": 0.9407,
+      "step": 15147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.061086828212248e-06,
+      "loss": 0.9595,
+      "step": 15148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.05982902786069e-06,
+      "loss": 0.9795,
+      "step": 15149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.058571301296924e-06,
+      "loss": 0.9899,
+      "step": 15150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.057313648544514e-06,
+      "loss": 0.9742,
+      "step": 15151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.056056069627005e-06,
+      "loss": 0.9892,
+      "step": 15152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.054798564567945e-06,
+      "loss": 1.0066,
+      "step": 15153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.053541133390891e-06,
+      "loss": 1.0294,
+      "step": 15154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.052283776119387e-06,
+      "loss": 0.983,
+      "step": 15155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.051026492776983e-06,
+      "loss": 0.9898,
+      "step": 15156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.049769283387218e-06,
+      "loss": 0.9242,
+      "step": 15157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04851214797364e-06,
+      "loss": 0.9587,
+      "step": 15158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.047255086559791e-06,
+      "loss": 0.9921,
+      "step": 15159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04599809916921e-06,
+      "loss": 1.0035,
+      "step": 15160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04474118582544e-06,
+      "loss": 1.0187,
+      "step": 15161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04348434655202e-06,
+      "loss": 0.9931,
+      "step": 15162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.04222758137248e-06,
+      "loss": 0.9802,
+      "step": 15163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.040970890310362e-06,
+      "loss": 1.0077,
+      "step": 15164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.0397142733892e-06,
+      "loss": 0.9609,
+      "step": 15165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.038457730632524e-06,
+      "loss": 1.0242,
+      "step": 15166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.037201262063867e-06,
+      "loss": 1.0029,
+      "step": 15167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.035944867706756e-06,
+      "loss": 0.9771,
+      "step": 15168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.034688547584723e-06,
+      "loss": 0.938,
+      "step": 15169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.033432301721292e-06,
+      "loss": 0.9797,
+      "step": 15170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.032176130139994e-06,
+      "loss": 1.0015,
+      "step": 15171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.03092003286435e-06,
+      "loss": 0.9856,
+      "step": 15172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.029664009917879e-06,
+      "loss": 1.0008,
+      "step": 15173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02840806132411e-06,
+      "loss": 0.9746,
+      "step": 15174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.027152187106561e-06,
+      "loss": 1.0456,
+      "step": 15175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.02589638728875e-06,
+      "loss": 1.0356,
+      "step": 15176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.024640661894195e-06,
+      "loss": 1.0072,
+      "step": 15177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.023385010946406e-06,
+      "loss": 0.969,
+      "step": 15178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.022129434468908e-06,
+      "loss": 0.9934,
+      "step": 15179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.020873932485208e-06,
+      "loss": 0.98,
+      "step": 15180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.019618505018817e-06,
+      "loss": 0.9815,
+      "step": 15181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.01836315209325e-06,
+      "loss": 0.9765,
+      "step": 15182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.017107873732011e-06,
+      "loss": 0.9528,
+      "step": 15183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.015852669958612e-06,
+      "loss": 0.956,
+      "step": 15184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.014597540796559e-06,
+      "loss": 1.0202,
+      "step": 15185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.013342486269352e-06,
+      "loss": 0.9732,
+      "step": 15186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.012087506400501e-06,
+      "loss": 0.9572,
+      "step": 15187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.0108326012135e-06,
+      "loss": 1.0184,
+      "step": 15188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.009577770731861e-06,
+      "loss": 0.9574,
+      "step": 15189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.008323014979075e-06,
+      "loss": 0.9657,
+      "step": 15190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.00706833397864e-06,
+      "loss": 1.0307,
+      "step": 15191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.005813727754055e-06,
+      "loss": 0.9738,
+      "step": 15192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.004559196328812e-06,
+      "loss": 0.972,
+      "step": 15193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.003304739726412e-06,
+      "loss": 1.0021,
+      "step": 15194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.00205035797034e-06,
+      "loss": 0.9668,
+      "step": 15195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 8.000796051084087e-06,
+      "loss": 1.059,
+      "step": 15196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.999541819091149e-06,
+      "loss": 0.9488,
+      "step": 15197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.998287662015002e-06,
+      "loss": 0.9848,
+      "step": 15198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.997033579879147e-06,
+      "loss": 0.9548,
+      "step": 15199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.995779572707063e-06,
+      "loss": 0.9994,
+      "step": 15200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.99452564052223e-06,
+      "loss": 0.8885,
+      "step": 15201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.993271783348136e-06,
+      "loss": 0.9416,
+      "step": 15202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.99201800120826e-06,
+      "loss": 1.0196,
+      "step": 15203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.990764294126082e-06,
+      "loss": 1.0258,
+      "step": 15204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.98951066212508e-06,
+      "loss": 0.9596,
+      "step": 15205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.988257105228728e-06,
+      "loss": 1.0332,
+      "step": 15206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.987003623460508e-06,
+      "loss": 0.974,
+      "step": 15207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.985750216843884e-06,
+      "loss": 0.9599,
+      "step": 15208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.98449688540234e-06,
+      "loss": 0.96,
+      "step": 15209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.983243629159344e-06,
+      "loss": 1.0415,
+      "step": 15210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.981990448138358e-06,
+      "loss": 0.9754,
+      "step": 15211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.98073734236286e-06,
+      "loss": 1.0103,
+      "step": 15212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.979484311856311e-06,
+      "loss": 1.0691,
+      "step": 15213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.978231356642181e-06,
+      "loss": 0.9976,
+      "step": 15214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.976978476743932e-06,
+      "loss": 0.999,
+      "step": 15215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.975725672185024e-06,
+      "loss": 0.9982,
+      "step": 15216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.974472942988922e-06,
+      "loss": 0.9712,
+      "step": 15217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.973220289179084e-06,
+      "loss": 1.0518,
+      "step": 15218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.971967710778972e-06,
+      "loss": 0.9743,
+      "step": 15219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.97071520781204e-06,
+      "loss": 1.0213,
+      "step": 15220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.969462780301742e-06,
+      "loss": 0.9997,
+      "step": 15221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.968210428271539e-06,
+      "loss": 0.9499,
+      "step": 15222
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.966958151744872e-06,
+      "loss": 1.0145,
+      "step": 15223
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.965705950745206e-06,
+      "loss": 0.9715,
+      "step": 15224
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.964453825295984e-06,
+      "loss": 0.9702,
+      "step": 15225
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.963201775420653e-06,
+      "loss": 1.0733,
+      "step": 15226
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.961949801142665e-06,
+      "loss": 1.0123,
+      "step": 15227
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.96069790248546e-06,
+      "loss": 1.0106,
+      "step": 15228
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.95944607947249e-06,
+      "loss": 0.9239,
+      "step": 15229
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.958194332127194e-06,
+      "loss": 0.9559,
+      "step": 15230
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.95694266047301e-06,
+      "loss": 0.9969,
+      "step": 15231
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.955691064533385e-06,
+      "loss": 1.0025,
+      "step": 15232
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.954439544331749e-06,
+      "loss": 0.9988,
+      "step": 15233
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.95318809989155e-06,
+      "loss": 1.011,
+      "step": 15234
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.951936731236218e-06,
+      "loss": 0.9959,
+      "step": 15235
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.950685438389184e-06,
+      "loss": 1.0219,
+      "step": 15236
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.949434221373887e-06,
+      "loss": 0.9651,
+      "step": 15237
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.948183080213756e-06,
+      "loss": 0.9988,
+      "step": 15238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.946932014932221e-06,
+      "loss": 0.9841,
+      "step": 15239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.945681025552713e-06,
+      "loss": 1.0563,
+      "step": 15240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.944430112098656e-06,
+      "loss": 0.9774,
+      "step": 15241
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.94317927459348e-06,
+      "loss": 1.0219,
+      "step": 15242
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.941928513060603e-06,
+      "loss": 1.0122,
+      "step": 15243
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.940677827523455e-06,
+      "loss": 0.9777,
+      "step": 15244
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.939427218005458e-06,
+      "loss": 0.9944,
+      "step": 15245
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.938176684530023e-06,
+      "loss": 0.9762,
+      "step": 15246
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.936926227120578e-06,
+      "loss": 1.0479,
+      "step": 15247
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.935675845800537e-06,
+      "loss": 1.0173,
+      "step": 15248
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.934425540593318e-06,
+      "loss": 0.9814,
+      "step": 15249
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.933175311522332e-06,
+      "loss": 0.957,
+      "step": 15250
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.931925158610993e-06,
+      "loss": 0.9831,
+      "step": 15251
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.930675081882716e-06,
+      "loss": 0.9622,
+      "step": 15252
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.929425081360909e-06,
+      "loss": 0.9958,
+      "step": 15253
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.928175157068977e-06,
+      "loss": 0.9726,
+      "step": 15254
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.926925309030334e-06,
+      "loss": 0.9202,
+      "step": 15255
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.92567553726838e-06,
+      "loss": 0.9557,
+      "step": 15256
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.924425841806525e-06,
+      "loss": 1.0408,
+      "step": 15257
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.923176222668172e-06,
+      "loss": 0.9869,
+      "step": 15258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.921926679876717e-06,
+      "loss": 0.9997,
+      "step": 15259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.920677213455566e-06,
+      "loss": 0.9817,
+      "step": 15260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.919427823428112e-06,
+      "loss": 0.9597,
+      "step": 15261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.918178509817758e-06,
+      "loss": 0.9996,
+      "step": 15262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.9169292726479e-06,
+      "loss": 1.0161,
+      "step": 15263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.915680111941926e-06,
+      "loss": 0.9749,
+      "step": 15264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.914431027723237e-06,
+      "loss": 1.0162,
+      "step": 15265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.913182020015217e-06,
+      "loss": 0.9616,
+      "step": 15266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.911933088841264e-06,
+      "loss": 1.0456,
+      "step": 15267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.910684234224764e-06,
+      "loss": 1.0481,
+      "step": 15268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.909435456189101e-06,
+      "loss": 0.9833,
+      "step": 15269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.908186754757666e-06,
+      "loss": 1.0467,
+      "step": 15270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.906938129953837e-06,
+      "loss": 1.0629,
+      "step": 15271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.905689581801005e-06,
+      "loss": 0.9343,
+      "step": 15272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.904441110322547e-06,
+      "loss": 0.9594,
+      "step": 15273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.903192715541842e-06,
+      "loss": 1.052,
+      "step": 15274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.901944397482272e-06,
+      "loss": 0.8622,
+      "step": 15275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.900696156167212e-06,
+      "loss": 0.961,
+      "step": 15276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.89944799162004e-06,
+      "loss": 0.9211,
+      "step": 15277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.898199903864127e-06,
+      "loss": 0.9762,
+      "step": 15278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.896951892922851e-06,
+      "loss": 0.968,
+      "step": 15279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.895703958819581e-06,
+      "loss": 1.0357,
+      "step": 15280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.894456101577684e-06,
+      "loss": 0.9621,
+      "step": 15281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.893208321220534e-06,
+      "loss": 0.9629,
+      "step": 15282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.891960617771497e-06,
+      "loss": 0.9902,
+      "step": 15283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.890712991253934e-06,
+      "loss": 1.0312,
+      "step": 15284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.889465441691216e-06,
+      "loss": 0.9359,
+      "step": 15285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.8882179691067e-06,
+      "loss": 0.957,
+      "step": 15286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.886970573523756e-06,
+      "loss": 1.0065,
+      "step": 15287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.885723254965734e-06,
+      "loss": 1.0014,
+      "step": 15288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.884476013455997e-06,
+      "loss": 0.971,
+      "step": 15289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.883228849017906e-06,
+      "loss": 0.9694,
+      "step": 15290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.881981761674808e-06,
+      "loss": 1.0414,
+      "step": 15291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.880734751450065e-06,
+      "loss": 0.9696,
+      "step": 15292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.87948781836703e-06,
+      "loss": 0.9575,
+      "step": 15293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.878240962449046e-06,
+      "loss": 1.0269,
+      "step": 15294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.876994183719474e-06,
+      "loss": 1.01,
+      "step": 15295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.875747482201652e-06,
+      "loss": 0.9626,
+      "step": 15296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.874500857918937e-06,
+      "loss": 1.0073,
+      "step": 15297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.873254310894667e-06,
+      "loss": 1.0312,
+      "step": 15298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.872007841152187e-06,
+      "loss": 0.9632,
+      "step": 15299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.870761448714844e-06,
+      "loss": 1.0544,
+      "step": 15300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.869515133605972e-06,
+      "loss": 1.0119,
+      "step": 15301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.868268895848921e-06,
+      "loss": 1.0057,
+      "step": 15302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.867022735467024e-06,
+      "loss": 0.9699,
+      "step": 15303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.865776652483614e-06,
+      "loss": 1.0385,
+      "step": 15304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.864530646922033e-06,
+      "loss": 0.9584,
+      "step": 15305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.863284718805611e-06,
+      "loss": 1.0437,
+      "step": 15306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.862038868157684e-06,
+      "loss": 0.9655,
+      "step": 15307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86079309500158e-06,
+      "loss": 0.9406,
+      "step": 15308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.859547399360627e-06,
+      "loss": 0.9807,
+      "step": 15309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.85830178125816e-06,
+      "loss": 0.9721,
+      "step": 15310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.857056240717496e-06,
+      "loss": 0.9963,
+      "step": 15311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.85581077776197e-06,
+      "loss": 0.966,
+      "step": 15312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.854565392414903e-06,
+      "loss": 0.9046,
+      "step": 15313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.853320084699615e-06,
+      "loss": 1.0154,
+      "step": 15314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.85207485463943e-06,
+      "loss": 1.0141,
+      "step": 15315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.850829702257658e-06,
+      "loss": 0.9544,
+      "step": 15316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.849584627577632e-06,
+      "loss": 0.9428,
+      "step": 15317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.848339630622659e-06,
+      "loss": 0.9992,
+      "step": 15318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.847094711416054e-06,
+      "loss": 0.9827,
+      "step": 15319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.845849869981138e-06,
+      "loss": 1.0023,
+      "step": 15320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.844605106341215e-06,
+      "loss": 1.0397,
+      "step": 15321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.843360420519597e-06,
+      "loss": 0.988,
+      "step": 15322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.842115812539599e-06,
+      "loss": 0.9855,
+      "step": 15323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.840871282424524e-06,
+      "loss": 1.0295,
+      "step": 15324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83962683019768e-06,
+      "loss": 0.9392,
+      "step": 15325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.838382455882372e-06,
+      "loss": 1.0277,
+      "step": 15326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.837138159501898e-06,
+      "loss": 1.0183,
+      "step": 15327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.835893941079571e-06,
+      "loss": 0.9814,
+      "step": 15328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.834649800638683e-06,
+      "loss": 0.9589,
+      "step": 15329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.833405738202536e-06,
+      "loss": 0.9633,
+      "step": 15330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.83216175379443e-06,
+      "loss": 1.0562,
+      "step": 15331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.830917847437649e-06,
+      "loss": 0.9517,
+      "step": 15332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.829674019155507e-06,
+      "loss": 0.9902,
+      "step": 15333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.828430268971279e-06,
+      "loss": 0.9296,
+      "step": 15334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.827186596908272e-06,
+      "loss": 1.0163,
+      "step": 15335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.825943002989767e-06,
+      "loss": 0.9717,
+      "step": 15336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.824699487239054e-06,
+      "loss": 0.9547,
+      "step": 15337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.823456049679421e-06,
+      "loss": 0.9978,
+      "step": 15338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.822212690334157e-06,
+      "loss": 1.0233,
+      "step": 15339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.820969409226544e-06,
+      "loss": 0.9912,
+      "step": 15340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.819726206379864e-06,
+      "loss": 1.0091,
+      "step": 15341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.8184830818174e-06,
+      "loss": 0.9589,
+      "step": 15342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.81724003556243e-06,
+      "loss": 1.0005,
+      "step": 15343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.815997067638235e-06,
+      "loss": 1.0213,
+      "step": 15344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.81475417806809e-06,
+      "loss": 0.9286,
+      "step": 15345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.813511366875278e-06,
+      "loss": 1.0113,
+      "step": 15346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.812268634083061e-06,
+      "loss": 1.0046,
+      "step": 15347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.811025979714718e-06,
+      "loss": 0.9668,
+      "step": 15348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.809783403793521e-06,
+      "loss": 1.0057,
+      "step": 15349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.808540906342739e-06,
+      "loss": 0.9669,
+      "step": 15350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.807298487385645e-06,
+      "loss": 1.0026,
+      "step": 15351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.80605614694549e-06,
+      "loss": 1.0064,
+      "step": 15352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.80481388504556e-06,
+      "loss": 0.9412,
+      "step": 15353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.803571701709104e-06,
+      "loss": 0.9612,
+      "step": 15354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.802329596959394e-06,
+      "loss": 0.9199,
+      "step": 15355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.801087570819686e-06,
+      "loss": 0.9566,
+      "step": 15356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.799845623313239e-06,
+      "loss": 1.0372,
+      "step": 15357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.798603754463315e-06,
+      "loss": 0.9635,
+      "step": 15358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.79736196429316e-06,
+      "loss": 0.9651,
+      "step": 15359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.796120252826046e-06,
+      "loss": 0.9947,
+      "step": 15360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.794878620085213e-06,
+      "loss": 0.9372,
+      "step": 15361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.793637066093917e-06,
+      "loss": 0.965,
+      "step": 15362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.79239559087541e-06,
+      "loss": 0.9761,
+      "step": 15363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.791154194452942e-06,
+      "loss": 0.9547,
+      "step": 15364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.789912876849756e-06,
+      "loss": 0.9409,
+      "step": 15365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.788671638089107e-06,
+      "loss": 1.0638,
+      "step": 15366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.78743047819423e-06,
+      "loss": 1.0128,
+      "step": 15367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.786189397188374e-06,
+      "loss": 0.9645,
+      "step": 15368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.784948395094776e-06,
+      "loss": 0.9567,
+      "step": 15369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.78370747193668e-06,
+      "loss": 0.958,
+      "step": 15370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.782466627737331e-06,
+      "loss": 1.0115,
+      "step": 15371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.78122586251995e-06,
+      "loss": 0.9646,
+      "step": 15372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77998517630779e-06,
+      "loss": 1.079,
+      "step": 15373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.778744569124074e-06,
+      "loss": 0.9982,
+      "step": 15374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77750404099204e-06,
+      "loss": 1.0183,
+      "step": 15375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.776263591934917e-06,
+      "loss": 0.9014,
+      "step": 15376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.775023221975936e-06,
+      "loss": 0.9988,
+      "step": 15377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77378293113833e-06,
+      "loss": 0.9841,
+      "step": 15378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.772542719445315e-06,
+      "loss": 0.9254,
+      "step": 15379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.77130258692013e-06,
+      "loss": 0.9566,
+      "step": 15380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.770062533585989e-06,
+      "loss": 0.9374,
+      "step": 15381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.768822559466117e-06,
+      "loss": 0.9631,
+      "step": 15382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.767582664583737e-06,
+      "loss": 0.9275,
+      "step": 15383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.766342848962067e-06,
+      "loss": 0.9266,
+      "step": 15384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.765103112624328e-06,
+      "loss": 0.9803,
+      "step": 15385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.763863455593732e-06,
+      "loss": 1.0037,
+      "step": 15386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.762623877893495e-06,
+      "loss": 0.9766,
+      "step": 15387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.761384379546834e-06,
+      "loss": 0.9288,
+      "step": 15388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.76014496057696e-06,
+      "loss": 0.9447,
+      "step": 15389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.758905621007077e-06,
+      "loss": 0.9991,
+      "step": 15390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.757666360860407e-06,
+      "loss": 1.0184,
+      "step": 15391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.756427180160145e-06,
+      "loss": 1.0283,
+      "step": 15392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.755188078929507e-06,
+      "loss": 1.0101,
+      "step": 15393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.753949057191693e-06,
+      "loss": 1.0003,
+      "step": 15394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.752710114969903e-06,
+      "loss": 1.0181,
+      "step": 15395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.75147125228735e-06,
+      "loss": 1.0877,
+      "step": 15396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.750232469167218e-06,
+      "loss": 0.9106,
+      "step": 15397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.748993765632723e-06,
+      "loss": 1.028,
+      "step": 15398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.747755141707048e-06,
+      "loss": 0.9142,
+      "step": 15399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.746516597413395e-06,
+      "loss": 0.9751,
+      "step": 15400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.745278132774957e-06,
+      "loss": 0.9541,
+      "step": 15401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.744039747814929e-06,
+      "loss": 0.997,
+      "step": 15402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.7428014425565e-06,
+      "loss": 0.9162,
+      "step": 15403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.741563217022864e-06,
+      "loss": 1.0178,
+      "step": 15404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.740325071237205e-06,
+      "loss": 0.9503,
+      "step": 15405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.73908700522271e-06,
+      "loss": 1.01,
+      "step": 15406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.737849019002564e-06,
+      "loss": 0.959,
+      "step": 15407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.736611112599952e-06,
+      "loss": 0.9608,
+      "step": 15408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.735373286038061e-06,
+      "loss": 0.9845,
+      "step": 15409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.73413553934006e-06,
+      "loss": 0.9706,
+      "step": 15410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.732897872529142e-06,
+      "loss": 1.079,
+      "step": 15411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.731660285628475e-06,
+      "loss": 0.9697,
+      "step": 15412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.73042277866124e-06,
+      "loss": 0.984,
+      "step": 15413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.72918535165061e-06,
+      "loss": 0.9732,
+      "step": 15414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.727948004619759e-06,
+      "loss": 0.9809,
+      "step": 15415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.726710737591863e-06,
+      "loss": 0.9583,
+      "step": 15416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.72547355059008e-06,
+      "loss": 1.0045,
+      "step": 15417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.724236443637594e-06,
+      "loss": 1.0097,
+      "step": 15418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.722999416757562e-06,
+      "loss": 0.9874,
+      "step": 15419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.721762469973155e-06,
+      "loss": 0.9794,
+      "step": 15420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.720525603307535e-06,
+      "loss": 0.9672,
+      "step": 15421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.719288816783864e-06,
+      "loss": 1.0179,
+      "step": 15422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.71805211042531e-06,
+      "loss": 1.0105,
+      "step": 15423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.716815484255024e-06,
+      "loss": 1.0053,
+      "step": 15424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.71557893829617e-06,
+      "loss": 0.9017,
+      "step": 15425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.7143424725719e-06,
+      "loss": 0.9722,
+      "step": 15426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.713106087105374e-06,
+      "loss": 0.9999,
+      "step": 15427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.711869781919742e-06,
+      "loss": 0.9347,
+      "step": 15428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.710633557038167e-06,
+      "loss": 1.0196,
+      "step": 15429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.709397412483781e-06,
+      "loss": 0.9338,
+      "step": 15430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.708161348279751e-06,
+      "loss": 0.9689,
+      "step": 15431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.706925364449215e-06,
+      "loss": 1.0386,
+      "step": 15432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.705689461015322e-06,
+      "loss": 0.9581,
+      "step": 15433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.70445363800122e-06,
+      "loss": 1.0312,
+      "step": 15434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.703217895430042e-06,
+      "loss": 0.9774,
+      "step": 15435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.701982233324945e-06,
+      "loss": 1.0036,
+      "step": 15436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.700746651709052e-06,
+      "loss": 0.8948,
+      "step": 15437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.69951115060552e-06,
+      "loss": 0.9808,
+      "step": 15438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.698275730037474e-06,
+      "loss": 1.0261,
+      "step": 15439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.697040390028054e-06,
+      "loss": 0.9855,
+      "step": 15440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.695805130600395e-06,
+      "loss": 1.0071,
+      "step": 15441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.694569951777622e-06,
+      "loss": 0.9563,
+      "step": 15442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.69333485358288e-06,
+      "loss": 0.9894,
+      "step": 15443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.692099836039287e-06,
+      "loss": 0.9944,
+      "step": 15444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.690864899169976e-06,
+      "loss": 0.973,
+      "step": 15445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.689630042998073e-06,
+      "loss": 1.027,
+      "step": 15446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.688395267546704e-06,
+      "loss": 0.9434,
+      "step": 15447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.68716057283899e-06,
+      "loss": 0.9717,
+      "step": 15448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.685925958898061e-06,
+      "loss": 1.0504,
+      "step": 15449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.684691425747028e-06,
+      "loss": 0.9984,
+      "step": 15450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.683456973409014e-06,
+      "loss": 0.9568,
+      "step": 15451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.682222601907138e-06,
+      "loss": 0.9968,
+      "step": 15452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.680988311264515e-06,
+      "loss": 1.0214,
+      "step": 15453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.679754101504262e-06,
+      "loss": 1.0419,
+      "step": 15454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.678519972649481e-06,
+      "loss": 1.031,
+      "step": 15455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.677285924723303e-06,
+      "loss": 0.9691,
+      "step": 15456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.676051957748822e-06,
+      "loss": 1.017,
+      "step": 15457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.674818071749153e-06,
+      "loss": 0.9449,
+      "step": 15458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.6735842667474e-06,
+      "loss": 0.9543,
+      "step": 15459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.67235054276667e-06,
+      "loss": 0.92,
+      "step": 15460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.671116899830075e-06,
+      "loss": 1.0207,
+      "step": 15461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.669883337960704e-06,
+      "loss": 0.9997,
+      "step": 15462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.668649857181664e-06,
+      "loss": 0.9494,
+      "step": 15463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.667416457516057e-06,
+      "loss": 1.0535,
+      "step": 15464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.666183138986976e-06,
+      "loss": 0.996,
+      "step": 15465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66494990161752e-06,
+      "loss": 0.9986,
+      "step": 15466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66371674543079e-06,
+      "loss": 1.0216,
+      "step": 15467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.662483670449866e-06,
+      "loss": 1.0175,
+      "step": 15468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.66125067669785e-06,
+      "loss": 0.9791,
+      "step": 15469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.660017764197827e-06,
+      "loss": 0.9365,
+      "step": 15470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.65878493297289e-06,
+      "loss": 0.9859,
+      "step": 15471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.657552183046127e-06,
+      "loss": 0.9783,
+      "step": 15472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.656319514440614e-06,
+      "loss": 0.9673,
+      "step": 15473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.65508692717945e-06,
+      "loss": 0.9527,
+      "step": 15474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.653854421285703e-06,
+      "loss": 0.99,
+      "step": 15475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.652621996782468e-06,
+      "loss": 0.9598,
+      "step": 15476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.651389653692816e-06,
+      "loss": 1.0117,
+      "step": 15477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.650157392039825e-06,
+      "loss": 0.9539,
+      "step": 15478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.648925211846581e-06,
+      "loss": 1.0237,
+      "step": 15479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.647693113136141e-06,
+      "loss": 0.9247,
+      "step": 15480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.646461095931601e-06,
+      "loss": 0.972,
+      "step": 15481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.645229160256017e-06,
+      "loss": 0.9141,
+      "step": 15482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.643997306132461e-06,
+      "loss": 1.0083,
+      "step": 15483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.642765533584007e-06,
+      "loss": 0.9736,
+      "step": 15484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.641533842633722e-06,
+      "loss": 0.9758,
+      "step": 15485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.640302233304669e-06,
+      "loss": 0.9577,
+      "step": 15486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.639070705619918e-06,
+      "loss": 0.9577,
+      "step": 15487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.637839259602525e-06,
+      "loss": 1.0272,
+      "step": 15488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.636607895275554e-06,
+      "loss": 0.9965,
+      "step": 15489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.635376612662064e-06,
+      "loss": 0.9696,
+      "step": 15490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.634145411785117e-06,
+      "loss": 1.0182,
+      "step": 15491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.632914292667769e-06,
+      "loss": 1.0115,
+      "step": 15492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.631683255333066e-06,
+      "loss": 1.0327,
+      "step": 15493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.630452299804077e-06,
+      "loss": 0.9144,
+      "step": 15494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.629221426103843e-06,
+      "loss": 0.9672,
+      "step": 15495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.627990634255416e-06,
+      "loss": 0.95,
+      "step": 15496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.626759924281848e-06,
+      "loss": 1.0029,
+      "step": 15497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.625529296206186e-06,
+      "loss": 0.9804,
+      "step": 15498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.62429875005148e-06,
+      "loss": 0.9792,
+      "step": 15499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.623068285840761e-06,
+      "loss": 0.9815,
+      "step": 15500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.621837903597089e-06,
+      "loss": 0.9909,
+      "step": 15501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.620607603343495e-06,
+      "loss": 0.9377,
+      "step": 15502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.61937738510302e-06,
+      "loss": 0.9406,
+      "step": 15503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.618147248898705e-06,
+      "loss": 0.9546,
+      "step": 15504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.616917194753586e-06,
+      "loss": 0.9282,
+      "step": 15505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.6156872226907e-06,
+      "loss": 1.0471,
+      "step": 15506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.614457332733077e-06,
+      "loss": 1.0379,
+      "step": 15507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.613227524903748e-06,
+      "loss": 0.9096,
+      "step": 15508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.611997799225748e-06,
+      "loss": 0.9709,
+      "step": 15509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.610768155722106e-06,
+      "loss": 0.9806,
+      "step": 15510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.609538594415846e-06,
+      "loss": 0.9691,
+      "step": 15511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.608309115330003e-06,
+      "loss": 0.9627,
+      "step": 15512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.607079718487584e-06,
+      "loss": 0.9426,
+      "step": 15513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.605850403911633e-06,
+      "loss": 1.0495,
+      "step": 15514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.6046211716251576e-06,
+      "loss": 0.9496,
+      "step": 15515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.603392021651179e-06,
+      "loss": 1.0027,
+      "step": 15516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.602162954012725e-06,
+      "loss": 0.9856,
+      "step": 15517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.600933968732795e-06,
+      "loss": 0.9871,
+      "step": 15518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.599705065834424e-06,
+      "loss": 0.9303,
+      "step": 15519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5984762453406075e-06,
+      "loss": 1.0221,
+      "step": 15520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5972475072743755e-06,
+      "loss": 0.9318,
+      "step": 15521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.596018851658725e-06,
+      "loss": 0.9527,
+      "step": 15522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.594790278516671e-06,
+      "loss": 1.0021,
+      "step": 15523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5935617878712174e-06,
+      "loss": 1.0152,
+      "step": 15524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.592333379745379e-06,
+      "loss": 0.9864,
+      "step": 15525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.591105054162149e-06,
+      "loss": 0.9633,
+      "step": 15526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.589876811144537e-06,
+      "loss": 0.9665,
+      "step": 15527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.58864865071554e-06,
+      "loss": 0.9831,
+      "step": 15528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.587420572898162e-06,
+      "loss": 0.9733,
+      "step": 15529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.586192577715403e-06,
+      "loss": 1.0256,
+      "step": 15530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.58496466519025e-06,
+      "loss": 0.9969,
+      "step": 15531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.583736835345712e-06,
+      "loss": 0.9895,
+      "step": 15532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.58250908820477e-06,
+      "loss": 0.9707,
+      "step": 15533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.581281423790424e-06,
+      "loss": 0.9643,
+      "step": 15534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5800538421256604e-06,
+      "loss": 0.9977,
+      "step": 15535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57882634323347e-06,
+      "loss": 0.979,
+      "step": 15536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5775989271368425e-06,
+      "loss": 0.9689,
+      "step": 15537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.576371593858755e-06,
+      "loss": 1.0466,
+      "step": 15538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.575144343422205e-06,
+      "loss": 0.9634,
+      "step": 15539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.573917175850166e-06,
+      "loss": 0.9429,
+      "step": 15540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57269009116562e-06,
+      "loss": 0.9443,
+      "step": 15541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.571463089391548e-06,
+      "loss": 1.0198,
+      "step": 15542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.57023617055093e-06,
+      "loss": 1.022,
+      "step": 15543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5690093346667454e-06,
+      "loss": 0.9605,
+      "step": 15544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.56778258176196e-06,
+      "loss": 0.9837,
+      "step": 15545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.566555911859552e-06,
+      "loss": 0.9369,
+      "step": 15546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.565329324982494e-06,
+      "loss": 1.0028,
+      "step": 15547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.564102821153754e-06,
+      "loss": 0.978,
+      "step": 15548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.562876400396304e-06,
+      "loss": 1.0107,
+      "step": 15549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.561650062733114e-06,
+      "loss": 1.0186,
+      "step": 15550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.560423808187136e-06,
+      "loss": 0.9861,
+      "step": 15551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.559197636781352e-06,
+      "loss": 1.032,
+      "step": 15552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.557971548538715e-06,
+      "loss": 0.9976,
+      "step": 15553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.556745543482185e-06,
+      "loss": 1.0205,
+      "step": 15554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.555519621634728e-06,
+      "loss": 0.9796,
+      "step": 15555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5542937830192885e-06,
+      "loss": 0.9204,
+      "step": 15556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.553068027658841e-06,
+      "loss": 0.9774,
+      "step": 15557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5518423555763245e-06,
+      "loss": 1.0489,
+      "step": 15558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.550616766794706e-06,
+      "loss": 1.0152,
+      "step": 15559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.549391261336925e-06,
+      "loss": 1.0195,
+      "step": 15560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.548165839225939e-06,
+      "loss": 1.0419,
+      "step": 15561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.546940500484692e-06,
+      "loss": 0.9772,
+      "step": 15562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.545715245136135e-06,
+      "loss": 1.0345,
+      "step": 15563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5444900732032145e-06,
+      "loss": 0.9945,
+      "step": 15564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.543264984708869e-06,
+      "loss": 0.9951,
+      "step": 15565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.542039979676042e-06,
+      "loss": 1.0061,
+      "step": 15566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.540815058127677e-06,
+      "loss": 0.9811,
+      "step": 15567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.539590220086709e-06,
+      "loss": 0.9746,
+      "step": 15568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5383654655760824e-06,
+      "loss": 1.0004,
+      "step": 15569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.53714079461873e-06,
+      "loss": 1.0109,
+      "step": 15570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.535916207237584e-06,
+      "loss": 1.0031,
+      "step": 15571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.534691703455577e-06,
+      "loss": 0.9999,
+      "step": 15572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.533467283295644e-06,
+      "loss": 1.0051,
+      "step": 15573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.53224294678071e-06,
+      "loss": 0.9886,
+      "step": 15574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.531018693933713e-06,
+      "loss": 1.0701,
+      "step": 15575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.529794524777562e-06,
+      "loss": 0.9283,
+      "step": 15576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.528570439335204e-06,
+      "loss": 0.9517,
+      "step": 15577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5273464376295425e-06,
+      "loss": 0.9417,
+      "step": 15578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5261225196835165e-06,
+      "loss": 0.9922,
+      "step": 15579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.524898685520033e-06,
+      "loss": 0.9073,
+      "step": 15580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.523674935162017e-06,
+      "loss": 1.0445,
+      "step": 15581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.52245126863239e-06,
+      "loss": 1.0035,
+      "step": 15582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.521227685954051e-06,
+      "loss": 0.9624,
+      "step": 15583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.520004187149936e-06,
+      "loss": 1.0034,
+      "step": 15584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.5187807722429435e-06,
+      "loss": 1.0177,
+      "step": 15585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.517557441255988e-06,
+      "loss": 0.9895,
+      "step": 15586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.516334194211979e-06,
+      "loss": 0.9837,
+      "step": 15587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.515111031133823e-06,
+      "loss": 0.9765,
+      "step": 15588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.513887952044428e-06,
+      "loss": 0.9773,
+      "step": 15589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.512664956966702e-06,
+      "loss": 1.037,
+      "step": 15590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.51144204592354e-06,
+      "loss": 0.9883,
+      "step": 15591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.510219218937849e-06,
+      "loss": 0.9819,
+      "step": 15592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.508996476032531e-06,
+      "loss": 0.9468,
+      "step": 15593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.507773817230473e-06,
+      "loss": 1.0185,
+      "step": 15594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.506551242554589e-06,
+      "loss": 1.0404,
+      "step": 15595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.505328752027756e-06,
+      "loss": 1.0226,
+      "step": 15596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.504106345672887e-06,
+      "loss": 1.0075,
+      "step": 15597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.502884023512857e-06,
+      "loss": 0.9883,
+      "step": 15598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.501661785570564e-06,
+      "loss": 0.9872,
+      "step": 15599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.500439631868896e-06,
+      "loss": 1.0217,
+      "step": 15600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.49921756243074e-06,
+      "loss": 0.9263,
+      "step": 15601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.497995577278989e-06,
+      "loss": 0.9596,
+      "step": 15602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.496773676436516e-06,
+      "loss": 0.9896,
+      "step": 15603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4955518599262075e-06,
+      "loss": 1.0092,
+      "step": 15604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.494330127770943e-06,
+      "loss": 1.0304,
+      "step": 15605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.493108479993608e-06,
+      "loss": 0.9909,
+      "step": 15606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4918869166170734e-06,
+      "loss": 1.0366,
+      "step": 15607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.490665437664225e-06,
+      "loss": 0.9191,
+      "step": 15608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.489444043157925e-06,
+      "loss": 1.0386,
+      "step": 15609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.488222733121054e-06,
+      "loss": 0.9677,
+      "step": 15610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.48700150757648e-06,
+      "loss": 0.9808,
+      "step": 15611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.485780366547078e-06,
+      "loss": 0.96,
+      "step": 15612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.484559310055715e-06,
+      "loss": 1.0063,
+      "step": 15613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.48333833812525e-06,
+      "loss": 0.9621,
+      "step": 15614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.482117450778562e-06,
+      "loss": 1.0096,
+      "step": 15615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.480896648038498e-06,
+      "loss": 1.0568,
+      "step": 15616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.479675929927937e-06,
+      "loss": 0.9758,
+      "step": 15617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.478455296469729e-06,
+      "loss": 1.0247,
+      "step": 15618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.477234747686735e-06,
+      "loss": 1.0119,
+      "step": 15619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.476014283601819e-06,
+      "loss": 0.9453,
+      "step": 15620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.474793904237818e-06,
+      "loss": 0.9924,
+      "step": 15621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.47357360961761e-06,
+      "loss": 0.926,
+      "step": 15622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4723533997640295e-06,
+      "loss": 0.9646,
+      "step": 15623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.471133274699936e-06,
+      "loss": 1.0257,
+      "step": 15624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.469913234448176e-06,
+      "loss": 0.9483,
+      "step": 15625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.468693279031599e-06,
+      "loss": 0.9965,
+      "step": 15626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.467473408473054e-06,
+      "loss": 0.9638,
+      "step": 15627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.466253622795377e-06,
+      "loss": 1.0364,
+      "step": 15628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.465033922021417e-06,
+      "loss": 0.9085,
+      "step": 15629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.463814306174014e-06,
+      "loss": 0.9794,
+      "step": 15630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.462594775276006e-06,
+      "loss": 0.9963,
+      "step": 15631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.461375329350236e-06,
+      "loss": 1.006,
+      "step": 15632
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.46015596841954e-06,
+      "loss": 0.9644,
+      "step": 15633
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.458936692506744e-06,
+      "loss": 0.9749,
+      "step": 15634
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.457717501634697e-06,
+      "loss": 0.9696,
+      "step": 15635
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.456498395826218e-06,
+      "loss": 0.9667,
+      "step": 15636
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.45527937510414e-06,
+      "loss": 0.9737,
+      "step": 15637
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.454060439491297e-06,
+      "loss": 0.9553,
+      "step": 15638
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4528415890105075e-06,
+      "loss": 0.928,
+      "step": 15639
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.451622823684606e-06,
+      "loss": 0.9708,
+      "step": 15640
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4504041435364075e-06,
+      "loss": 0.9602,
+      "step": 15641
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.449185548588744e-06,
+      "loss": 0.977,
+      "step": 15642
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.44796703886443e-06,
+      "loss": 0.9088,
+      "step": 15643
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.446748614386284e-06,
+      "loss": 0.9913,
+      "step": 15644
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.445530275177124e-06,
+      "loss": 0.9517,
+      "step": 15645
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.44431202125977e-06,
+      "loss": 0.9465,
+      "step": 15646
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.443093852657035e-06,
+      "loss": 0.9696,
+      "step": 15647
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.441875769391726e-06,
+      "loss": 0.9798,
+      "step": 15648
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.440657771486662e-06,
+      "loss": 1.0442,
+      "step": 15649
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.439439858964645e-06,
+      "loss": 0.9072,
+      "step": 15650
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4382220318484875e-06,
+      "loss": 1.0054,
+      "step": 15651
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.437004290160995e-06,
+      "loss": 0.9464,
+      "step": 15652
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.435786633924978e-06,
+      "loss": 0.9176,
+      "step": 15653
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.434569063163227e-06,
+      "loss": 1.0163,
+      "step": 15654
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.433351577898552e-06,
+      "loss": 1.0171,
+      "step": 15655
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.432134178153752e-06,
+      "loss": 0.9638,
+      "step": 15656
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.430916863951623e-06,
+      "loss": 0.9984,
+      "step": 15657
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4296996353149665e-06,
+      "loss": 0.9721,
+      "step": 15658
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4284824922665694e-06,
+      "loss": 0.984,
+      "step": 15659
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.427265434829236e-06,
+      "loss": 0.9345,
+      "step": 15660
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.426048463025749e-06,
+      "loss": 0.9826,
+      "step": 15661
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.424831576878902e-06,
+      "loss": 0.9361,
+      "step": 15662
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.423614776411483e-06,
+      "loss": 1.0788,
+      "step": 15663
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.422398061646282e-06,
+      "loss": 0.9446,
+      "step": 15664
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.421181432606083e-06,
+      "loss": 0.9981,
+      "step": 15665
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4199648893136674e-06,
+      "loss": 1.0047,
+      "step": 15666
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.41874843179182e-06,
+      "loss": 0.9119,
+      "step": 15667
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.41753206006332e-06,
+      "loss": 0.9248,
+      "step": 15668
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.416315774150946e-06,
+      "loss": 0.9553,
+      "step": 15669
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.415099574077478e-06,
+      "loss": 0.9798,
+      "step": 15670
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.413883459865694e-06,
+      "loss": 0.9601,
+      "step": 15671
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.412667431538356e-06,
+      "loss": 1.0237,
+      "step": 15672
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.411451489118255e-06,
+      "loss": 0.9873,
+      "step": 15673
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.410235632628147e-06,
+      "loss": 0.9521,
+      "step": 15674
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.409019862090808e-06,
+      "loss": 0.9313,
+      "step": 15675
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.407804177529008e-06,
+      "loss": 1.0016,
+      "step": 15676
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4065885789655034e-06,
+      "loss": 0.9815,
+      "step": 15677
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.4053730664230716e-06,
+      "loss": 0.9355,
+      "step": 15678
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.404157639924461e-06,
+      "loss": 1.0229,
+      "step": 15679
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.402942299492449e-06,
+      "loss": 1.022,
+      "step": 15680
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.401727045149785e-06,
+      "loss": 0.9517,
+      "step": 15681
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.400511876919227e-06,
+      "loss": 0.9751,
+      "step": 15682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.399296794823537e-06,
+      "loss": 0.9908,
+      "step": 15683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.398081798885466e-06,
+      "loss": 0.9375,
+      "step": 15684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3968668891277715e-06,
+      "loss": 0.9749,
+      "step": 15685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.395652065573199e-06,
+      "loss": 0.9,
+      "step": 15686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.394437328244501e-06,
+      "loss": 0.9555,
+      "step": 15687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3932226771644275e-06,
+      "loss": 1.0138,
+      "step": 15688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.392008112355724e-06,
+      "loss": 0.9441,
+      "step": 15689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.390793633841136e-06,
+      "loss": 0.9876,
+      "step": 15690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3895792416434095e-06,
+      "loss": 0.9948,
+      "step": 15691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3883649357852815e-06,
+      "loss": 0.9881,
+      "step": 15692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.387150716289496e-06,
+      "loss": 1.0311,
+      "step": 15693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.385936583178789e-06,
+      "loss": 0.9881,
+      "step": 15694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.384722536475901e-06,
+      "loss": 0.9919,
+      "step": 15695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3835085762035694e-06,
+      "loss": 0.9129,
+      "step": 15696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.382294702384517e-06,
+      "loss": 0.9845,
+      "step": 15697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.38108091504149e-06,
+      "loss": 0.8866,
+      "step": 15698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3798672141972055e-06,
+      "loss": 0.9692,
+      "step": 15699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.378653599874409e-06,
+      "loss": 0.9938,
+      "step": 15700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.377440072095813e-06,
+      "loss": 0.9684,
+      "step": 15701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.376226630884151e-06,
+      "loss": 0.9622,
+      "step": 15702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.375013276262149e-06,
+      "loss": 0.9589,
+      "step": 15703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.373800008252519e-06,
+      "loss": 0.9074,
+      "step": 15704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.372586826877994e-06,
+      "loss": 0.9806,
+      "step": 15705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.371373732161285e-06,
+      "loss": 0.9876,
+      "step": 15706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.370160724125112e-06,
+      "loss": 0.9601,
+      "step": 15707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.368947802792195e-06,
+      "loss": 0.9574,
+      "step": 15708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.367734968185242e-06,
+      "loss": 0.9784,
+      "step": 15709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.366522220326971e-06,
+      "loss": 1.0009,
+      "step": 15710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.365309559240093e-06,
+      "loss": 0.9964,
+      "step": 15711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.364096984947313e-06,
+      "loss": 1.0642,
+      "step": 15712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.362884497471344e-06,
+      "loss": 1.0004,
+      "step": 15713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.361672096834886e-06,
+      "loss": 0.9501,
+      "step": 15714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.360459783060651e-06,
+      "loss": 0.9376,
+      "step": 15715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.359247556171341e-06,
+      "loss": 1.0298,
+      "step": 15716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.358035416189649e-06,
+      "loss": 1.0569,
+      "step": 15717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.356823363138288e-06,
+      "loss": 1.0659,
+      "step": 15718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.355611397039946e-06,
+      "loss": 0.9974,
+      "step": 15719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.354399517917322e-06,
+      "loss": 1.0468,
+      "step": 15720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.353187725793114e-06,
+      "loss": 0.9523,
+      "step": 15721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.351976020690011e-06,
+      "loss": 0.9615,
+      "step": 15722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.35076440263071e-06,
+      "loss": 1.032,
+      "step": 15723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.349552871637891e-06,
+      "loss": 0.9965,
+      "step": 15724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3483414277342565e-06,
+      "loss": 1.0037,
+      "step": 15725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.347130070942484e-06,
+      "loss": 1.0205,
+      "step": 15726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34591880128526e-06,
+      "loss": 1.0056,
+      "step": 15727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.344707618785266e-06,
+      "loss": 0.8929,
+      "step": 15728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.343496523465188e-06,
+      "loss": 1.0108,
+      "step": 15729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.34228551534771e-06,
+      "loss": 1.0082,
+      "step": 15730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3410745944555e-06,
+      "loss": 0.9873,
+      "step": 15731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339863760811242e-06,
+      "loss": 1.0215,
+      "step": 15732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.338653014437607e-06,
+      "loss": 0.9996,
+      "step": 15733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3374423553572775e-06,
+      "loss": 0.928,
+      "step": 15734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.336231783592911e-06,
+      "loss": 0.9829,
+      "step": 15735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3350212991671945e-06,
+      "loss": 0.982,
+      "step": 15736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.333810902102781e-06,
+      "loss": 0.9601,
+      "step": 15737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.332600592422353e-06,
+      "loss": 1.0175,
+      "step": 15738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.331390370148563e-06,
+      "loss": 1.0022,
+      "step": 15739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.330180235304082e-06,
+      "loss": 1.0411,
+      "step": 15740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.328970187911574e-06,
+      "loss": 1.0012,
+      "step": 15741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.32776022799369e-06,
+      "loss": 0.9874,
+      "step": 15742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3265503555731015e-06,
+      "loss": 0.9894,
+      "step": 15743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3253405706724565e-06,
+      "loss": 1.0356,
+      "step": 15744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.324130873314413e-06,
+      "loss": 1.0662,
+      "step": 15745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.322921263521626e-06,
+      "loss": 1.0017,
+      "step": 15746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.321711741316749e-06,
+      "loss": 1.0199,
+      "step": 15747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.32050230672243e-06,
+      "loss": 0.9732,
+      "step": 15748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.319292959761324e-06,
+      "loss": 0.9851,
+      "step": 15749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.31808370045607e-06,
+      "loss": 0.9817,
+      "step": 15750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.316874528829318e-06,
+      "loss": 1.0044,
+      "step": 15751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.315665444903714e-06,
+      "loss": 0.9199,
+      "step": 15752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.314456448701898e-06,
+      "loss": 1.0368,
+      "step": 15753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.313247540246516e-06,
+      "loss": 0.9806,
+      "step": 15754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.312038719560194e-06,
+      "loss": 1.0331,
+      "step": 15755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.310829986665587e-06,
+      "loss": 0.9622,
+      "step": 15756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.309621341585319e-06,
+      "loss": 1.0375,
+      "step": 15757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3084127843420285e-06,
+      "loss": 0.9854,
+      "step": 15758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.307204314958347e-06,
+      "loss": 1.0014,
+      "step": 15759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.305995933456909e-06,
+      "loss": 0.9855,
+      "step": 15760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.304787639860343e-06,
+      "loss": 0.9964,
+      "step": 15761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.303579434191267e-06,
+      "loss": 0.9634,
+      "step": 15762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.302371316472323e-06,
+      "loss": 1.0123,
+      "step": 15763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.3011632867261254e-06,
+      "loss": 1.0463,
+      "step": 15764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.299955344975298e-06,
+      "loss": 1.0727,
+      "step": 15765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.298747491242465e-06,
+      "loss": 1.0333,
+      "step": 15766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.297539725550243e-06,
+      "loss": 1.005,
+      "step": 15767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.296332047921256e-06,
+      "loss": 1.0459,
+      "step": 15768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2951244583781115e-06,
+      "loss": 1.0194,
+      "step": 15769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.293916956943428e-06,
+      "loss": 1.0198,
+      "step": 15770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.292709543639818e-06,
+      "loss": 0.9626,
+      "step": 15771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.291502218489892e-06,
+      "loss": 1.0481,
+      "step": 15772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.290294981516264e-06,
+      "loss": 0.9947,
+      "step": 15773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.28908783274154e-06,
+      "loss": 0.965,
+      "step": 15774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2878807721883186e-06,
+      "loss": 1.0318,
+      "step": 15775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.28667379987922e-06,
+      "loss": 0.9766,
+      "step": 15776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2854669158368325e-06,
+      "loss": 1.0511,
+      "step": 15777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.284260120083766e-06,
+      "loss": 0.9237,
+      "step": 15778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.28305341264262e-06,
+      "loss": 1.037,
+      "step": 15779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.281846793535982e-06,
+      "loss": 0.9961,
+      "step": 15780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.280640262786465e-06,
+      "loss": 0.9695,
+      "step": 15781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.279433820416647e-06,
+      "loss": 0.9624,
+      "step": 15782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.278227466449138e-06,
+      "loss": 0.972,
+      "step": 15783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.277021200906517e-06,
+      "loss": 0.9867,
+      "step": 15784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.275815023811377e-06,
+      "loss": 0.992,
+      "step": 15785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.274608935186307e-06,
+      "loss": 1.0278,
+      "step": 15786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.273402935053895e-06,
+      "loss": 0.9793,
+      "step": 15787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.272197023436727e-06,
+      "loss": 1.0207,
+      "step": 15788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.270991200357377e-06,
+      "loss": 0.9876,
+      "step": 15789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.269785465838438e-06,
+      "loss": 0.9872,
+      "step": 15790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.268579819902481e-06,
+      "loss": 0.9279,
+      "step": 15791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.267374262572089e-06,
+      "loss": 0.9949,
+      "step": 15792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.266168793869838e-06,
+      "loss": 0.9957,
+      "step": 15793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.264963413818305e-06,
+      "loss": 1.0081,
+      "step": 15794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2637581224400575e-06,
+      "loss": 1.0148,
+      "step": 15795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.262552919757671e-06,
+      "loss": 0.9643,
+      "step": 15796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.261347805793713e-06,
+      "loss": 0.9637,
+      "step": 15797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2601427805707544e-06,
+      "loss": 0.9463,
+      "step": 15798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.258937844111365e-06,
+      "loss": 0.9196,
+      "step": 15799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.257732996438096e-06,
+      "loss": 0.9444,
+      "step": 15800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.256528237573528e-06,
+      "loss": 0.9868,
+      "step": 15801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.255323567540211e-06,
+      "loss": 0.9911,
+      "step": 15802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2541189863607104e-06,
+      "loss": 0.9644,
+      "step": 15803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.252914494057579e-06,
+      "loss": 0.9937,
+      "step": 15804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.25171009065338e-06,
+      "loss": 1.006,
+      "step": 15805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.250505776170667e-06,
+      "loss": 0.9838,
+      "step": 15806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.24930155063199e-06,
+      "loss": 0.993,
+      "step": 15807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.248097414059901e-06,
+      "loss": 1.0424,
+      "step": 15808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.246893366476951e-06,
+      "loss": 1.0005,
+      "step": 15809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.245689407905688e-06,
+      "loss": 0.9556,
+      "step": 15810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.24448553836866e-06,
+      "loss": 0.9488,
+      "step": 15811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2432817578884134e-06,
+      "loss": 1.0016,
+      "step": 15812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2420780664874864e-06,
+      "loss": 1.0164,
+      "step": 15813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.240874464188422e-06,
+      "loss": 0.9516,
+      "step": 15814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.239670951013762e-06,
+      "loss": 0.9655,
+      "step": 15815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.238467526986044e-06,
+      "loss": 0.9292,
+      "step": 15816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.237264192127807e-06,
+      "loss": 0.9688,
+      "step": 15817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.236060946461577e-06,
+      "loss": 1.0017,
+      "step": 15818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.234857790009901e-06,
+      "loss": 0.94,
+      "step": 15819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.233654722795297e-06,
+      "loss": 1.0491,
+      "step": 15820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.232451744840306e-06,
+      "loss": 0.9737,
+      "step": 15821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.23124885616745e-06,
+      "loss": 0.9231,
+      "step": 15822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.230046056799256e-06,
+      "loss": 0.986,
+      "step": 15823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.228843346758253e-06,
+      "loss": 0.9917,
+      "step": 15824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2276407260669535e-06,
+      "loss": 1.0618,
+      "step": 15825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.226438194747896e-06,
+      "loss": 0.9213,
+      "step": 15826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.225235752823585e-06,
+      "loss": 0.9939,
+      "step": 15827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.224033400316545e-06,
+      "loss": 0.985,
+      "step": 15828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.222831137249295e-06,
+      "loss": 0.9629,
+      "step": 15829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2216289636443445e-06,
+      "loss": 0.9983,
+      "step": 15830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.22042687952421e-06,
+      "loss": 0.9815,
+      "step": 15831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.219224884911404e-06,
+      "loss": 0.9606,
+      "step": 15832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2180229798284344e-06,
+      "loss": 1.0378,
+      "step": 15833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.216821164297808e-06,
+      "loss": 0.9443,
+      "step": 15834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.21561943834203e-06,
+      "loss": 1.0107,
+      "step": 15835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2144178019836115e-06,
+      "loss": 1.0112,
+      "step": 15836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2132162552450534e-06,
+      "loss": 0.9718,
+      "step": 15837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.212014798148849e-06,
+      "loss": 0.969,
+      "step": 15838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2108134307175134e-06,
+      "loss": 0.9489,
+      "step": 15839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.209612152973531e-06,
+      "loss": 1.004,
+      "step": 15840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.208410964939403e-06,
+      "loss": 1.0268,
+      "step": 15841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.2072098666376254e-06,
+      "loss": 0.9381,
+      "step": 15842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.20600885809069e-06,
+      "loss": 1.0097,
+      "step": 15843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.204807939321092e-06,
+      "loss": 0.8976,
+      "step": 15844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.203607110351309e-06,
+      "loss": 0.9771,
+      "step": 15845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.202406371203848e-06,
+      "loss": 0.9415,
+      "step": 15846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.201205721901178e-06,
+      "loss": 0.9594,
+      "step": 15847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.200005162465791e-06,
+      "loss": 0.9894,
+      "step": 15848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19880469292017e-06,
+      "loss": 1.0419,
+      "step": 15849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.197604313286795e-06,
+      "loss": 1.0696,
+      "step": 15850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19640402358815e-06,
+      "loss": 0.9916,
+      "step": 15851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.195203823846706e-06,
+      "loss": 0.9753,
+      "step": 15852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.19400371408494e-06,
+      "loss": 0.9481,
+      "step": 15853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.192803694325333e-06,
+      "loss": 0.9844,
+      "step": 15854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1916037645903485e-06,
+      "loss": 0.977,
+      "step": 15855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.190403924902466e-06,
+      "loss": 0.9469,
+      "step": 15856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.189204175284154e-06,
+      "loss": 0.9719,
+      "step": 15857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.188004515757871e-06,
+      "loss": 0.9517,
+      "step": 15858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1868049463461e-06,
+      "loss": 0.9733,
+      "step": 15859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.185605467071289e-06,
+      "loss": 1.0447,
+      "step": 15860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1844060779559064e-06,
+      "loss": 1.0568,
+      "step": 15861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1832067790224195e-06,
+      "loss": 0.9765,
+      "step": 15862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.182007570293274e-06,
+      "loss": 1.0308,
+      "step": 15863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.180808451790944e-06,
+      "loss": 0.9403,
+      "step": 15864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1796094235378674e-06,
+      "loss": 1.0351,
+      "step": 15865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.178410485556519e-06,
+      "loss": 0.9837,
+      "step": 15866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.177211637869335e-06,
+      "loss": 0.9813,
+      "step": 15867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.176012880498772e-06,
+      "loss": 0.9533,
+      "step": 15868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.174814213467278e-06,
+      "loss": 0.9997,
+      "step": 15869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.173615636797308e-06,
+      "loss": 0.9996,
+      "step": 15870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.172417150511294e-06,
+      "loss": 0.9907,
+      "step": 15871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.171218754631691e-06,
+      "loss": 0.9565,
+      "step": 15872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.170020449180937e-06,
+      "loss": 0.9547,
+      "step": 15873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.168822234181472e-06,
+      "loss": 0.9602,
+      "step": 15874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.167624109655743e-06,
+      "loss": 0.9528,
+      "step": 15875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.166426075626172e-06,
+      "loss": 0.9409,
+      "step": 15876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.165228132115212e-06,
+      "loss": 0.9582,
+      "step": 15877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1640302791452845e-06,
+      "loss": 1.0079,
+      "step": 15878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.162832516738825e-06,
+      "loss": 0.9942,
+      "step": 15879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.161634844918266e-06,
+      "loss": 0.9382,
+      "step": 15880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.160437263706035e-06,
+      "loss": 0.9584,
+      "step": 15881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.159239773124565e-06,
+      "loss": 0.9761,
+      "step": 15882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.158042373196265e-06,
+      "loss": 0.9383,
+      "step": 15883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1568450639435815e-06,
+      "loss": 0.9367,
+      "step": 15884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.155647845388917e-06,
+      "loss": 0.9766,
+      "step": 15885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.154450717554702e-06,
+      "loss": 1.065,
+      "step": 15886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.153253680463352e-06,
+      "loss": 1.0321,
+      "step": 15887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.152056734137284e-06,
+      "loss": 1.0206,
+      "step": 15888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.15085987859892e-06,
+      "loss": 0.9089,
+      "step": 15889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.149663113870663e-06,
+      "loss": 0.9625,
+      "step": 15890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.14846643997493e-06,
+      "loss": 1.0046,
+      "step": 15891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.147269856934131e-06,
+      "loss": 0.993,
+      "step": 15892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.146073364770673e-06,
+      "loss": 0.9525,
+      "step": 15893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.144876963506965e-06,
+      "loss": 0.97,
+      "step": 15894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.1436806531654144e-06,
+      "loss": 1.0126,
+      "step": 15895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.142484433768413e-06,
+      "loss": 1.0263,
+      "step": 15896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.14128830533838e-06,
+      "loss": 1.0003,
+      "step": 15897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.140092267897701e-06,
+      "loss": 0.9927,
+      "step": 15898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.138896321468781e-06,
+      "loss": 1.0403,
+      "step": 15899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.137700466074019e-06,
+      "loss": 1.0021,
+      "step": 15900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.136504701735795e-06,
+      "loss": 1.048,
+      "step": 15901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.135309028476523e-06,
+      "loss": 0.96,
+      "step": 15902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.134113446318576e-06,
+      "loss": 0.9187,
+      "step": 15903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.132917955284358e-06,
+      "loss": 0.9617,
+      "step": 15904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.131722555396251e-06,
+      "loss": 0.9266,
+      "step": 15905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.130527246676639e-06,
+      "loss": 0.9459,
+      "step": 15906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.129332029147909e-06,
+      "loss": 0.9709,
+      "step": 15907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.128136902832443e-06,
+      "loss": 0.9852,
+      "step": 15908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.12694186775263e-06,
+      "loss": 0.9004,
+      "step": 15909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.125746923930836e-06,
+      "loss": 0.993,
+      "step": 15910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.124552071389448e-06,
+      "loss": 0.9505,
+      "step": 15911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.123357310150837e-06,
+      "loss": 1.001,
+      "step": 15912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.12216264023738e-06,
+      "loss": 0.9659,
+      "step": 15913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.120968061671452e-06,
+      "loss": 1.0031,
+      "step": 15914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.119773574475423e-06,
+      "loss": 1.0357,
+      "step": 15915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.118579178671657e-06,
+      "loss": 1.0004,
+      "step": 15916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.117384874282524e-06,
+      "loss": 1.0285,
+      "step": 15917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.116190661330393e-06,
+      "loss": 1.0357,
+      "step": 15918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.114996539837627e-06,
+      "loss": 0.9705,
+      "step": 15919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.113802509826588e-06,
+      "loss": 0.9538,
+      "step": 15920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.112608571319632e-06,
+      "loss": 0.9817,
+      "step": 15921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.111414724339128e-06,
+      "loss": 0.9803,
+      "step": 15922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.110220968907419e-06,
+      "loss": 1.0012,
+      "step": 15923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.109027305046876e-06,
+      "loss": 1.0116,
+      "step": 15924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.107833732779843e-06,
+      "loss": 0.9893,
+      "step": 15925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.106640252128674e-06,
+      "loss": 1.0143,
+      "step": 15926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.105446863115727e-06,
+      "loss": 1.0037,
+      "step": 15927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.104253565763332e-06,
+      "loss": 0.9801,
+      "step": 15928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.103060360093857e-06,
+      "loss": 0.9511,
+      "step": 15929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.101867246129633e-06,
+      "loss": 1.0246,
+      "step": 15930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.100674223893009e-06,
+      "loss": 0.902,
+      "step": 15931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.099481293406328e-06,
+      "loss": 0.9868,
+      "step": 15932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.098288454691926e-06,
+      "loss": 0.9704,
+      "step": 15933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.097095707772145e-06,
+      "loss": 0.9772,
+      "step": 15934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.095903052669323e-06,
+      "loss": 0.9808,
+      "step": 15935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.09471048940579e-06,
+      "loss": 0.9927,
+      "step": 15936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.09351801800388e-06,
+      "loss": 1.0002,
+      "step": 15937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.092325638485929e-06,
+      "loss": 0.9939,
+      "step": 15938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.091133350874258e-06,
+      "loss": 1.0128,
+      "step": 15939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.089941155191207e-06,
+      "loss": 0.9791,
+      "step": 15940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.088749051459089e-06,
+      "loss": 1.0085,
+      "step": 15941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.087557039700243e-06,
+      "loss": 1.0368,
+      "step": 15942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.08636511993698e-06,
+      "loss": 0.9798,
+      "step": 15943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.085173292191626e-06,
+      "loss": 0.9692,
+      "step": 15944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.083981556486501e-06,
+      "loss": 0.9353,
+      "step": 15945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.08278991284392e-06,
+      "loss": 1.0226,
+      "step": 15946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.081598361286206e-06,
+      "loss": 1.0457,
+      "step": 15947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0804069018356644e-06,
+      "loss": 0.9371,
+      "step": 15948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0792155345146095e-06,
+      "loss": 1.0112,
+      "step": 15949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.078024259345354e-06,
+      "loss": 0.9762,
+      "step": 15950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.076833076350208e-06,
+      "loss": 0.978,
+      "step": 15951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.075641985551477e-06,
+      "loss": 1.0004,
+      "step": 15952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.074450986971472e-06,
+      "loss": 0.9614,
+      "step": 15953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.073260080632488e-06,
+      "loss": 0.9855,
+      "step": 15954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.07206926655683e-06,
+      "loss": 1.0043,
+      "step": 15955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0708785447668024e-06,
+      "loss": 0.9512,
+      "step": 15956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.069687915284699e-06,
+      "loss": 1.0032,
+      "step": 15957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.068497378132825e-06,
+      "loss": 0.9701,
+      "step": 15958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0673069333334605e-06,
+      "loss": 1.0506,
+      "step": 15959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.066116580908916e-06,
+      "loss": 1.016,
+      "step": 15960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.064926320881468e-06,
+      "loss": 0.9983,
+      "step": 15961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.063736153273422e-06,
+      "loss": 0.9518,
+      "step": 15962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.062546078107056e-06,
+      "loss": 0.9445,
+      "step": 15963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.061356095404658e-06,
+      "loss": 0.9609,
+      "step": 15964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.060166205188516e-06,
+      "loss": 0.9883,
+      "step": 15965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.058976407480906e-06,
+      "loss": 0.9297,
+      "step": 15966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.057786702304121e-06,
+      "loss": 0.9991,
+      "step": 15967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.056597089680431e-06,
+      "loss": 0.9832,
+      "step": 15968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.055407569632116e-06,
+      "loss": 0.9903,
+      "step": 15969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.054218142181453e-06,
+      "loss": 0.8961,
+      "step": 15970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.053028807350719e-06,
+      "loss": 0.9681,
+      "step": 15971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0518395651621825e-06,
+      "loss": 1.0601,
+      "step": 15972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.050650415638122e-06,
+      "loss": 0.9928,
+      "step": 15973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0494613588007975e-06,
+      "loss": 0.9821,
+      "step": 15974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.048272394672479e-06,
+      "loss": 0.9502,
+      "step": 15975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.0470835232754326e-06,
+      "loss": 0.9873,
+      "step": 15976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.045894744631927e-06,
+      "loss": 0.9195,
+      "step": 15977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.044706058764222e-06,
+      "loss": 0.9497,
+      "step": 15978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.04351746569457e-06,
+      "loss": 1.0078,
+      "step": 15979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.042328965445245e-06,
+      "loss": 0.9739,
+      "step": 15980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.041140558038493e-06,
+      "loss": 0.9689,
+      "step": 15981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0399522434965705e-06,
+      "loss": 0.9742,
+      "step": 15982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.038764021841738e-06,
+      "loss": 0.9871,
+      "step": 15983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0375758930962355e-06,
+      "loss": 1.0105,
+      "step": 15984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.036387857282326e-06,
+      "loss": 0.9794,
+      "step": 15985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.035199914422245e-06,
+      "loss": 1.0057,
+      "step": 15986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.034012064538254e-06,
+      "loss": 0.9565,
+      "step": 15987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0328243076525855e-06,
+      "loss": 0.9695,
+      "step": 15988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.031636643787489e-06,
+      "loss": 0.9524,
+      "step": 15989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.030449072965204e-06,
+      "loss": 0.9844,
+      "step": 15990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.029261595207971e-06,
+      "loss": 0.9749,
+      "step": 15991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0280742105380285e-06,
+      "loss": 1.0689,
+      "step": 15992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.026886918977609e-06,
+      "loss": 0.9626,
+      "step": 15993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.025699720548951e-06,
+      "loss": 1.0181,
+      "step": 15994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.024512615274284e-06,
+      "loss": 0.9585,
+      "step": 15995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.023325603175841e-06,
+      "loss": 0.9355,
+      "step": 15996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.022138684275852e-06,
+      "loss": 0.9575,
+      "step": 15997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.020951858596548e-06,
+      "loss": 0.975,
+      "step": 15998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.019765126160146e-06,
+      "loss": 1.0302,
+      "step": 15999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.018578486988875e-06,
+      "loss": 1.0627,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "eval_text_loss": 0.9386805891990662,
+      "eval_text_runtime": 1.5595,
+      "eval_text_samples_per_second": 47.452,
+      "eval_text_steps_per_second": 1.282,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "eval_svit_loss": 1.0897661447525024,
+      "eval_svit_runtime": 3.7361,
+      "eval_svit_samples_per_second": 182.278,
+      "eval_svit_steps_per_second": 2.944,
+      "step": 16000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0173919411049535e-06,
+      "loss": 1.0247,
+      "step": 16001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.016205488530607e-06,
+      "loss": 0.9781,
+      "step": 16002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.0150191292880564e-06,
+      "loss": 1.0561,
+      "step": 16003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.013832863399507e-06,
+      "loss": 0.9609,
+      "step": 16004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.012646690887189e-06,
+      "loss": 0.986,
+      "step": 16005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.011460611773305e-06,
+      "loss": 1.0064,
+      "step": 16006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.010274626080071e-06,
+      "loss": 0.9958,
+      "step": 16007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.009088733829695e-06,
+      "loss": 0.9671,
+      "step": 16008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.007902935044388e-06,
+      "loss": 0.9805,
+      "step": 16009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.006717229746359e-06,
+      "loss": 0.956,
+      "step": 16010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.005531617957804e-06,
+      "loss": 0.9495,
+      "step": 16011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.004346099700932e-06,
+      "loss": 0.9673,
+      "step": 16012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.003160674997943e-06,
+      "loss": 0.9471,
+      "step": 16013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.001975343871038e-06,
+      "loss": 0.9891,
+      "step": 16014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.000790106342412e-06,
+      "loss": 0.9425,
+      "step": 16015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.999604962434265e-06,
+      "loss": 1.0003,
+      "step": 16016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.998419912168784e-06,
+      "loss": 0.9685,
+      "step": 16017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.997234955568172e-06,
+      "loss": 0.9416,
+      "step": 16018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.99605009265461e-06,
+      "loss": 0.9948,
+      "step": 16019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.994865323450292e-06,
+      "loss": 1.0145,
+      "step": 16020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9936806479774085e-06,
+      "loss": 0.9225,
+      "step": 16021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.992496066258133e-06,
+      "loss": 1.0021,
+      "step": 16022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.991311578314664e-06,
+      "loss": 0.9693,
+      "step": 16023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.990127184169171e-06,
+      "loss": 1.0076,
+      "step": 16024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.988942883843844e-06,
+      "loss": 0.9542,
+      "step": 16025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.987758677360855e-06,
+      "loss": 0.9274,
+      "step": 16026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.986574564742383e-06,
+      "loss": 0.9883,
+      "step": 16027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.985390546010603e-06,
+      "loss": 0.9562,
+      "step": 16028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.984206621187687e-06,
+      "loss": 0.9576,
+      "step": 16029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.98302279029581e-06,
+      "loss": 0.9521,
+      "step": 16030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.981839053357136e-06,
+      "loss": 1.0007,
+      "step": 16031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.980655410393837e-06,
+      "loss": 1.018,
+      "step": 16032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.979471861428078e-06,
+      "loss": 0.9694,
+      "step": 16033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.978288406482024e-06,
+      "loss": 0.9845,
+      "step": 16034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.977105045577835e-06,
+      "loss": 0.9937,
+      "step": 16035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.97592177873768e-06,
+      "loss": 1.0237,
+      "step": 16036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9747386059837064e-06,
+      "loss": 1.0175,
+      "step": 16037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9735555273380785e-06,
+      "loss": 0.9589,
+      "step": 16038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.972372542822951e-06,
+      "loss": 1.0175,
+      "step": 16039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9711896524604755e-06,
+      "loss": 1.0101,
+      "step": 16040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.97000685627281e-06,
+      "loss": 0.9425,
+      "step": 16041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.968824154282095e-06,
+      "loss": 0.9069,
+      "step": 16042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.96764154651049e-06,
+      "loss": 0.9895,
+      "step": 16043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.96645903298013e-06,
+      "loss": 1.0024,
+      "step": 16044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.965276613713175e-06,
+      "loss": 0.9351,
+      "step": 16045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.964094288731755e-06,
+      "loss": 1.0441,
+      "step": 16046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.962912058058015e-06,
+      "loss": 1.0234,
+      "step": 16047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9617299217141e-06,
+      "loss": 0.9649,
+      "step": 16048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9605478797221365e-06,
+      "loss": 0.9739,
+      "step": 16049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.959365932104277e-06,
+      "loss": 0.986,
+      "step": 16050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.95818407888264e-06,
+      "loss": 0.9262,
+      "step": 16051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9570023200793666e-06,
+      "loss": 0.9901,
+      "step": 16052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.955820655716587e-06,
+      "loss": 0.9888,
+      "step": 16053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.954639085816428e-06,
+      "loss": 0.9575,
+      "step": 16054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.953457610401018e-06,
+      "loss": 0.9734,
+      "step": 16055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.952276229492488e-06,
+      "loss": 0.9717,
+      "step": 16056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.95109494311295e-06,
+      "loss": 1.0175,
+      "step": 16057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.949913751284536e-06,
+      "loss": 0.9881,
+      "step": 16058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.948732654029361e-06,
+      "loss": 0.9591,
+      "step": 16059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.947551651369545e-06,
+      "loss": 0.9848,
+      "step": 16060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.946370743327208e-06,
+      "loss": 1.0312,
+      "step": 16061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.945189929924456e-06,
+      "loss": 0.9761,
+      "step": 16062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.944009211183416e-06,
+      "loss": 1.0025,
+      "step": 16063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.942828587126187e-06,
+      "loss": 0.955,
+      "step": 16064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.941648057774882e-06,
+      "loss": 0.9454,
+      "step": 16065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.940467623151613e-06,
+      "loss": 0.9754,
+      "step": 16066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.939287283278479e-06,
+      "loss": 0.9834,
+      "step": 16067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.938107038177596e-06,
+      "loss": 1.0285,
+      "step": 16068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.936926887871049e-06,
+      "loss": 0.954,
+      "step": 16069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.935746832380956e-06,
+      "loss": 0.9158,
+      "step": 16070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.934566871729404e-06,
+      "loss": 0.9358,
+      "step": 16071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.933387005938495e-06,
+      "loss": 1.0137,
+      "step": 16072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.932207235030325e-06,
+      "loss": 0.9448,
+      "step": 16073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.931027559026991e-06,
+      "loss": 0.9751,
+      "step": 16074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.929847977950576e-06,
+      "loss": 0.9044,
+      "step": 16075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.928668491823173e-06,
+      "loss": 1.0271,
+      "step": 16076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.927489100666875e-06,
+      "loss": 0.9069,
+      "step": 16077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.926309804503763e-06,
+      "loss": 0.9669,
+      "step": 16078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.92513060335593e-06,
+      "loss": 1.0108,
+      "step": 16079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9239514972454446e-06,
+      "loss": 0.9575,
+      "step": 16080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9227724861944055e-06,
+      "loss": 0.9771,
+      "step": 16081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.921593570224873e-06,
+      "loss": 0.9726,
+      "step": 16082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9204147493589445e-06,
+      "loss": 1.0128,
+      "step": 16083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.91923602361868e-06,
+      "loss": 0.9651,
+      "step": 16084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.918057393026162e-06,
+      "loss": 0.9456,
+      "step": 16085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.916878857603464e-06,
+      "loss": 0.9615,
+      "step": 16086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.915700417372646e-06,
+      "loss": 1.0136,
+      "step": 16087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.91452207235579e-06,
+      "loss": 0.9266,
+      "step": 16088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9133438225749536e-06,
+      "loss": 0.9765,
+      "step": 16089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.912165668052206e-06,
+      "loss": 0.9719,
+      "step": 16090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.910987608809608e-06,
+      "loss": 0.9884,
+      "step": 16091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9098096448692245e-06,
+      "loss": 1.0486,
+      "step": 16092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.908631776253112e-06,
+      "loss": 0.9376,
+      "step": 16093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.907454002983335e-06,
+      "loss": 1.0329,
+      "step": 16094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.906276325081941e-06,
+      "loss": 1.0096,
+      "step": 16095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.905098742570987e-06,
+      "loss": 1.0178,
+      "step": 16096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.903921255472529e-06,
+      "loss": 0.9496,
+      "step": 16097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.9027438638086145e-06,
+      "loss": 0.9795,
+      "step": 16098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.901566567601298e-06,
+      "loss": 0.9723,
+      "step": 16099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.900389366872615e-06,
+      "loss": 1.0007,
+      "step": 16100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.899212261644626e-06,
+      "loss": 0.9782,
+      "step": 16101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.898035251939364e-06,
+      "loss": 0.959,
+      "step": 16102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.896858337778874e-06,
+      "loss": 0.9443,
+      "step": 16103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8956815191851964e-06,
+      "loss": 0.9982,
+      "step": 16104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8945047961803696e-06,
+      "loss": 0.9485,
+      "step": 16105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.893328168786435e-06,
+      "loss": 0.9811,
+      "step": 16106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.892151637025413e-06,
+      "loss": 0.958,
+      "step": 16107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.890975200919355e-06,
+      "loss": 0.9917,
+      "step": 16108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.889798860490279e-06,
+      "loss": 0.9293,
+      "step": 16109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.888622615760217e-06,
+      "loss": 0.9765,
+      "step": 16110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.887446466751201e-06,
+      "loss": 0.9799,
+      "step": 16111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.886270413485254e-06,
+      "loss": 1.0443,
+      "step": 16112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.885094455984403e-06,
+      "loss": 0.9934,
+      "step": 16113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8839185942706615e-06,
+      "loss": 1.019,
+      "step": 16114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.88274282836606e-06,
+      "loss": 1.0369,
+      "step": 16115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.881567158292609e-06,
+      "loss": 0.9496,
+      "step": 16116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.880391584072331e-06,
+      "loss": 0.9972,
+      "step": 16117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.879216105727239e-06,
+      "loss": 1.0079,
+      "step": 16118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.87804072327935e-06,
+      "loss": 1.0184,
+      "step": 16119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.876865436750665e-06,
+      "loss": 0.9397,
+      "step": 16120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.875690246163208e-06,
+      "loss": 0.9659,
+      "step": 16121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.874515151538976e-06,
+      "loss": 0.9657,
+      "step": 16122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8733401528999775e-06,
+      "loss": 0.9583,
+      "step": 16123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8721652502682214e-06,
+      "loss": 0.9637,
+      "step": 16124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8709904436657e-06,
+      "loss": 0.9794,
+      "step": 16125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.869815733114428e-06,
+      "loss": 1.0109,
+      "step": 16126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8686411186363895e-06,
+      "loss": 0.9939,
+      "step": 16127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.867466600253595e-06,
+      "loss": 1.0209,
+      "step": 16128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8662921779880314e-06,
+      "loss": 1.0121,
+      "step": 16129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.865117851861692e-06,
+      "loss": 1.0097,
+      "step": 16130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8639436218965736e-06,
+      "loss": 1.0032,
+      "step": 16131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.862769488114662e-06,
+      "loss": 0.9988,
+      "step": 16132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.86159545053795e-06,
+      "loss": 0.9993,
+      "step": 16133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.860421509188417e-06,
+      "loss": 0.9417,
+      "step": 16134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.859247664088051e-06,
+      "loss": 0.9828,
+      "step": 16135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.858073915258835e-06,
+      "loss": 1.0483,
+      "step": 16136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.856900262722749e-06,
+      "loss": 0.9893,
+      "step": 16137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8557267065017725e-06,
+      "loss": 0.9703,
+      "step": 16138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.854553246617888e-06,
+      "loss": 1.0555,
+      "step": 16139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8533798830930596e-06,
+      "loss": 0.9601,
+      "step": 16140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.852206615949269e-06,
+      "loss": 1.0084,
+      "step": 16141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8510334452084855e-06,
+      "loss": 0.9471,
+      "step": 16142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.849860370892679e-06,
+      "loss": 0.9992,
+      "step": 16143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.848687393023822e-06,
+      "loss": 0.973,
+      "step": 16144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.84751451162387e-06,
+      "loss": 0.9735,
+      "step": 16145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8463417267148034e-06,
+      "loss": 1.016,
+      "step": 16146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8451690383185715e-06,
+      "loss": 0.9621,
+      "step": 16147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.843996446457141e-06,
+      "loss": 0.9383,
+      "step": 16148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.842823951152473e-06,
+      "loss": 0.9277,
+      "step": 16149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.841651552426518e-06,
+      "loss": 1.0178,
+      "step": 16150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.840479250301242e-06,
+      "loss": 0.9428,
+      "step": 16151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.839307044798588e-06,
+      "loss": 1.0135,
+      "step": 16152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.838134935940513e-06,
+      "loss": 1.0556,
+      "step": 16153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.836962923748967e-06,
+      "loss": 1.087,
+      "step": 16154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8357910082459e-06,
+      "loss": 0.9841,
+      "step": 16155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.834619189453255e-06,
+      "loss": 1.0101,
+      "step": 16156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.83344746739298e-06,
+      "loss": 1.0484,
+      "step": 16157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.832275842087012e-06,
+      "loss": 0.9967,
+      "step": 16158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8311043135573015e-06,
+      "loss": 1.0127,
+      "step": 16159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.829932881825781e-06,
+      "loss": 1.0452,
+      "step": 16160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.82876154691439e-06,
+      "loss": 0.964,
+      "step": 16161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.827590308845065e-06,
+      "loss": 0.9981,
+      "step": 16162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.826419167639733e-06,
+      "loss": 1.0241,
+      "step": 16163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8252481233203365e-06,
+      "loss": 1.0015,
+      "step": 16164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8240771759087955e-06,
+      "loss": 1.0056,
+      "step": 16165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8229063254270515e-06,
+      "loss": 0.9573,
+      "step": 16166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.821735571897019e-06,
+      "loss": 0.9736,
+      "step": 16167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.820564915340627e-06,
+      "loss": 0.9434,
+      "step": 16168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.819394355779801e-06,
+      "loss": 1.0088,
+      "step": 16169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.818223893236455e-06,
+      "loss": 0.9942,
+      "step": 16170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.817053527732518e-06,
+      "loss": 0.969,
+      "step": 16171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.815883259289899e-06,
+      "loss": 0.9538,
+      "step": 16172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.814713087930517e-06,
+      "loss": 0.9283,
+      "step": 16173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.813543013676288e-06,
+      "loss": 0.9213,
+      "step": 16174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8123730365491215e-06,
+      "loss": 1.0116,
+      "step": 16175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.811203156570927e-06,
+      "loss": 0.9971,
+      "step": 16176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.8100333737636195e-06,
+      "loss": 0.9462,
+      "step": 16177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.808863688149096e-06,
+      "loss": 0.9874,
+      "step": 16178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.807694099749264e-06,
+      "loss": 0.9585,
+      "step": 16179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.806524608586029e-06,
+      "loss": 0.9563,
+      "step": 16180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.805355214681293e-06,
+      "loss": 0.9742,
+      "step": 16181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.804185918056955e-06,
+      "loss": 0.9509,
+      "step": 16182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.803016718734904e-06,
+      "loss": 0.9133,
+      "step": 16183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.801847616737049e-06,
+      "loss": 0.9768,
+      "step": 16184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.800678612085275e-06,
+      "loss": 1.0149,
+      "step": 16185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7995097048014765e-06,
+      "loss": 0.9643,
+      "step": 16186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7983408949075425e-06,
+      "loss": 1.014,
+      "step": 16187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.797172182425364e-06,
+      "loss": 0.9783,
+      "step": 16188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7960035673768294e-06,
+      "loss": 0.9986,
+      "step": 16189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.794835049783811e-06,
+      "loss": 0.9642,
+      "step": 16190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7936666296682106e-06,
+      "loss": 0.9602,
+      "step": 16191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.792498307051894e-06,
+      "loss": 0.9586,
+      "step": 16192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.791330081956746e-06,
+      "loss": 0.9516,
+      "step": 16193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7901619544046425e-06,
+      "loss": 1.0252,
+      "step": 16194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.788993924417464e-06,
+      "loss": 0.9941,
+      "step": 16195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.787825992017081e-06,
+      "loss": 0.961,
+      "step": 16196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.786658157225364e-06,
+      "loss": 0.9929,
+      "step": 16197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7854904200641814e-06,
+      "loss": 1.0365,
+      "step": 16198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.784322780555407e-06,
+      "loss": 1.0193,
+      "step": 16199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7831552387209014e-06,
+      "loss": 1.0387,
+      "step": 16200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.781987794582536e-06,
+      "loss": 0.975,
+      "step": 16201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.780820448162172e-06,
+      "loss": 0.9376,
+      "step": 16202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.77965319948166e-06,
+      "loss": 0.9472,
+      "step": 16203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.778486048562877e-06,
+      "loss": 1.0013,
+      "step": 16204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.777318995427666e-06,
+      "loss": 1.0104,
+      "step": 16205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.7761520400978875e-06,
+      "loss": 0.9505,
+      "step": 16206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.774985182595398e-06,
+      "loss": 0.9386,
+      "step": 16207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.77381842294204e-06,
+      "loss": 1.0118,
+      "step": 16208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.772651761159676e-06,
+      "loss": 1.0072,
+      "step": 16209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.77148519727014e-06,
+      "loss": 0.9475,
+      "step": 16210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.770318731295293e-06,
+      "loss": 0.9582,
+      "step": 16211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.769152363256971e-06,
+      "loss": 0.9765,
+      "step": 16212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.767986093177018e-06,
+      "loss": 1.0176,
+      "step": 16213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.766819921077272e-06,
+      "loss": 1.0376,
+      "step": 16214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7656538469795825e-06,
+      "loss": 0.949,
+      "step": 16215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7644878709057734e-06,
+      "loss": 0.9601,
+      "step": 16216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.763321992877685e-06,
+      "loss": 1.0191,
+      "step": 16217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.762156212917153e-06,
+      "loss": 0.9574,
+      "step": 16218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.760990531046006e-06,
+      "loss": 0.9174,
+      "step": 16219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7598249472860795e-06,
+      "loss": 0.972,
+      "step": 16220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.75865946165919e-06,
+      "loss": 0.9761,
+      "step": 16221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.757494074187179e-06,
+      "loss": 0.9978,
+      "step": 16222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.756328784891857e-06,
+      "loss": 0.9754,
+      "step": 16223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7551635937950545e-06,
+      "loss": 0.9967,
+      "step": 16224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7539985009185896e-06,
+      "loss": 0.9371,
+      "step": 16225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7528335062842805e-06,
+      "loss": 0.9208,
+      "step": 16226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.751668609913947e-06,
+      "loss": 0.9808,
+      "step": 16227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.750503811829395e-06,
+      "loss": 0.9764,
+      "step": 16228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.749339112052454e-06,
+      "loss": 0.9685,
+      "step": 16229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.74817451060492e-06,
+      "loss": 0.9542,
+      "step": 16230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.747010007508609e-06,
+      "loss": 0.9987,
+      "step": 16231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.745845602785329e-06,
+      "loss": 1.0249,
+      "step": 16232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.744681296456884e-06,
+      "loss": 0.9433,
+      "step": 16233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.743517088545084e-06,
+      "loss": 1.0141,
+      "step": 16234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.742352979071723e-06,
+      "loss": 0.9447,
+      "step": 16235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.741188968058603e-06,
+      "loss": 0.9531,
+      "step": 16236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.740025055527526e-06,
+      "loss": 0.9458,
+      "step": 16237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.738861241500287e-06,
+      "loss": 0.9665,
+      "step": 16238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.73769752599868e-06,
+      "loss": 0.9636,
+      "step": 16239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.736533909044503e-06,
+      "loss": 0.9557,
+      "step": 16240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.735370390659533e-06,
+      "loss": 1.0039,
+      "step": 16241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.734206970865578e-06,
+      "loss": 0.963,
+      "step": 16242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.733043649684414e-06,
+      "loss": 0.9877,
+      "step": 16243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7318804271378285e-06,
+      "loss": 1.0144,
+      "step": 16244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7307173032476085e-06,
+      "loss": 1.0326,
+      "step": 16245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.729554278035526e-06,
+      "loss": 0.902,
+      "step": 16246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.728391351523376e-06,
+      "loss": 0.9665,
+      "step": 16247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7272285237329215e-06,
+      "loss": 0.9405,
+      "step": 16248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.726065794685953e-06,
+      "loss": 0.9379,
+      "step": 16249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.724903164404235e-06,
+      "loss": 0.9907,
+      "step": 16250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.723740632909543e-06,
+      "loss": 0.9793,
+      "step": 16251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.722578200223649e-06,
+      "loss": 0.9721,
+      "step": 16252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.72141586636832e-06,
+      "loss": 0.9863,
+      "step": 16253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.720253631365329e-06,
+      "loss": 0.9501,
+      "step": 16254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.719091495236432e-06,
+      "loss": 0.9242,
+      "step": 16255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.717929458003398e-06,
+      "loss": 0.9908,
+      "step": 16256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.716767519687986e-06,
+      "loss": 0.9822,
+      "step": 16257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.715605680311958e-06,
+      "loss": 0.9957,
+      "step": 16258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.71444393989707e-06,
+      "loss": 1.0107,
+      "step": 16259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.713282298465081e-06,
+      "loss": 0.9706,
+      "step": 16260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.712120756037742e-06,
+      "loss": 1.0099,
+      "step": 16261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.710959312636805e-06,
+      "loss": 0.9694,
+      "step": 16262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.709797968284023e-06,
+      "loss": 0.9748,
+      "step": 16263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.708636723001141e-06,
+      "loss": 0.9056,
+      "step": 16264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.707475576809913e-06,
+      "loss": 0.9617,
+      "step": 16265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.706314529732071e-06,
+      "loss": 0.9513,
+      "step": 16266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.705153581789372e-06,
+      "loss": 0.992,
+      "step": 16267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.703992733003544e-06,
+      "loss": 0.9891,
+      "step": 16268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.702831983396342e-06,
+      "loss": 0.9401,
+      "step": 16269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.70167133298949e-06,
+      "loss": 0.9937,
+      "step": 16270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.7005107818047276e-06,
+      "loss": 1.0059,
+      "step": 16271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6993503298637935e-06,
+      "loss": 0.9722,
+      "step": 16272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.698189977188407e-06,
+      "loss": 0.9511,
+      "step": 16273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.697029723800314e-06,
+      "loss": 1.0326,
+      "step": 16274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.695869569721231e-06,
+      "loss": 1.0032,
+      "step": 16275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.694709514972889e-06,
+      "loss": 1.0138,
+      "step": 16276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.693549559577011e-06,
+      "loss": 0.9265,
+      "step": 16277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.692389703555321e-06,
+      "loss": 0.9758,
+      "step": 16278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.691229946929537e-06,
+      "loss": 0.9492,
+      "step": 16279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.690070289721384e-06,
+      "loss": 0.9743,
+      "step": 16280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6889107319525695e-06,
+      "loss": 0.8871,
+      "step": 16281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.687751273644815e-06,
+      "loss": 1.0168,
+      "step": 16282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.686591914819835e-06,
+      "loss": 0.9811,
+      "step": 16283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.685432655499331e-06,
+      "loss": 0.961,
+      "step": 16284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6842734957050285e-06,
+      "loss": 0.9668,
+      "step": 16285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.683114435458617e-06,
+      "loss": 0.9582,
+      "step": 16286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.681955474781821e-06,
+      "loss": 0.9505,
+      "step": 16287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6807966136963305e-06,
+      "loss": 0.9783,
+      "step": 16288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.679637852223852e-06,
+      "loss": 0.9552,
+      "step": 16289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.678479190386086e-06,
+      "loss": 0.9942,
+      "step": 16290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.677320628204731e-06,
+      "loss": 1.0104,
+      "step": 16291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.676162165701486e-06,
+      "loss": 0.9606,
+      "step": 16292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.675003802898039e-06,
+      "loss": 0.989,
+      "step": 16293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.673845539816088e-06,
+      "loss": 0.9458,
+      "step": 16294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.672687376477321e-06,
+      "loss": 1.0163,
+      "step": 16295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.671529312903428e-06,
+      "loss": 1.0079,
+      "step": 16296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.670371349116097e-06,
+      "loss": 1.0138,
+      "step": 16297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.669213485137017e-06,
+      "loss": 1.0675,
+      "step": 16298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6680557209878615e-06,
+      "loss": 0.9358,
+      "step": 16299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.666898056690319e-06,
+      "loss": 0.9598,
+      "step": 16300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.665740492266066e-06,
+      "loss": 0.9754,
+      "step": 16301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.664583027736784e-06,
+      "loss": 0.9973,
+      "step": 16302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.663425663124149e-06,
+      "loss": 1.0288,
+      "step": 16303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6622683984498275e-06,
+      "loss": 1.0002,
+      "step": 16304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6611112337355025e-06,
+      "loss": 0.9807,
+      "step": 16305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.659954169002832e-06,
+      "loss": 1.0166,
+      "step": 16306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.658797204273498e-06,
+      "loss": 0.9568,
+      "step": 16307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.657640339569159e-06,
+      "loss": 1.0066,
+      "step": 16308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.656483574911478e-06,
+      "loss": 0.9714,
+      "step": 16309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.655326910322126e-06,
+      "loss": 0.9605,
+      "step": 16310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.65417034582275e-06,
+      "loss": 0.9717,
+      "step": 16311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.653013881435027e-06,
+      "loss": 0.9688,
+      "step": 16312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6518575171806e-06,
+      "loss": 1.015,
+      "step": 16313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.65070125308113e-06,
+      "loss": 0.9281,
+      "step": 16314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.649545089158269e-06,
+      "loss": 1.0017,
+      "step": 16315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.648389025433669e-06,
+      "loss": 0.956,
+      "step": 16316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.647233061928979e-06,
+      "loss": 0.9558,
+      "step": 16317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.646077198665853e-06,
+      "loss": 0.8844,
+      "step": 16318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6449214356659285e-06,
+      "loss": 1.0187,
+      "step": 16319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.643765772950849e-06,
+      "loss": 0.9434,
+      "step": 16320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.642610210542264e-06,
+      "loss": 0.9651,
+      "step": 16321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.641454748461807e-06,
+      "loss": 1.0208,
+      "step": 16322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.640299386731126e-06,
+      "loss": 0.9829,
+      "step": 16323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.639144125371841e-06,
+      "loss": 0.9674,
+      "step": 16324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.637988964405605e-06,
+      "loss": 0.9603,
+      "step": 16325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.636833903854039e-06,
+      "loss": 1.0179,
+      "step": 16326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6356789437387764e-06,
+      "loss": 0.9426,
+      "step": 16327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.634524084081447e-06,
+      "loss": 1.0471,
+      "step": 16328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.63336932490368e-06,
+      "loss": 0.9289,
+      "step": 16329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.632214666227101e-06,
+      "loss": 0.9927,
+      "step": 16330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.631060108073322e-06,
+      "loss": 0.9704,
+      "step": 16331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.629905650463982e-06,
+      "loss": 0.9227,
+      "step": 16332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.628751293420688e-06,
+      "loss": 0.9403,
+      "step": 16333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6275970369650625e-06,
+      "loss": 0.9453,
+      "step": 16334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.62644288111872e-06,
+      "loss": 1.0741,
+      "step": 16335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.625288825903275e-06,
+      "loss": 1.0058,
+      "step": 16336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.624134871340343e-06,
+      "loss": 1.0163,
+      "step": 16337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.622981017451527e-06,
+      "loss": 0.9866,
+      "step": 16338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.621827264258439e-06,
+      "loss": 0.9679,
+      "step": 16339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.620673611782686e-06,
+      "loss": 0.9831,
+      "step": 16340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6195200600458705e-06,
+      "loss": 0.9487,
+      "step": 16341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.618366609069598e-06,
+      "loss": 1.0112,
+      "step": 16342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.61721325887547e-06,
+      "loss": 0.968,
+      "step": 16343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6160600094850764e-06,
+      "loss": 1.0153,
+      "step": 16344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.614906860920028e-06,
+      "loss": 0.9227,
+      "step": 16345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.613753813201912e-06,
+      "loss": 0.9992,
+      "step": 16346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.612600866352319e-06,
+      "loss": 1.0004,
+      "step": 16347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.611448020392847e-06,
+      "loss": 0.9759,
+      "step": 16348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.610295275345076e-06,
+      "loss": 0.9657,
+      "step": 16349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.6091426312306065e-06,
+      "loss": 0.9812,
+      "step": 16350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.607990088071014e-06,
+      "loss": 0.9541,
+      "step": 16351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.606837645887885e-06,
+      "loss": 1.0175,
+      "step": 16352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.605685304702802e-06,
+      "loss": 0.9605,
+      "step": 16353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.604533064537344e-06,
+      "loss": 0.9885,
+      "step": 16354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.603380925413094e-06,
+      "loss": 0.9716,
+      "step": 16355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.60222888735162e-06,
+      "loss": 0.9517,
+      "step": 16356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.601076950374502e-06,
+      "loss": 0.9703,
+      "step": 16357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.599925114503309e-06,
+      "loss": 1.004,
+      "step": 16358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.598773379759612e-06,
+      "loss": 0.9488,
+      "step": 16359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.597621746164983e-06,
+      "loss": 0.9787,
+      "step": 16360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.596470213740988e-06,
+      "loss": 0.96,
+      "step": 16361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5953187825091845e-06,
+      "loss": 0.966,
+      "step": 16362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.594167452491148e-06,
+      "loss": 0.99,
+      "step": 16363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.59301622370843e-06,
+      "loss": 0.9973,
+      "step": 16364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.59186509618259e-06,
+      "loss": 1.0054,
+      "step": 16365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.590714069935193e-06,
+      "loss": 1.0119,
+      "step": 16366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.589563144987781e-06,
+      "loss": 0.9065,
+      "step": 16367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.588412321361922e-06,
+      "loss": 0.9334,
+      "step": 16368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.587261599079155e-06,
+      "loss": 1.0535,
+      "step": 16369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.586110978161041e-06,
+      "loss": 0.9783,
+      "step": 16370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.584960458629121e-06,
+      "loss": 1.0399,
+      "step": 16371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.58381004050494e-06,
+      "loss": 1.0148,
+      "step": 16372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.582659723810045e-06,
+      "loss": 1.0087,
+      "step": 16373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.581509508565975e-06,
+      "loss": 0.9986,
+      "step": 16374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.580359394794278e-06,
+      "loss": 0.965,
+      "step": 16375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.579209382516483e-06,
+      "loss": 1.0099,
+      "step": 16376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.578059471754128e-06,
+      "loss": 1.0004,
+      "step": 16377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.57690966252875e-06,
+      "loss": 1.0178,
+      "step": 16378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5757599548618814e-06,
+      "loss": 0.9195,
+      "step": 16379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.574610348775051e-06,
+      "loss": 0.8927,
+      "step": 16380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.573460844289793e-06,
+      "loss": 0.9864,
+      "step": 16381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.572311441427627e-06,
+      "loss": 0.9882,
+      "step": 16382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5711621402100804e-06,
+      "loss": 1.0352,
+      "step": 16383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.570012940658676e-06,
+      "loss": 0.9897,
+      "step": 16384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.568863842794936e-06,
+      "loss": 1.0118,
+      "step": 16385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.567714846640383e-06,
+      "loss": 1.0185,
+      "step": 16386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.566565952216525e-06,
+      "loss": 0.9346,
+      "step": 16387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.565417159544888e-06,
+      "loss": 1.0411,
+      "step": 16388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5642684686469735e-06,
+      "loss": 1.0341,
+      "step": 16389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.563119879544308e-06,
+      "loss": 0.9924,
+      "step": 16390
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.561971392258388e-06,
+      "loss": 0.9971,
+      "step": 16391
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.560823006810729e-06,
+      "loss": 0.9123,
+      "step": 16392
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.559674723222837e-06,
+      "loss": 0.998,
+      "step": 16393
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.558526541516204e-06,
+      "loss": 0.9666,
+      "step": 16394
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.557378461712351e-06,
+      "loss": 0.9642,
+      "step": 16395
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.556230483832767e-06,
+      "loss": 0.946,
+      "step": 16396
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.555082607898948e-06,
+      "loss": 0.9735,
+      "step": 16397
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.553934833932397e-06,
+      "loss": 0.9333,
+      "step": 16398
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.552787161954604e-06,
+      "loss": 0.9326,
+      "step": 16399
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.551639591987063e-06,
+      "loss": 1.015,
+      "step": 16400
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.550492124051269e-06,
+      "loss": 0.9899,
+      "step": 16401
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.549344758168704e-06,
+      "loss": 0.9242,
+      "step": 16402
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.548197494360854e-06,
+      "loss": 0.9591,
+      "step": 16403
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.54705033264921e-06,
+      "loss": 0.994,
+      "step": 16404
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.545903273055251e-06,
+      "loss": 0.9194,
+      "step": 16405
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.544756315600463e-06,
+      "loss": 0.9435,
+      "step": 16406
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.543609460306314e-06,
+      "loss": 0.9898,
+      "step": 16407
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.542462707194296e-06,
+      "loss": 0.9204,
+      "step": 16408
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.541316056285871e-06,
+      "loss": 0.9969,
+      "step": 16409
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.540169507602519e-06,
+      "loss": 0.9639,
+      "step": 16410
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.539023061165712e-06,
+      "loss": 0.9872,
+      "step": 16411
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.537876716996918e-06,
+      "loss": 0.9382,
+      "step": 16412
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.536730475117607e-06,
+      "loss": 1.0396,
+      "step": 16413
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5355843355492375e-06,
+      "loss": 0.9985,
+      "step": 16414
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.534438298313283e-06,
+      "loss": 0.9826,
+      "step": 16415
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.533292363431199e-06,
+      "loss": 0.9622,
+      "step": 16416
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.532146530924448e-06,
+      "loss": 1.0185,
+      "step": 16417
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.531000800814487e-06,
+      "loss": 0.9394,
+      "step": 16418
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.529855173122774e-06,
+      "loss": 0.9954,
+      "step": 16419
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.52870964787076e-06,
+      "loss": 0.9425,
+      "step": 16420
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.527564225079898e-06,
+      "loss": 0.9362,
+      "step": 16421
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.526418904771641e-06,
+      "loss": 0.9535,
+      "step": 16422
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.525273686967435e-06,
+      "loss": 0.9879,
+      "step": 16423
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.524128571688732e-06,
+      "loss": 1.0466,
+      "step": 16424
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.522983558956964e-06,
+      "loss": 0.9359,
+      "step": 16425
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.521838648793588e-06,
+      "loss": 0.9843,
+      "step": 16426
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.520693841220032e-06,
+      "loss": 0.9764,
+      "step": 16427
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.51954913625775e-06,
+      "loss": 0.9731,
+      "step": 16428
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.518404533928166e-06,
+      "loss": 1.0054,
+      "step": 16429
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.51726003425272e-06,
+      "loss": 0.9595,
+      "step": 16430
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.516115637252847e-06,
+      "loss": 0.9926,
+      "step": 16431
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.514971342949969e-06,
+      "loss": 0.9652,
+      "step": 16432
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.513827151365529e-06,
+      "loss": 1.02,
+      "step": 16433
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.512683062520945e-06,
+      "loss": 0.9248,
+      "step": 16434
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.511539076437644e-06,
+      "loss": 1.0184,
+      "step": 16435
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.51039519313705e-06,
+      "loss": 0.9883,
+      "step": 16436
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.509251412640586e-06,
+      "loss": 0.9861,
+      "step": 16437
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5081077349696705e-06,
+      "loss": 1.0375,
+      "step": 16438
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.506964160145723e-06,
+      "loss": 1.0301,
+      "step": 16439
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.505820688190156e-06,
+      "loss": 0.9644,
+      "step": 16440
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5046773191243836e-06,
+      "loss": 0.969,
+      "step": 16441
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.503534052969818e-06,
+      "loss": 0.9319,
+      "step": 16442
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.502390889747872e-06,
+      "loss": 0.9582,
+      "step": 16443
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5012478294799546e-06,
+      "loss": 0.9689,
+      "step": 16444
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.5001048721874625e-06,
+      "loss": 0.9461,
+      "step": 16445
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.498962017891814e-06,
+      "loss": 1.0018,
+      "step": 16446
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.497819266614399e-06,
+      "loss": 0.9917,
+      "step": 16447
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.496676618376624e-06,
+      "loss": 1.0054,
+      "step": 16448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.495534073199886e-06,
+      "loss": 0.9367,
+      "step": 16449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4943916311055825e-06,
+      "loss": 0.9952,
+      "step": 16450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4932492921151105e-06,
+      "loss": 1.012,
+      "step": 16451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.492107056249853e-06,
+      "loss": 1.0215,
+      "step": 16452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.490964923531214e-06,
+      "loss": 0.9234,
+      "step": 16453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.489822893980572e-06,
+      "loss": 0.9615,
+      "step": 16454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.488680967619317e-06,
+      "loss": 0.9969,
+      "step": 16455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.487539144468833e-06,
+      "loss": 0.9275,
+      "step": 16456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.486397424550507e-06,
+      "loss": 0.9698,
+      "step": 16457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.485255807885721e-06,
+      "loss": 1.0248,
+      "step": 16458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.484114294495845e-06,
+      "loss": 0.9859,
+      "step": 16459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.482972884402263e-06,
+      "loss": 0.9633,
+      "step": 16460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.481831577626349e-06,
+      "loss": 1.0003,
+      "step": 16461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.480690374189474e-06,
+      "loss": 0.953,
+      "step": 16462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.479549274113015e-06,
+      "loss": 1.0015,
+      "step": 16463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.478408277418339e-06,
+      "loss": 0.9913,
+      "step": 16464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.477267384126807e-06,
+      "loss": 0.9808,
+      "step": 16465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.476126594259796e-06,
+      "loss": 0.9655,
+      "step": 16466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.474985907838661e-06,
+      "loss": 0.9511,
+      "step": 16467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.473845324884766e-06,
+      "loss": 0.9573,
+      "step": 16468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.472704845419476e-06,
+      "loss": 0.996,
+      "step": 16469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.471564469464136e-06,
+      "loss": 0.995,
+      "step": 16470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.470424197040117e-06,
+      "loss": 0.9461,
+      "step": 16471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.469284028168759e-06,
+      "loss": 0.9489,
+      "step": 16472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4681439628714275e-06,
+      "loss": 0.9517,
+      "step": 16473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.467004001169463e-06,
+      "loss": 0.9167,
+      "step": 16474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.465864143084216e-06,
+      "loss": 0.9491,
+      "step": 16475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.464724388637032e-06,
+      "loss": 0.9668,
+      "step": 16476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.463584737849258e-06,
+      "loss": 1.0112,
+      "step": 16477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.462445190742239e-06,
+      "loss": 1.0331,
+      "step": 16478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.461305747337307e-06,
+      "loss": 0.974,
+      "step": 16479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.460166407655805e-06,
+      "loss": 0.9265,
+      "step": 16480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.459027171719068e-06,
+      "loss": 1.015,
+      "step": 16481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.457888039548433e-06,
+      "loss": 0.9484,
+      "step": 16482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.45674901116523e-06,
+      "loss": 1.0229,
+      "step": 16483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4556100865907945e-06,
+      "loss": 1.0165,
+      "step": 16484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4544712658464485e-06,
+      "loss": 0.967,
+      "step": 16485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.453332548953522e-06,
+      "loss": 0.9202,
+      "step": 16486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.452193935933339e-06,
+      "loss": 0.9564,
+      "step": 16487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.451055426807222e-06,
+      "loss": 0.9541,
+      "step": 16488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.449917021596498e-06,
+      "loss": 0.9252,
+      "step": 16489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.448778720322473e-06,
+      "loss": 0.9935,
+      "step": 16490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.44764052300648e-06,
+      "loss": 0.9789,
+      "step": 16491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.446502429669821e-06,
+      "loss": 0.9741,
+      "step": 16492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.445364440333814e-06,
+      "loss": 0.9944,
+      "step": 16493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.444226555019771e-06,
+      "loss": 1.0837,
+      "step": 16494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.443088773749002e-06,
+      "loss": 0.9349,
+      "step": 16495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.441951096542814e-06,
+      "loss": 0.9515,
+      "step": 16496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.440813523422509e-06,
+      "loss": 0.9801,
+      "step": 16497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.439676054409393e-06,
+      "loss": 0.94,
+      "step": 16498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.438538689524767e-06,
+      "loss": 0.9655,
+      "step": 16499
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.437401428789928e-06,
+      "loss": 1.0087,
+      "step": 16500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.436264272226179e-06,
+      "loss": 1.012,
+      "step": 16501
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.435127219854814e-06,
+      "loss": 0.9629,
+      "step": 16502
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.43399027169712e-06,
+      "loss": 0.9858,
+      "step": 16503
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.432853427774399e-06,
+      "loss": 1.0188,
+      "step": 16504
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.431716688107935e-06,
+      "loss": 1.0371,
+      "step": 16505
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.430580052719015e-06,
+      "loss": 1.0231,
+      "step": 16506
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.429443521628929e-06,
+      "loss": 0.9562,
+      "step": 16507
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.428307094858952e-06,
+      "loss": 0.9158,
+      "step": 16508
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4271707724303795e-06,
+      "loss": 1.0,
+      "step": 16509
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.426034554364477e-06,
+      "loss": 0.9709,
+      "step": 16510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.424898440682536e-06,
+      "loss": 1.003,
+      "step": 16511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.423762431405822e-06,
+      "loss": 0.9682,
+      "step": 16512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4226265265556155e-06,
+      "loss": 1.0289,
+      "step": 16513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.421490726153184e-06,
+      "loss": 1.003,
+      "step": 16514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.4203550302198035e-06,
+      "loss": 0.9758,
+      "step": 16515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.419219438776739e-06,
+      "loss": 0.9455,
+      "step": 16516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.418083951845255e-06,
+      "loss": 0.9064,
+      "step": 16517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.416948569446617e-06,
+      "loss": 0.9802,
+      "step": 16518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.415813291602088e-06,
+      "loss": 0.9842,
+      "step": 16519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.414678118332928e-06,
+      "loss": 0.9644,
+      "step": 16520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.413543049660396e-06,
+      "loss": 1.0352,
+      "step": 16521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.412408085605751e-06,
+      "loss": 0.99,
+      "step": 16522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.411273226190242e-06,
+      "loss": 0.9892,
+      "step": 16523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.410138471435125e-06,
+      "loss": 0.9369,
+      "step": 16524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.409003821361649e-06,
+      "loss": 0.9797,
+      "step": 16525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.407869275991063e-06,
+      "loss": 0.9474,
+      "step": 16526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.40673483534462e-06,
+      "loss": 1.0031,
+      "step": 16527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.40560049944355e-06,
+      "loss": 0.985,
+      "step": 16528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.404466268309114e-06,
+      "loss": 0.9368,
+      "step": 16529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.403332141962533e-06,
+      "loss": 1.0149,
+      "step": 16530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.402198120425067e-06,
+      "loss": 0.9561,
+      "step": 16531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.401064203717936e-06,
+      "loss": 0.8889,
+      "step": 16532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.399930391862383e-06,
+      "loss": 1.0135,
+      "step": 16533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.398796684879643e-06,
+      "loss": 0.9993,
+      "step": 16534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.397663082790935e-06,
+      "loss": 0.914,
+      "step": 16535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.396529585617504e-06,
+      "loss": 0.9375,
+      "step": 16536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.395396193380565e-06,
+      "loss": 0.9704,
+      "step": 16537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.394262906101347e-06,
+      "loss": 0.9547,
+      "step": 16538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.393129723801076e-06,
+      "loss": 0.987,
+      "step": 16539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.391996646500968e-06,
+      "loss": 0.884,
+      "step": 16540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.39086367422225e-06,
+      "loss": 0.9623,
+      "step": 16541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.38973080698613e-06,
+      "loss": 0.9803,
+      "step": 16542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.388598044813828e-06,
+      "loss": 0.9071,
+      "step": 16543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.387465387726555e-06,
+      "loss": 0.9917,
+      "step": 16544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.386332835745526e-06,
+      "loss": 0.9989,
+      "step": 16545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3852003888919465e-06,
+      "loss": 0.9842,
+      "step": 16546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.384068047187031e-06,
+      "loss": 0.9941,
+      "step": 16547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.382935810651973e-06,
+      "loss": 0.9214,
+      "step": 16548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.381803679307988e-06,
+      "loss": 0.9755,
+      "step": 16549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.380671653176269e-06,
+      "loss": 0.9802,
+      "step": 16550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.37953973227802e-06,
+      "loss": 0.977,
+      "step": 16551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.37840791663444e-06,
+      "loss": 0.9686,
+      "step": 16552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.377276206266714e-06,
+      "loss": 0.9782,
+      "step": 16553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.376144601196051e-06,
+      "loss": 0.9819,
+      "step": 16554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.37501310144363e-06,
+      "loss": 1.0051,
+      "step": 16555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.373881707030649e-06,
+      "loss": 1.0651,
+      "step": 16556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.37275041797829e-06,
+      "loss": 0.9119,
+      "step": 16557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.371619234307742e-06,
+      "loss": 0.9885,
+      "step": 16558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.370488156040186e-06,
+      "loss": 0.9822,
+      "step": 16559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.36935718319681e-06,
+      "loss": 0.968,
+      "step": 16560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.368226315798786e-06,
+      "loss": 0.9643,
+      "step": 16561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.367095553867297e-06,
+      "loss": 0.9662,
+      "step": 16562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3659648974235145e-06,
+      "loss": 0.9735,
+      "step": 16563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3648343464886155e-06,
+      "loss": 0.9875,
+      "step": 16564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.363703901083776e-06,
+      "loss": 0.9483,
+      "step": 16565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3625735612301515e-06,
+      "loss": 0.9875,
+      "step": 16566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.36144332694893e-06,
+      "loss": 0.9774,
+      "step": 16567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.360313198261261e-06,
+      "loss": 0.9503,
+      "step": 16568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3591831751883146e-06,
+      "loss": 1.0038,
+      "step": 16569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.358053257751254e-06,
+      "loss": 1.0438,
+      "step": 16570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.356923445971239e-06,
+      "loss": 0.9446,
+      "step": 16571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.355793739869429e-06,
+      "loss": 0.9664,
+      "step": 16572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.35466413946697e-06,
+      "loss": 0.9503,
+      "step": 16573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.353534644785029e-06,
+      "loss": 0.9313,
+      "step": 16574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.352405255844752e-06,
+      "loss": 0.9819,
+      "step": 16575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.351275972667291e-06,
+      "loss": 1.048,
+      "step": 16576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.350146795273792e-06,
+      "loss": 1.0135,
+      "step": 16577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.349017723685402e-06,
+      "loss": 0.9753,
+      "step": 16578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.347888757923268e-06,
+      "loss": 1.0359,
+      "step": 16579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.346759898008528e-06,
+      "loss": 0.9624,
+      "step": 16580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.345631143962323e-06,
+      "loss": 0.9729,
+      "step": 16581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.344502495805793e-06,
+      "loss": 1.0528,
+      "step": 16582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.343373953560073e-06,
+      "loss": 0.8751,
+      "step": 16583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.342245517246297e-06,
+      "loss": 0.996,
+      "step": 16584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.341117186885603e-06,
+      "loss": 0.9744,
+      "step": 16585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.339988962499109e-06,
+      "loss": 0.9517,
+      "step": 16586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.338860844107956e-06,
+      "loss": 1.0159,
+      "step": 16587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.337732831733263e-06,
+      "loss": 0.9858,
+      "step": 16588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.336604925396157e-06,
+      "loss": 0.9648,
+      "step": 16589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.335477125117762e-06,
+      "loss": 1.023,
+      "step": 16590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.334349430919188e-06,
+      "loss": 0.9979,
+      "step": 16591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.33322184282157e-06,
+      "loss": 0.9184,
+      "step": 16592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.332094360846009e-06,
+      "loss": 0.9765,
+      "step": 16593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.330966985013633e-06,
+      "loss": 1.0292,
+      "step": 16594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.329839715345543e-06,
+      "loss": 0.9448,
+      "step": 16595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.328712551862855e-06,
+      "loss": 0.9755,
+      "step": 16596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.327585494586677e-06,
+      "loss": 0.965,
+      "step": 16597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3264585435381145e-06,
+      "loss": 0.9791,
+      "step": 16598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.325331698738276e-06,
+      "loss": 0.9565,
+      "step": 16599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.324204960208257e-06,
+      "loss": 0.9749,
+      "step": 16600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.323078327969162e-06,
+      "loss": 1.0352,
+      "step": 16601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.321951802042088e-06,
+      "loss": 0.9646,
+      "step": 16602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.320825382448134e-06,
+      "loss": 0.9874,
+      "step": 16603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.319699069208391e-06,
+      "loss": 0.9423,
+      "step": 16604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.31857286234396e-06,
+      "loss": 0.9822,
+      "step": 16605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.31744676187592e-06,
+      "loss": 0.9674,
+      "step": 16606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.316320767825364e-06,
+      "loss": 0.9236,
+      "step": 16607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.31519488021338e-06,
+      "loss": 0.9899,
+      "step": 16608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.314069099061052e-06,
+      "loss": 0.9732,
+      "step": 16609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.312943424389464e-06,
+      "loss": 0.9569,
+      "step": 16610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.31181785621969e-06,
+      "loss": 0.9195,
+      "step": 16611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.310692394572819e-06,
+      "loss": 1.0638,
+      "step": 16612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.309567039469917e-06,
+      "loss": 1.0032,
+      "step": 16613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.308441790932069e-06,
+      "loss": 0.9782,
+      "step": 16614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.307316648980339e-06,
+      "loss": 0.9492,
+      "step": 16615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3061916136358e-06,
+      "loss": 0.9897,
+      "step": 16616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.305066684919527e-06,
+      "loss": 0.9758,
+      "step": 16617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.303941862852573e-06,
+      "loss": 1.0039,
+      "step": 16618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.302817147456017e-06,
+      "loss": 0.9216,
+      "step": 16619
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.3016925387509125e-06,
+      "loss": 1.0392,
+      "step": 16620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.300568036758322e-06,
+      "loss": 0.952,
+      "step": 16621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.299443641499307e-06,
+      "loss": 1.0302,
+      "step": 16622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2983193529949234e-06,
+      "loss": 0.9531,
+      "step": 16623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.297195171266219e-06,
+      "loss": 0.9843,
+      "step": 16624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.296071096334258e-06,
+      "loss": 0.9838,
+      "step": 16625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.29494712822008e-06,
+      "loss": 0.9926,
+      "step": 16626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.293823266944741e-06,
+      "loss": 0.977,
+      "step": 16627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.292699512529288e-06,
+      "loss": 0.9975,
+      "step": 16628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.291575864994754e-06,
+      "loss": 0.9837,
+      "step": 16629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.290452324362199e-06,
+      "loss": 0.918,
+      "step": 16630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.289328890652648e-06,
+      "loss": 0.9402,
+      "step": 16631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.288205563887154e-06,
+      "loss": 0.9663,
+      "step": 16632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.287082344086742e-06,
+      "loss": 0.9372,
+      "step": 16633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.285959231272451e-06,
+      "loss": 1.0124,
+      "step": 16634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.284836225465313e-06,
+      "loss": 1.0242,
+      "step": 16635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.283713326686358e-06,
+      "loss": 1.006,
+      "step": 16636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.282590534956622e-06,
+      "loss": 0.9693,
+      "step": 16637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.281467850297118e-06,
+      "loss": 0.9989,
+      "step": 16638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.280345272728879e-06,
+      "loss": 1.0152,
+      "step": 16639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.279222802272927e-06,
+      "loss": 0.9622,
+      "step": 16640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.278100438950281e-06,
+      "loss": 0.9993,
+      "step": 16641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2769781827819605e-06,
+      "loss": 0.9841,
+      "step": 16642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.275856033788986e-06,
+      "loss": 1.0536,
+      "step": 16643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.274733991992365e-06,
+      "loss": 0.93,
+      "step": 16644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.273612057413113e-06,
+      "loss": 1.0053,
+      "step": 16645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.272490230072238e-06,
+      "loss": 1.0428,
+      "step": 16646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.271368509990755e-06,
+      "loss": 0.9438,
+      "step": 16647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.270246897189667e-06,
+      "loss": 1.0039,
+      "step": 16648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.269125391689972e-06,
+      "loss": 0.8944,
+      "step": 16649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.268003993512686e-06,
+      "loss": 0.9761,
+      "step": 16650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.266882702678794e-06,
+      "loss": 0.9998,
+      "step": 16651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.265761519209311e-06,
+      "loss": 0.9436,
+      "step": 16652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.264640443125221e-06,
+      "loss": 0.9449,
+      "step": 16653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.263519474447522e-06,
+      "loss": 1.0435,
+      "step": 16654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.262398613197211e-06,
+      "loss": 0.9818,
+      "step": 16655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.261277859395267e-06,
+      "loss": 0.933,
+      "step": 16656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.260157213062692e-06,
+      "loss": 0.9782,
+      "step": 16657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.259036674220463e-06,
+      "loss": 0.9844,
+      "step": 16658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.257916242889568e-06,
+      "loss": 0.8904,
+      "step": 16659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.256795919090989e-06,
+      "loss": 1.0291,
+      "step": 16660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.255675702845707e-06,
+      "loss": 1.0423,
+      "step": 16661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.254555594174698e-06,
+      "loss": 1.0618,
+      "step": 16662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.253435593098945e-06,
+      "loss": 0.9924,
+      "step": 16663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.252315699639413e-06,
+      "loss": 1.0012,
+      "step": 16664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.251195913817079e-06,
+      "loss": 0.9808,
+      "step": 16665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.250076235652912e-06,
+      "loss": 0.9316,
+      "step": 16666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.248956665167881e-06,
+      "loss": 0.9648,
+      "step": 16667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.247837202382957e-06,
+      "loss": 0.9935,
+      "step": 16668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2467178473190924e-06,
+      "loss": 1.0478,
+      "step": 16669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.245598599997263e-06,
+      "loss": 0.9993,
+      "step": 16670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.244479460438421e-06,
+      "loss": 0.994,
+      "step": 16671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.243360428663526e-06,
+      "loss": 0.9286,
+      "step": 16672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.242241504693533e-06,
+      "loss": 0.9653,
+      "step": 16673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.241122688549399e-06,
+      "loss": 0.9689,
+      "step": 16674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.240003980252076e-06,
+      "loss": 0.9916,
+      "step": 16675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.238885379822507e-06,
+      "loss": 0.9869,
+      "step": 16676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.237766887281654e-06,
+      "loss": 0.9859,
+      "step": 16677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.236648502650451e-06,
+      "loss": 1.0029,
+      "step": 16678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2355302259498455e-06,
+      "loss": 0.9053,
+      "step": 16679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.234412057200779e-06,
+      "loss": 0.9655,
+      "step": 16680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.233293996424195e-06,
+      "loss": 0.9667,
+      "step": 16681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.232176043641032e-06,
+      "loss": 1.0155,
+      "step": 16682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.231058198872218e-06,
+      "loss": 1.0477,
+      "step": 16683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.229940462138695e-06,
+      "loss": 0.9467,
+      "step": 16684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2288228334613895e-06,
+      "loss": 1.0166,
+      "step": 16685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.227705312861233e-06,
+      "loss": 1.0327,
+      "step": 16686
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.226587900359155e-06,
+      "loss": 0.9603,
+      "step": 16687
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.225470595976086e-06,
+      "loss": 0.9415,
+      "step": 16688
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.224353399732936e-06,
+      "loss": 0.9834,
+      "step": 16689
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.223236311650642e-06,
+      "loss": 1.0204,
+      "step": 16690
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.222119331750113e-06,
+      "loss": 0.9492,
+      "step": 16691
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.22100246005227e-06,
+      "loss": 0.8789,
+      "step": 16692
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.219885696578034e-06,
+      "loss": 0.9511,
+      "step": 16693
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.218769041348307e-06,
+      "loss": 1.0036,
+      "step": 16694
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.217652494384014e-06,
+      "loss": 0.9467,
+      "step": 16695
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.216536055706056e-06,
+      "loss": 0.9799,
+      "step": 16696
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.215419725335343e-06,
+      "loss": 0.9601,
+      "step": 16697
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.21430350329278e-06,
+      "loss": 0.943,
+      "step": 16698
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.21318738959927e-06,
+      "loss": 0.9821,
+      "step": 16699
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.212071384275716e-06,
+      "loss": 0.973,
+      "step": 16700
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.210955487343022e-06,
+      "loss": 0.9312,
+      "step": 16701
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.209839698822076e-06,
+      "loss": 0.9193,
+      "step": 16702
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.208724018733778e-06,
+      "loss": 0.8971,
+      "step": 16703
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.207608447099021e-06,
+      "loss": 0.967,
+      "step": 16704
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.206492983938698e-06,
+      "loss": 0.9738,
+      "step": 16705
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.205377629273701e-06,
+      "loss": 0.9969,
+      "step": 16706
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.204262383124905e-06,
+      "loss": 1.0228,
+      "step": 16707
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.20314724551321e-06,
+      "loss": 0.9911,
+      "step": 16708
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.202032216459492e-06,
+      "loss": 0.9783,
+      "step": 16709
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.200917295984632e-06,
+      "loss": 0.9907,
+      "step": 16710
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1998024841095125e-06,
+      "loss": 0.9231,
+      "step": 16711
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.198687780855002e-06,
+      "loss": 1.0179,
+      "step": 16712
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.19757318624199e-06,
+      "loss": 1.0047,
+      "step": 16713
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.196458700291334e-06,
+      "loss": 0.8789,
+      "step": 16714
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.195344323023919e-06,
+      "loss": 0.9939,
+      "step": 16715
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.194230054460606e-06,
+      "loss": 1.041,
+      "step": 16716
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.193115894622263e-06,
+      "loss": 0.9116,
+      "step": 16717
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.192001843529755e-06,
+      "loss": 0.9532,
+      "step": 16718
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.190887901203945e-06,
+      "loss": 1.0262,
+      "step": 16719
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1897740676657e-06,
+      "loss": 0.9738,
+      "step": 16720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.188660342935867e-06,
+      "loss": 0.9367,
+      "step": 16721
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.18754672703531e-06,
+      "loss": 1.0236,
+      "step": 16722
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1864332199848826e-06,
+      "loss": 1.0057,
+      "step": 16723
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.185319821805438e-06,
+      "loss": 0.8913,
+      "step": 16724
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.184206532517827e-06,
+      "loss": 0.9817,
+      "step": 16725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1830933521429e-06,
+      "loss": 1.014,
+      "step": 16726
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.181980280701499e-06,
+      "loss": 0.979,
+      "step": 16727
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.180867318214469e-06,
+      "loss": 0.9646,
+      "step": 16728
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.179754464702657e-06,
+      "loss": 0.9424,
+      "step": 16729
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1786417201869e-06,
+      "loss": 1.0075,
+      "step": 16730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.177529084688039e-06,
+      "loss": 0.9173,
+      "step": 16731
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.176416558226902e-06,
+      "loss": 1.0464,
+      "step": 16732
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.175304140824338e-06,
+      "loss": 0.926,
+      "step": 16733
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.174191832501165e-06,
+      "loss": 0.9406,
+      "step": 16734
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1730796332782244e-06,
+      "loss": 0.9553,
+      "step": 16735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.171967543176338e-06,
+      "loss": 0.9786,
+      "step": 16736
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.170855562216332e-06,
+      "loss": 0.9378,
+      "step": 16737
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.169743690419036e-06,
+      "loss": 0.9296,
+      "step": 16738
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.168631927805261e-06,
+      "loss": 0.985,
+      "step": 16739
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.167520274395842e-06,
+      "loss": 0.9685,
+      "step": 16740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1664087302115855e-06,
+      "loss": 0.942,
+      "step": 16741
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1652972952733105e-06,
+      "loss": 0.9896,
+      "step": 16742
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.164185969601832e-06,
+      "loss": 0.8638,
+      "step": 16743
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.163074753217961e-06,
+      "loss": 0.9702,
+      "step": 16744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1619636461425075e-06,
+      "loss": 0.9566,
+      "step": 16745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1608526483962834e-06,
+      "loss": 0.9936,
+      "step": 16746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.159741760000087e-06,
+      "loss": 0.8859,
+      "step": 16747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.158630980974726e-06,
+      "loss": 1.0055,
+      "step": 16748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.157520311340998e-06,
+      "loss": 1.0149,
+      "step": 16749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.156409751119709e-06,
+      "loss": 0.9797,
+      "step": 16750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.155299300331658e-06,
+      "loss": 0.9313,
+      "step": 16751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.154188958997628e-06,
+      "loss": 0.9928,
+      "step": 16752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.153078727138427e-06,
+      "loss": 1.0132,
+      "step": 16753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.151968604774837e-06,
+      "loss": 0.9673,
+      "step": 16754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.150858591927649e-06,
+      "loss": 0.9988,
+      "step": 16755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1497486886176515e-06,
+      "loss": 0.9808,
+      "step": 16756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.148638894865631e-06,
+      "loss": 0.9481,
+      "step": 16757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1475292106923705e-06,
+      "loss": 0.9226,
+      "step": 16758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1464196361186445e-06,
+      "loss": 0.9081,
+      "step": 16759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.145310171165246e-06,
+      "loss": 0.9829,
+      "step": 16760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.144200815852937e-06,
+      "loss": 1.029,
+      "step": 16761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.143091570202501e-06,
+      "loss": 0.9534,
+      "step": 16762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.141982434234709e-06,
+      "loss": 0.9908,
+      "step": 16763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.140873407970333e-06,
+      "loss": 0.9978,
+      "step": 16764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.139764491430139e-06,
+      "loss": 0.9565,
+      "step": 16765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.138655684634893e-06,
+      "loss": 1.0159,
+      "step": 16766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.137546987605364e-06,
+      "loss": 0.9808,
+      "step": 16767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.136438400362313e-06,
+      "loss": 0.9852,
+      "step": 16768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.135329922926503e-06,
+      "loss": 1.0357,
+      "step": 16769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.13422155531868e-06,
+      "loss": 0.9297,
+      "step": 16770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1331132975596195e-06,
+      "loss": 0.9751,
+      "step": 16771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.13200514967006e-06,
+      "loss": 0.9782,
+      "step": 16772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1308971116707665e-06,
+      "loss": 0.9521,
+      "step": 16773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.12978918358248e-06,
+      "loss": 0.9291,
+      "step": 16774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.128681365425952e-06,
+      "loss": 0.9547,
+      "step": 16775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.127573657221932e-06,
+      "loss": 0.975,
+      "step": 16776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.126466058991152e-06,
+      "loss": 0.9819,
+      "step": 16777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.12535857075437e-06,
+      "loss": 0.974,
+      "step": 16778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.124251192532316e-06,
+      "loss": 0.9818,
+      "step": 16779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.123143924345732e-06,
+      "loss": 1.0239,
+      "step": 16780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.122036766215352e-06,
+      "loss": 0.9213,
+      "step": 16781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.120929718161909e-06,
+      "loss": 0.9869,
+      "step": 16782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.119822780206136e-06,
+      "loss": 0.9784,
+      "step": 16783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1187159523687675e-06,
+      "loss": 0.9522,
+      "step": 16784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.117609234670522e-06,
+      "loss": 0.9555,
+      "step": 16785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.116502627132129e-06,
+      "loss": 0.9371,
+      "step": 16786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.115396129774314e-06,
+      "loss": 0.9752,
+      "step": 16787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1142897426177965e-06,
+      "loss": 0.9187,
+      "step": 16788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1131834656833e-06,
+      "loss": 0.994,
+      "step": 16789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.11207729899153e-06,
+      "loss": 0.9325,
+      "step": 16790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.110971242563216e-06,
+      "loss": 0.9795,
+      "step": 16791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.109865296419063e-06,
+      "loss": 0.9496,
+      "step": 16792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.108759460579784e-06,
+      "loss": 0.9583,
+      "step": 16793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.107653735066087e-06,
+      "loss": 1.0098,
+      "step": 16794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.10654811989868e-06,
+      "loss": 0.9255,
+      "step": 16795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.105442615098273e-06,
+      "loss": 0.9636,
+      "step": 16796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1043372206855555e-06,
+      "loss": 0.9058,
+      "step": 16797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.103231936681245e-06,
+      "loss": 0.9604,
+      "step": 16798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.1021267631060275e-06,
+      "loss": 1.0242,
+      "step": 16799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.101021699980605e-06,
+      "loss": 0.9876,
+      "step": 16800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0999167473256695e-06,
+      "loss": 0.9833,
+      "step": 16801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0988119051619165e-06,
+      "loss": 0.9965,
+      "step": 16802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.097707173510039e-06,
+      "loss": 0.9448,
+      "step": 16803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.096602552390716e-06,
+      "loss": 0.9891,
+      "step": 16804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.095498041824641e-06,
+      "loss": 0.958,
+      "step": 16805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.094393641832495e-06,
+      "loss": 0.98,
+      "step": 16806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.093289352434961e-06,
+      "loss": 0.9312,
+      "step": 16807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.092185173652721e-06,
+      "loss": 0.9307,
+      "step": 16808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.091081105506456e-06,
+      "loss": 0.9296,
+      "step": 16809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.089977148016829e-06,
+      "loss": 0.9207,
+      "step": 16810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.088873301204531e-06,
+      "loss": 0.9601,
+      "step": 16811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.087769565090221e-06,
+      "loss": 1.0151,
+      "step": 16812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.086665939694572e-06,
+      "loss": 0.9321,
+      "step": 16813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.085562425038259e-06,
+      "loss": 1.0141,
+      "step": 16814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0844590211419325e-06,
+      "loss": 0.9479,
+      "step": 16815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.083355728026272e-06,
+      "loss": 1.0022,
+      "step": 16816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.082252545711925e-06,
+      "loss": 0.8896,
+      "step": 16817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0811494742195634e-06,
+      "loss": 0.9594,
+      "step": 16818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.080046513569837e-06,
+      "loss": 0.9324,
+      "step": 16819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.078943663783401e-06,
+      "loss": 0.9465,
+      "step": 16820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.07784092488091e-06,
+      "loss": 0.9853,
+      "step": 16821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.076738296883016e-06,
+      "loss": 0.933,
+      "step": 16822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.075635779810372e-06,
+      "loss": 0.9976,
+      "step": 16823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.074533373683614e-06,
+      "loss": 1.0017,
+      "step": 16824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.073431078523395e-06,
+      "loss": 0.926,
+      "step": 16825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.072328894350355e-06,
+      "loss": 1.0113,
+      "step": 16826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.071226821185135e-06,
+      "loss": 0.9851,
+      "step": 16827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.070124859048374e-06,
+      "loss": 0.8996,
+      "step": 16828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.069023007960712e-06,
+      "loss": 0.9466,
+      "step": 16829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.067921267942777e-06,
+      "loss": 1.0052,
+      "step": 16830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.066819639015204e-06,
+      "loss": 0.9613,
+      "step": 16831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0657181211986235e-06,
+      "loss": 0.9639,
+      "step": 16832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0646167145136634e-06,
+      "loss": 0.9411,
+      "step": 16833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0635154189809525e-06,
+      "loss": 0.9609,
+      "step": 16834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.062414234621108e-06,
+      "loss": 0.9842,
+      "step": 16835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.061313161454761e-06,
+      "loss": 1.0241,
+      "step": 16836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.060212199502523e-06,
+      "loss": 0.9785,
+      "step": 16837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.059111348785016e-06,
+      "loss": 0.9624,
+      "step": 16838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0580106093228535e-06,
+      "loss": 1.0121,
+      "step": 16839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.056909981136653e-06,
+      "loss": 0.9117,
+      "step": 16840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.055809464247027e-06,
+      "loss": 0.9845,
+      "step": 16841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.054709058674578e-06,
+      "loss": 0.9568,
+      "step": 16842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.053608764439916e-06,
+      "loss": 0.8677,
+      "step": 16843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.052508581563647e-06,
+      "loss": 0.9399,
+      "step": 16844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.051408510066374e-06,
+      "loss": 0.9459,
+      "step": 16845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0503085499687e-06,
+      "loss": 0.9324,
+      "step": 16846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.049208701291227e-06,
+      "loss": 1.01,
+      "step": 16847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0481089640545376e-06,
+      "loss": 0.883,
+      "step": 16848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.047009338279244e-06,
+      "loss": 0.9828,
+      "step": 16849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.04590982398593e-06,
+      "loss": 0.9665,
+      "step": 16850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.044810421195186e-06,
+      "loss": 0.9704,
+      "step": 16851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.043711129927606e-06,
+      "loss": 1.0297,
+      "step": 16852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.042611950203766e-06,
+      "loss": 1.0405,
+      "step": 16853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0415128820442655e-06,
+      "loss": 0.979,
+      "step": 16854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.04041392546967e-06,
+      "loss": 1.0237,
+      "step": 16855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.039315080500574e-06,
+      "loss": 0.98,
+      "step": 16856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.038216347157549e-06,
+      "loss": 0.9461,
+      "step": 16857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.03711772546117e-06,
+      "loss": 1.0215,
+      "step": 16858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.036019215432013e-06,
+      "loss": 0.9221,
+      "step": 16859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.034920817090651e-06,
+      "loss": 0.9636,
+      "step": 16860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.033822530457654e-06,
+      "loss": 0.8943,
+      "step": 16861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.032724355553585e-06,
+      "loss": 1.0149,
+      "step": 16862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.031626292399015e-06,
+      "loss": 1.0003,
+      "step": 16863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0305283410145015e-06,
+      "loss": 1.0498,
+      "step": 16864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.029430501420612e-06,
+      "loss": 1.0156,
+      "step": 16865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.028332773637904e-06,
+      "loss": 0.9446,
+      "step": 16866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.027235157686936e-06,
+      "loss": 1.0352,
+      "step": 16867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.026137653588258e-06,
+      "loss": 0.9393,
+      "step": 16868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.025040261362428e-06,
+      "loss": 0.9475,
+      "step": 16869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.023942981029992e-06,
+      "loss": 0.9194,
+      "step": 16870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.022845812611504e-06,
+      "loss": 0.966,
+      "step": 16871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.021748756127511e-06,
+      "loss": 0.9641,
+      "step": 16872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0206518115985495e-06,
+      "loss": 0.9809,
+      "step": 16873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.019554979045175e-06,
+      "loss": 1.0168,
+      "step": 16874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.018458258487913e-06,
+      "loss": 0.9911,
+      "step": 16875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.017361649947316e-06,
+      "loss": 0.9799,
+      "step": 16876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.016265153443913e-06,
+      "loss": 0.9349,
+      "step": 16877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.015168768998237e-06,
+      "loss": 0.9604,
+      "step": 16878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.014072496630827e-06,
+      "loss": 0.9451,
+      "step": 16879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0129763363621985e-06,
+      "loss": 1.0244,
+      "step": 16880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.011880288212899e-06,
+      "loss": 0.9682,
+      "step": 16881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.010784352203438e-06,
+      "loss": 0.9935,
+      "step": 16882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.009688528354348e-06,
+      "loss": 0.9751,
+      "step": 16883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0085928166861455e-06,
+      "loss": 0.9886,
+      "step": 16884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0074972172193545e-06,
+      "loss": 0.9059,
+      "step": 16885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.006401729974488e-06,
+      "loss": 0.9821,
+      "step": 16886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.0053063549720684e-06,
+      "loss": 1.0116,
+      "step": 16887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.004211092232599e-06,
+      "loss": 0.9716,
+      "step": 16888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.003115941776597e-06,
+      "loss": 0.9572,
+      "step": 16889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.002020903624571e-06,
+      "loss": 0.9708,
+      "step": 16890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.000925977797025e-06,
+      "loss": 0.9077,
+      "step": 16891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.999831164314469e-06,
+      "loss": 0.9163,
+      "step": 16892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9987364631973965e-06,
+      "loss": 0.9769,
+      "step": 16893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9976418744663205e-06,
+      "loss": 0.9154,
+      "step": 16894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.996547398141729e-06,
+      "loss": 1.0023,
+      "step": 16895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.995453034244123e-06,
+      "loss": 0.9658,
+      "step": 16896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.994358782793997e-06,
+      "loss": 1.0242,
+      "step": 16897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9932646438118375e-06,
+      "loss": 1.0169,
+      "step": 16898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9921706173181435e-06,
+      "loss": 1.014,
+      "step": 16899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.991076703333396e-06,
+      "loss": 0.9963,
+      "step": 16900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.989982901878083e-06,
+      "loss": 0.9839,
+      "step": 16901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.988889212972688e-06,
+      "loss": 0.9895,
+      "step": 16902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.987795636637693e-06,
+      "loss": 0.9768,
+      "step": 16903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.986702172893579e-06,
+      "loss": 0.9245,
+      "step": 16904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.985608821760822e-06,
+      "loss": 0.9664,
+      "step": 16905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9845155832598955e-06,
+      "loss": 0.9556,
+      "step": 16906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.983422457411274e-06,
+      "loss": 1.0145,
+      "step": 16907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.982329444235426e-06,
+      "loss": 0.9756,
+      "step": 16908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.981236543752826e-06,
+      "loss": 0.9625,
+      "step": 16909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.980143755983939e-06,
+      "loss": 1.004,
+      "step": 16910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.979051080949224e-06,
+      "loss": 0.9405,
+      "step": 16911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9779585186691525e-06,
+      "loss": 0.9022,
+      "step": 16912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.976866069164178e-06,
+      "loss": 0.9892,
+      "step": 16913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.975773732454762e-06,
+      "loss": 1.0135,
+      "step": 16914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.974681508561358e-06,
+      "loss": 1.0588,
+      "step": 16915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.973589397504425e-06,
+      "loss": 1.0597,
+      "step": 16916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.972497399304414e-06,
+      "loss": 0.9767,
+      "step": 16917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.971405513981768e-06,
+      "loss": 1.0312,
+      "step": 16918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.970313741556945e-06,
+      "loss": 0.9887,
+      "step": 16919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.969222082050383e-06,
+      "loss": 1.0045,
+      "step": 16920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.968130535482528e-06,
+      "loss": 0.921,
+      "step": 16921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.967039101873822e-06,
+      "loss": 0.9268,
+      "step": 16922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.965947781244703e-06,
+      "loss": 1.0183,
+      "step": 16923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9648565736156136e-06,
+      "loss": 0.9201,
+      "step": 16924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.963765479006981e-06,
+      "loss": 0.9406,
+      "step": 16925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.962674497439243e-06,
+      "loss": 1.0363,
+      "step": 16926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9615836289328265e-06,
+      "loss": 0.9567,
+      "step": 16927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9604928735081645e-06,
+      "loss": 0.928,
+      "step": 16928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9594022311856825e-06,
+      "loss": 1.0126,
+      "step": 16929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.958311701985808e-06,
+      "loss": 0.9935,
+      "step": 16930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.957221285928953e-06,
+      "loss": 0.9543,
+      "step": 16931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9561309830355515e-06,
+      "loss": 1.0003,
+      "step": 16932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.955040793326011e-06,
+      "loss": 0.9418,
+      "step": 16933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.953950716820752e-06,
+      "loss": 0.9708,
+      "step": 16934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.952860753540192e-06,
+      "loss": 0.9613,
+      "step": 16935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9517709035047295e-06,
+      "loss": 0.9982,
+      "step": 16936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.950681166734794e-06,
+      "loss": 1.0397,
+      "step": 16937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.949591543250773e-06,
+      "loss": 0.9109,
+      "step": 16938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.948502033073089e-06,
+      "loss": 1.0071,
+      "step": 16939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.947412636222133e-06,
+      "loss": 1.025,
+      "step": 16940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.946323352718311e-06,
+      "loss": 0.9832,
+      "step": 16941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.945234182582024e-06,
+      "loss": 0.9432,
+      "step": 16942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.944145125833666e-06,
+      "loss": 0.966,
+      "step": 16943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.943056182493637e-06,
+      "loss": 1.0086,
+      "step": 16944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.941967352582321e-06,
+      "loss": 0.9733,
+      "step": 16945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.940878636120113e-06,
+      "loss": 0.9754,
+      "step": 16946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.939790033127402e-06,
+      "loss": 0.9397,
+      "step": 16947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.938701543624574e-06,
+      "loss": 0.9692,
+      "step": 16948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.937613167632011e-06,
+      "loss": 0.9715,
+      "step": 16949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.936524905170103e-06,
+      "loss": 1.0134,
+      "step": 16950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.935436756259219e-06,
+      "loss": 0.9988,
+      "step": 16951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.934348720919743e-06,
+      "loss": 0.9443,
+      "step": 16952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.933260799172049e-06,
+      "loss": 1.0018,
+      "step": 16953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.932172991036511e-06,
+      "loss": 0.9707,
+      "step": 16954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.931085296533504e-06,
+      "loss": 1.0038,
+      "step": 16955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.929997715683388e-06,
+      "loss": 0.9745,
+      "step": 16956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.928910248506542e-06,
+      "loss": 0.9881,
+      "step": 16957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.927822895023318e-06,
+      "loss": 0.9361,
+      "step": 16958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9267356552540944e-06,
+      "loss": 1.0063,
+      "step": 16959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.925648529219219e-06,
+      "loss": 0.902,
+      "step": 16960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.924561516939055e-06,
+      "loss": 0.9862,
+      "step": 16961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.923474618433965e-06,
+      "loss": 0.9219,
+      "step": 16962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9223878337242865e-06,
+      "loss": 1.0129,
+      "step": 16963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.921301162830393e-06,
+      "loss": 0.9811,
+      "step": 16964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.92021460577262e-06,
+      "loss": 0.9734,
+      "step": 16965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.919128162571319e-06,
+      "loss": 1.0079,
+      "step": 16966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.918041833246838e-06,
+      "loss": 1.0281,
+      "step": 16967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.916955617819523e-06,
+      "loss": 0.9298,
+      "step": 16968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.915869516309704e-06,
+      "loss": 0.9178,
+      "step": 16969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.914783528737735e-06,
+      "loss": 0.9575,
+      "step": 16970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913697655123946e-06,
+      "loss": 0.9852,
+      "step": 16971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9126118954886715e-06,
+      "loss": 0.8954,
+      "step": 16972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.911526249852251e-06,
+      "loss": 0.9123,
+      "step": 16973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.910440718235002e-06,
+      "loss": 0.9628,
+      "step": 16974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.90935530065727e-06,
+      "loss": 1.0477,
+      "step": 16975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.908269997139367e-06,
+      "loss": 1.0216,
+      "step": 16976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.907184807701631e-06,
+      "loss": 0.953,
+      "step": 16977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9060997323643745e-06,
+      "loss": 0.9853,
+      "step": 16978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.90501477114792e-06,
+      "loss": 0.9396,
+      "step": 16979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.903929924072588e-06,
+      "loss": 0.9253,
+      "step": 16980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.902845191158692e-06,
+      "loss": 0.9634,
+      "step": 16981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.90176057242655e-06,
+      "loss": 1.0858,
+      "step": 16982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.900676067896469e-06,
+      "loss": 0.9707,
+      "step": 16983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.899591677588757e-06,
+      "loss": 1.0163,
+      "step": 16984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.898507401523728e-06,
+      "loss": 1.0263,
+      "step": 16985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.897423239721681e-06,
+      "loss": 1.0367,
+      "step": 16986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.896339192202924e-06,
+      "loss": 1.0252,
+      "step": 16987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.895255258987758e-06,
+      "loss": 1.0108,
+      "step": 16988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.894171440096479e-06,
+      "loss": 0.9935,
+      "step": 16989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8930877355493835e-06,
+      "loss": 0.9543,
+      "step": 16990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.892004145366766e-06,
+      "loss": 1.0201,
+      "step": 16991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.890920669568921e-06,
+      "loss": 0.9652,
+      "step": 16992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.889837308176143e-06,
+      "loss": 0.9458,
+      "step": 16993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.888754061208708e-06,
+      "loss": 0.9249,
+      "step": 16994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.887670928686914e-06,
+      "loss": 1.0286,
+      "step": 16995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.886587910631034e-06,
+      "loss": 0.9375,
+      "step": 16996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.885505007061365e-06,
+      "loss": 0.9911,
+      "step": 16997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.884422217998172e-06,
+      "loss": 0.9419,
+      "step": 16998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.883339543461738e-06,
+      "loss": 1.0036,
+      "step": 16999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.882256983472343e-06,
+      "loss": 0.9104,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "eval_text_loss": 0.9338606595993042,
+      "eval_text_runtime": 1.56,
+      "eval_text_samples_per_second": 47.437,
+      "eval_text_steps_per_second": 1.282,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "eval_svit_loss": 1.0850073099136353,
+      "eval_svit_runtime": 4.0854,
+      "eval_svit_samples_per_second": 166.69,
+      "eval_svit_steps_per_second": 2.693,
+      "step": 17000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8811745380502454e-06,
+      "loss": 1.0106,
+      "step": 17001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.880092207215736e-06,
+      "loss": 0.9442,
+      "step": 17002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8790099909890685e-06,
+      "loss": 0.9637,
+      "step": 17003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.877927889390518e-06,
+      "loss": 0.9431,
+      "step": 17004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.876845902440343e-06,
+      "loss": 1.0526,
+      "step": 17005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.87576403015881e-06,
+      "loss": 0.9782,
+      "step": 17006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.874682272566179e-06,
+      "loss": 0.9904,
+      "step": 17007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.873600629682708e-06,
+      "loss": 0.9861,
+      "step": 17008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8725191015286526e-06,
+      "loss": 0.9373,
+      "step": 17009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8714376881242645e-06,
+      "loss": 0.9633,
+      "step": 17010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.870356389489798e-06,
+      "loss": 0.9948,
+      "step": 17011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.869275205645501e-06,
+      "loss": 0.9898,
+      "step": 17012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.868194136611626e-06,
+      "loss": 1.0067,
+      "step": 17013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.867113182408409e-06,
+      "loss": 0.9523,
+      "step": 17014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.866032343056101e-06,
+      "loss": 0.9759,
+      "step": 17015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.864951618574941e-06,
+      "loss": 1.0019,
+      "step": 17016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8638710089851634e-06,
+      "loss": 0.9899,
+      "step": 17017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8627905143070114e-06,
+      "loss": 0.9556,
+      "step": 17018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.861710134560714e-06,
+      "loss": 0.9735,
+      "step": 17019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.860629869766511e-06,
+      "loss": 0.9976,
+      "step": 17020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.859549719944618e-06,
+      "loss": 0.933,
+      "step": 17021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.858469685115283e-06,
+      "loss": 0.9642,
+      "step": 17022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8573897652987155e-06,
+      "loss": 0.997,
+      "step": 17023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.856309960515145e-06,
+      "loss": 0.9789,
+      "step": 17024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.855230270784795e-06,
+      "loss": 1.0241,
+      "step": 17025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.854150696127882e-06,
+      "loss": 1.0315,
+      "step": 17026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.853071236564627e-06,
+      "loss": 0.9814,
+      "step": 17027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.85199189211524e-06,
+      "loss": 0.9881,
+      "step": 17028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.850912662799937e-06,
+      "loss": 0.9813,
+      "step": 17029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.849833548638927e-06,
+      "loss": 0.9516,
+      "step": 17030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.848754549652422e-06,
+      "loss": 0.9599,
+      "step": 17031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8476756658606245e-06,
+      "loss": 0.9922,
+      "step": 17032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.846596897283743e-06,
+      "loss": 0.968,
+      "step": 17033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.845518243941971e-06,
+      "loss": 0.9087,
+      "step": 17034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.844439705855523e-06,
+      "loss": 0.9756,
+      "step": 17035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8433612830445845e-06,
+      "loss": 0.9506,
+      "step": 17036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8422829755293555e-06,
+      "loss": 1.0304,
+      "step": 17037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.841204783330032e-06,
+      "loss": 0.9533,
+      "step": 17038
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.840126706466796e-06,
+      "loss": 1.0038,
+      "step": 17039
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.839048744959852e-06,
+      "loss": 0.975,
+      "step": 17040
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.837970898829371e-06,
+      "loss": 1.059,
+      "step": 17041
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.836893168095549e-06,
+      "loss": 0.8948,
+      "step": 17042
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.835815552778564e-06,
+      "loss": 0.9803,
+      "step": 17043
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8347380528985975e-06,
+      "loss": 0.9785,
+      "step": 17044
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.833660668475827e-06,
+      "loss": 0.9855,
+      "step": 17045
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.832583399530434e-06,
+      "loss": 1.039,
+      "step": 17046
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8315062460825865e-06,
+      "loss": 0.925,
+      "step": 17047
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.830429208152457e-06,
+      "loss": 0.948,
+      "step": 17048
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.829352285760215e-06,
+      "loss": 0.9512,
+      "step": 17049
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.82827547892603e-06,
+      "loss": 0.9674,
+      "step": 17050
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8271987876700715e-06,
+      "loss": 0.9599,
+      "step": 17051
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.826122212012491e-06,
+      "loss": 0.8957,
+      "step": 17052
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.825045751973464e-06,
+      "loss": 0.9789,
+      "step": 17053
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.823969407573137e-06,
+      "loss": 0.9634,
+      "step": 17054
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.822893178831675e-06,
+      "loss": 1.0016,
+      "step": 17055
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.821817065769231e-06,
+      "loss": 0.9997,
+      "step": 17056
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.820741068405949e-06,
+      "loss": 0.992,
+      "step": 17057
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8196651867619935e-06,
+      "loss": 0.9342,
+      "step": 17058
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.818589420857499e-06,
+      "loss": 0.9985,
+      "step": 17059
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8175137707126245e-06,
+      "loss": 0.9672,
+      "step": 17060
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.816438236347503e-06,
+      "loss": 0.9613,
+      "step": 17061
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.81536281778228e-06,
+      "loss": 0.9551,
+      "step": 17062
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.814287515037097e-06,
+      "loss": 1.0236,
+      "step": 17063
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8132123281320875e-06,
+      "loss": 0.9789,
+      "step": 17064
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.812137257087392e-06,
+      "loss": 0.9747,
+      "step": 17065
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.811062301923138e-06,
+      "loss": 0.9832,
+      "step": 17066
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8099874626594566e-06,
+      "loss": 0.9807,
+      "step": 17067
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.808912739316478e-06,
+      "loss": 0.9089,
+      "step": 17068
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.807838131914329e-06,
+      "loss": 0.9416,
+      "step": 17069
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.806763640473132e-06,
+      "loss": 1.0012,
+      "step": 17070
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.805689265013015e-06,
+      "loss": 0.9977,
+      "step": 17071
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.804615005554085e-06,
+      "loss": 0.9797,
+      "step": 17072
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.803540862116477e-06,
+      "loss": 1.0043,
+      "step": 17073
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.802466834720292e-06,
+      "loss": 0.8939,
+      "step": 17074
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.801392923385651e-06,
+      "loss": 0.946,
+      "step": 17075
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.800319128132664e-06,
+      "loss": 1.0073,
+      "step": 17076
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.799245448981433e-06,
+      "loss": 0.9448,
+      "step": 17077
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.798171885952079e-06,
+      "loss": 1.0072,
+      "step": 17078
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.79709843906469e-06,
+      "loss": 0.9279,
+      "step": 17079
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.796025108339385e-06,
+      "loss": 1.0238,
+      "step": 17080
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.794951893796254e-06,
+      "loss": 0.9899,
+      "step": 17081
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.793878795455395e-06,
+      "loss": 0.9519,
+      "step": 17082
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.792805813336911e-06,
+      "loss": 0.967,
+      "step": 17083
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.791732947460885e-06,
+      "loss": 1.0032,
+      "step": 17084
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.790660197847422e-06,
+      "loss": 0.987,
+      "step": 17085
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.789587564516601e-06,
+      "loss": 1.0081,
+      "step": 17086
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7885150474885125e-06,
+      "loss": 0.9963,
+      "step": 17087
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.787442646783242e-06,
+      "loss": 0.9592,
+      "step": 17088
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.78637036242087e-06,
+      "loss": 0.9913,
+      "step": 17089
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.785298194421481e-06,
+      "loss": 0.9465,
+      "step": 17090
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.784226142805156e-06,
+      "loss": 0.9775,
+      "step": 17091
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.783154207591963e-06,
+      "loss": 0.9883,
+      "step": 17092
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.782082388801981e-06,
+      "loss": 0.9524,
+      "step": 17093
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7810106864552804e-06,
+      "loss": 1.0331,
+      "step": 17094
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.779939100571933e-06,
+      "loss": 0.9753,
+      "step": 17095
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.778867631172009e-06,
+      "loss": 0.963,
+      "step": 17096
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.777796278275564e-06,
+      "loss": 0.9687,
+      "step": 17097
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7767250419026735e-06,
+      "loss": 1.0515,
+      "step": 17098
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7756539220733894e-06,
+      "loss": 0.994,
+      "step": 17099
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.774582918807775e-06,
+      "loss": 1.0024,
+      "step": 17100
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7735120321258855e-06,
+      "loss": 0.9698,
+      "step": 17101
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.772441262047775e-06,
+      "loss": 0.9877,
+      "step": 17102
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.771370608593503e-06,
+      "loss": 1.0114,
+      "step": 17103
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.770300071783108e-06,
+      "loss": 0.9531,
+      "step": 17104
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.769229651636644e-06,
+      "loss": 0.9493,
+      "step": 17105
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.768159348174156e-06,
+      "loss": 0.9597,
+      "step": 17106
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.767089161415688e-06,
+      "loss": 0.9447,
+      "step": 17107
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.766019091381283e-06,
+      "loss": 0.9276,
+      "step": 17108
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.764949138090982e-06,
+      "loss": 0.8884,
+      "step": 17109
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7638793015648145e-06,
+      "loss": 0.8984,
+      "step": 17110
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.76280958182282e-06,
+      "loss": 1.0232,
+      "step": 17111
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.761739978885029e-06,
+      "loss": 0.9919,
+      "step": 17112
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.760670492771477e-06,
+      "loss": 0.9324,
+      "step": 17113
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.759601123502191e-06,
+      "loss": 0.9365,
+      "step": 17114
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.758531871097188e-06,
+      "loss": 0.9822,
+      "step": 17115
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.757462735576506e-06,
+      "loss": 0.9423,
+      "step": 17116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.756393716960154e-06,
+      "loss": 0.9943,
+      "step": 17117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.755324815268166e-06,
+      "loss": 0.9787,
+      "step": 17118
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.754256030520547e-06,
+      "loss": 1.0022,
+      "step": 17119
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.753187362737315e-06,
+      "loss": 0.9606,
+      "step": 17120
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.752118811938488e-06,
+      "loss": 0.9684,
+      "step": 17121
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.751050378144066e-06,
+      "loss": 0.9236,
+      "step": 17122
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.749982061374071e-06,
+      "loss": 1.0228,
+      "step": 17123
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7489138616485015e-06,
+      "loss": 1.0115,
+      "step": 17124
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.747845778987361e-06,
+      "loss": 0.9925,
+      "step": 17125
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.746777813410656e-06,
+      "loss": 1.0371,
+      "step": 17126
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.745709964938383e-06,
+      "loss": 0.9313,
+      "step": 17127
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.744642233590541e-06,
+      "loss": 0.9518,
+      "step": 17128
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7435746193871295e-06,
+      "loss": 0.953,
+      "step": 17129
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.742507122348133e-06,
+      "loss": 1.016,
+      "step": 17130
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.741439742493548e-06,
+      "loss": 1.0323,
+      "step": 17131
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.740372479843362e-06,
+      "loss": 0.9976,
+      "step": 17132
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.739305334417562e-06,
+      "loss": 0.9631,
+      "step": 17133
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.738238306236136e-06,
+      "loss": 1.0451,
+      "step": 17134
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.737171395319057e-06,
+      "loss": 0.9252,
+      "step": 17135
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.736104601686318e-06,
+      "loss": 0.9029,
+      "step": 17136
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.735037925357886e-06,
+      "loss": 0.9219,
+      "step": 17137
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7339713663537415e-06,
+      "loss": 1.0146,
+      "step": 17138
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.732904924693856e-06,
+      "loss": 0.9918,
+      "step": 17139
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.731838600398203e-06,
+      "loss": 1.0135,
+      "step": 17140
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.730772393486753e-06,
+      "loss": 0.9803,
+      "step": 17141
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.729706303979466e-06,
+      "loss": 0.9857,
+      "step": 17142
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.728640331896315e-06,
+      "loss": 0.9769,
+      "step": 17143
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.727574477257258e-06,
+      "loss": 0.9534,
+      "step": 17144
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.726508740082255e-06,
+      "loss": 1.0075,
+      "step": 17145
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.725443120391265e-06,
+      "loss": 0.9419,
+      "step": 17146
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.724377618204245e-06,
+      "loss": 1.0349,
+      "step": 17147
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.72331223354115e-06,
+      "loss": 1.0057,
+      "step": 17148
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.722246966421928e-06,
+      "loss": 0.9102,
+      "step": 17149
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.721181816866528e-06,
+      "loss": 0.9472,
+      "step": 17150
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.720116784894899e-06,
+      "loss": 0.932,
+      "step": 17151
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.719051870526987e-06,
+      "loss": 1.0029,
+      "step": 17152
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7179870737827314e-06,
+      "loss": 1.0312,
+      "step": 17153
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.716922394682079e-06,
+      "loss": 0.9545,
+      "step": 17154
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.715857833244958e-06,
+      "loss": 0.9627,
+      "step": 17155
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.714793389491317e-06,
+      "loss": 0.9949,
+      "step": 17156
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.713729063441079e-06,
+      "loss": 1.0366,
+      "step": 17157
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7126648551141805e-06,
+      "loss": 0.9706,
+      "step": 17158
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.711600764530553e-06,
+      "loss": 0.9382,
+      "step": 17159
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.710536791710115e-06,
+      "loss": 0.9444,
+      "step": 17160
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.709472936672805e-06,
+      "loss": 0.9453,
+      "step": 17161
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.708409199438531e-06,
+      "loss": 0.9703,
+      "step": 17162
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.707345580027229e-06,
+      "loss": 0.9324,
+      "step": 17163
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.706282078458806e-06,
+      "loss": 0.9872,
+      "step": 17164
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.705218694753182e-06,
+      "loss": 0.9474,
+      "step": 17165
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.70415542893027e-06,
+      "loss": 0.933,
+      "step": 17166
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.703092281009985e-06,
+      "loss": 1.0393,
+      "step": 17167
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.702029251012237e-06,
+      "loss": 0.988,
+      "step": 17168
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.700966338956928e-06,
+      "loss": 0.9253,
+      "step": 17169
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.699903544863965e-06,
+      "loss": 0.9632,
+      "step": 17170
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.698840868753253e-06,
+      "loss": 0.9525,
+      "step": 17171
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.697778310644691e-06,
+      "loss": 0.9243,
+      "step": 17172
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.69671587055818e-06,
+      "loss": 0.9207,
+      "step": 17173
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.695653548513618e-06,
+      "loss": 0.9828,
+      "step": 17174
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.694591344530892e-06,
+      "loss": 0.9475,
+      "step": 17175
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.693529258629899e-06,
+      "loss": 0.9939,
+      "step": 17176
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6924672908305275e-06,
+      "loss": 0.9489,
+      "step": 17177
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.691405441152665e-06,
+      "loss": 0.8912,
+      "step": 17178
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6903437096162e-06,
+      "loss": 1.0228,
+      "step": 17179
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.689282096241006e-06,
+      "loss": 0.9958,
+      "step": 17180
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.688220601046978e-06,
+      "loss": 0.9844,
+      "step": 17181
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.687159224053984e-06,
+      "loss": 1.0294,
+      "step": 17182
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.686097965281904e-06,
+      "loss": 0.9505,
+      "step": 17183
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.685036824750611e-06,
+      "loss": 0.97,
+      "step": 17184
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.683975802479978e-06,
+      "loss": 0.9691,
+      "step": 17185
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.68291489848988e-06,
+      "loss": 0.9443,
+      "step": 17186
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.681854112800175e-06,
+      "loss": 0.9469,
+      "step": 17187
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.680793445430733e-06,
+      "loss": 0.8482,
+      "step": 17188
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6797328964014165e-06,
+      "loss": 0.9296,
+      "step": 17189
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.678672465732088e-06,
+      "loss": 1.0244,
+      "step": 17190
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.677612153442604e-06,
+      "loss": 0.9663,
+      "step": 17191
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.676551959552828e-06,
+      "loss": 0.8751,
+      "step": 17192
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.675491884082599e-06,
+      "loss": 1.0001,
+      "step": 17193
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6744319270517865e-06,
+      "loss": 1.0123,
+      "step": 17194
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6733720884802306e-06,
+      "loss": 0.998,
+      "step": 17195
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.672312368387779e-06,
+      "loss": 0.9559,
+      "step": 17196
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.671252766794283e-06,
+      "loss": 0.9882,
+      "step": 17197
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6701932837195755e-06,
+      "loss": 0.9201,
+      "step": 17198
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.66913391918351e-06,
+      "loss": 0.9593,
+      "step": 17199
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.668074673205914e-06,
+      "loss": 0.9624,
+      "step": 17200
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.667015545806633e-06,
+      "loss": 0.9445,
+      "step": 17201
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.665956537005497e-06,
+      "loss": 0.9503,
+      "step": 17202
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.664897646822337e-06,
+      "loss": 0.9845,
+      "step": 17203
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.663838875276985e-06,
+      "loss": 0.9533,
+      "step": 17204
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.662780222389268e-06,
+      "loss": 0.993,
+      "step": 17205
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.661721688179016e-06,
+      "loss": 0.9579,
+      "step": 17206
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.660663272666044e-06,
+      "loss": 0.9743,
+      "step": 17207
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.659604975870177e-06,
+      "loss": 0.9365,
+      "step": 17208
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6585467978112335e-06,
+      "loss": 0.99,
+      "step": 17209
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.65748873850903e-06,
+      "loss": 0.9161,
+      "step": 17210
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.656430797983381e-06,
+      "loss": 0.9222,
+      "step": 17211
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.655372976254102e-06,
+      "loss": 0.9955,
+      "step": 17212
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.654315273340997e-06,
+      "loss": 0.9792,
+      "step": 17213
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.653257689263875e-06,
+      "loss": 0.9905,
+      "step": 17214
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.652200224042544e-06,
+      "loss": 0.9712,
+      "step": 17215
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6511428776968044e-06,
+      "loss": 0.9063,
+      "step": 17216
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.650085650246463e-06,
+      "loss": 0.9213,
+      "step": 17217
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.649028541711308e-06,
+      "loss": 0.9724,
+      "step": 17218
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.647971552111148e-06,
+      "loss": 0.9889,
+      "step": 17219
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.646914681465765e-06,
+      "loss": 1.0196,
+      "step": 17220
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.645857929794963e-06,
+      "loss": 0.9955,
+      "step": 17221
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.644801297118526e-06,
+      "loss": 1.029,
+      "step": 17222
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.64374478345624e-06,
+      "loss": 0.9746,
+      "step": 17223
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.642688388827895e-06,
+      "loss": 0.9854,
+      "step": 17224
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6416321132532665e-06,
+      "loss": 0.9606,
+      "step": 17225
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.640575956752145e-06,
+      "loss": 0.9635,
+      "step": 17226
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6395199193443035e-06,
+      "loss": 0.9447,
+      "step": 17227
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.638464001049518e-06,
+      "loss": 1.0236,
+      "step": 17228
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.637408201887567e-06,
+      "loss": 0.9677,
+      "step": 17229
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6363525218782164e-06,
+      "loss": 0.974,
+      "step": 17230
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.635296961041242e-06,
+      "loss": 1.013,
+      "step": 17231
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.63424151939641e-06,
+      "loss": 0.9417,
+      "step": 17232
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.633186196963484e-06,
+      "loss": 1.051,
+      "step": 17233
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.632130993762226e-06,
+      "loss": 1.0079,
+      "step": 17234
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.631075909812398e-06,
+      "loss": 0.9592,
+      "step": 17235
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.63002094513376e-06,
+      "loss": 0.9519,
+      "step": 17236
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.628966099746071e-06,
+      "loss": 0.9311,
+      "step": 17237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6279113736690734e-06,
+      "loss": 0.9811,
+      "step": 17238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.626856766922535e-06,
+      "loss": 0.93,
+      "step": 17239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.625802279526194e-06,
+      "loss": 0.94,
+      "step": 17240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.624747911499802e-06,
+      "loss": 0.9196,
+      "step": 17241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.623693662863106e-06,
+      "loss": 0.9745,
+      "step": 17242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.622639533635842e-06,
+      "loss": 1.0239,
+      "step": 17243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.621585523837761e-06,
+      "loss": 0.9759,
+      "step": 17244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.620531633488594e-06,
+      "loss": 0.881,
+      "step": 17245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.619477862608079e-06,
+      "loss": 0.9327,
+      "step": 17246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.618424211215949e-06,
+      "loss": 0.9134,
+      "step": 17247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.617370679331938e-06,
+      "loss": 0.9539,
+      "step": 17248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.616317266975774e-06,
+      "loss": 0.9483,
+      "step": 17249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.615263974167189e-06,
+      "loss": 0.9909,
+      "step": 17250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6142108009259e-06,
+      "loss": 0.9401,
+      "step": 17251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.613157747271635e-06,
+      "loss": 0.9581,
+      "step": 17252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.612104813224113e-06,
+      "loss": 0.972,
+      "step": 17253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6110519988030526e-06,
+      "loss": 0.9512,
+      "step": 17254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.609999304028173e-06,
+      "loss": 0.9279,
+      "step": 17255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.608946728919179e-06,
+      "loss": 0.9532,
+      "step": 17256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6078942734957955e-06,
+      "loss": 0.9842,
+      "step": 17257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.606841937777717e-06,
+      "loss": 0.9856,
+      "step": 17258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.605789721784666e-06,
+      "loss": 0.9697,
+      "step": 17259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6047376255363365e-06,
+      "loss": 0.9321,
+      "step": 17260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.603685649052436e-06,
+      "loss": 1.0126,
+      "step": 17261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.602633792352664e-06,
+      "loss": 0.9326,
+      "step": 17262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.601582055456712e-06,
+      "loss": 0.8961,
+      "step": 17263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.600530438384291e-06,
+      "loss": 0.9608,
+      "step": 17264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.599478941155082e-06,
+      "loss": 0.928,
+      "step": 17265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.598427563788778e-06,
+      "loss": 0.8577,
+      "step": 17266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.597376306305072e-06,
+      "loss": 0.9457,
+      "step": 17267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.596325168723649e-06,
+      "loss": 1.033,
+      "step": 17268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.595274151064197e-06,
+      "loss": 0.952,
+      "step": 17269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5942232533463926e-06,
+      "loss": 0.9672,
+      "step": 17270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.593172475589919e-06,
+      "loss": 0.9501,
+      "step": 17271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.592121817814454e-06,
+      "loss": 0.9476,
+      "step": 17272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.591071280039674e-06,
+      "loss": 0.9078,
+      "step": 17273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.590020862285251e-06,
+      "loss": 0.9002,
+      "step": 17274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.588970564570861e-06,
+      "loss": 0.9214,
+      "step": 17275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.587920386916162e-06,
+      "loss": 1.0124,
+      "step": 17276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.586870329340835e-06,
+      "loss": 0.9712,
+      "step": 17277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.585820391864535e-06,
+      "loss": 1.004,
+      "step": 17278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.584770574506926e-06,
+      "loss": 0.9662,
+      "step": 17279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.583720877287672e-06,
+      "loss": 1.029,
+      "step": 17280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.58267130022642e-06,
+      "loss": 0.9591,
+      "step": 17281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.58162184334284e-06,
+      "loss": 0.9271,
+      "step": 17282
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.580572506656571e-06,
+      "loss": 0.9615,
+      "step": 17283
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.579523290187278e-06,
+      "loss": 1.0104,
+      "step": 17284
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5784741939546e-06,
+      "loss": 0.9729,
+      "step": 17285
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5774252179781855e-06,
+      "loss": 0.9287,
+      "step": 17286
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5763763622776795e-06,
+      "loss": 0.9146,
+      "step": 17287
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.575327626872724e-06,
+      "loss": 0.9508,
+      "step": 17288
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.574279011782963e-06,
+      "loss": 0.9937,
+      "step": 17289
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.573230517028025e-06,
+      "loss": 0.8948,
+      "step": 17290
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.57218214262755e-06,
+      "loss": 0.9501,
+      "step": 17291
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5711338886011715e-06,
+      "loss": 0.921,
+      "step": 17292
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5700857549685195e-06,
+      "loss": 0.9709,
+      "step": 17293
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.569037741749222e-06,
+      "loss": 0.9451,
+      "step": 17294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.56798984896291e-06,
+      "loss": 0.9593,
+      "step": 17295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.566942076629198e-06,
+      "loss": 1.0086,
+      "step": 17296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.565894424767715e-06,
+      "loss": 0.9571,
+      "step": 17297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.564846893398078e-06,
+      "loss": 0.8726,
+      "step": 17298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5637994825399035e-06,
+      "loss": 0.9122,
+      "step": 17299
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.56275219221281e-06,
+      "loss": 1.063,
+      "step": 17300
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.561705022436402e-06,
+      "loss": 1.0076,
+      "step": 17301
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.560657973230302e-06,
+      "loss": 0.9967,
+      "step": 17302
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.559611044614105e-06,
+      "loss": 1.0549,
+      "step": 17303
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.55856423660743e-06,
+      "loss": 0.9429,
+      "step": 17304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.557517549229872e-06,
+      "loss": 0.9473,
+      "step": 17305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.556470982501034e-06,
+      "loss": 1.022,
+      "step": 17306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5554245364405175e-06,
+      "loss": 0.9793,
+      "step": 17307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.554378211067911e-06,
+      "loss": 0.936,
+      "step": 17308
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.553332006402823e-06,
+      "loss": 0.9188,
+      "step": 17309
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.552285922464835e-06,
+      "loss": 1.0034,
+      "step": 17310
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.55123995927354e-06,
+      "loss": 0.9174,
+      "step": 17311
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.550194116848527e-06,
+      "loss": 0.9801,
+      "step": 17312
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5491483952093834e-06,
+      "loss": 0.9924,
+      "step": 17313
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5481027943756846e-06,
+      "loss": 0.967,
+      "step": 17314
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.547057314367022e-06,
+      "loss": 0.8956,
+      "step": 17315
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.546011955202967e-06,
+      "loss": 0.9676,
+      "step": 17316
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5449667169030974e-06,
+      "loss": 0.9934,
+      "step": 17317
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.543921599486993e-06,
+      "loss": 0.9606,
+      "step": 17318
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.542876602974214e-06,
+      "loss": 0.9573,
+      "step": 17319
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5418317273843425e-06,
+      "loss": 0.9648,
+      "step": 17320
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.540786972736936e-06,
+      "loss": 0.9002,
+      "step": 17321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.53974233905157e-06,
+      "loss": 0.9435,
+      "step": 17322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.538697826347799e-06,
+      "loss": 0.9964,
+      "step": 17323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5376534346451865e-06,
+      "loss": 0.9764,
+      "step": 17324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.53660916396329e-06,
+      "loss": 1.0016,
+      "step": 17325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.535565014321668e-06,
+      "loss": 0.9714,
+      "step": 17326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.534520985739875e-06,
+      "loss": 0.9878,
+      "step": 17327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.533477078237458e-06,
+      "loss": 0.968,
+      "step": 17328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.532433291833967e-06,
+      "loss": 0.9664,
+      "step": 17329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.531389626548952e-06,
+      "loss": 0.9555,
+      "step": 17330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.530346082401956e-06,
+      "loss": 0.9961,
+      "step": 17331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.529302659412522e-06,
+      "loss": 0.9612,
+      "step": 17332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.528259357600194e-06,
+      "loss": 0.9109,
+      "step": 17333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5272161769845034e-06,
+      "loss": 0.96,
+      "step": 17334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.526173117584989e-06,
+      "loss": 0.9919,
+      "step": 17335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.525130179421182e-06,
+      "loss": 0.963,
+      "step": 17336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.524087362512617e-06,
+      "loss": 0.9655,
+      "step": 17337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.523044666878826e-06,
+      "loss": 1.0405,
+      "step": 17338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.522002092539321e-06,
+      "loss": 1.0166,
+      "step": 17339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.520959639513646e-06,
+      "loss": 0.9711,
+      "step": 17340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.519917307821306e-06,
+      "loss": 0.9744,
+      "step": 17341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.518875097481834e-06,
+      "loss": 0.9365,
+      "step": 17342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51783300851474e-06,
+      "loss": 0.9281,
+      "step": 17343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51679104093954e-06,
+      "loss": 1.0205,
+      "step": 17344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.515749194775752e-06,
+      "loss": 0.9544,
+      "step": 17345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.514707470042875e-06,
+      "loss": 0.9755,
+      "step": 17346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.513665866760431e-06,
+      "loss": 0.9299,
+      "step": 17347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.51262438494792e-06,
+      "loss": 0.9436,
+      "step": 17348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.511583024624844e-06,
+      "loss": 0.9771,
+      "step": 17349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.510541785810708e-06,
+      "loss": 0.9572,
+      "step": 17350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5095006685250095e-06,
+      "loss": 0.9693,
+      "step": 17351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.508459672787246e-06,
+      "loss": 0.967,
+      "step": 17352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.507418798616917e-06,
+      "loss": 0.9792,
+      "step": 17353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.506378046033507e-06,
+      "loss": 0.9985,
+      "step": 17354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.505337415056509e-06,
+      "loss": 0.9458,
+      "step": 17355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.504296905705412e-06,
+      "loss": 1.03,
+      "step": 17356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.503256517999701e-06,
+      "loss": 0.9897,
+      "step": 17357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.502216251958863e-06,
+      "loss": 0.9346,
+      "step": 17358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.501176107602371e-06,
+      "loss": 1.0013,
+      "step": 17359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.500136084949714e-06,
+      "loss": 0.9428,
+      "step": 17360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.499096184020359e-06,
+      "loss": 0.9294,
+      "step": 17361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.498056404833786e-06,
+      "loss": 1.0512,
+      "step": 17362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.497016747409465e-06,
+      "loss": 0.9827,
+      "step": 17363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.495977211766868e-06,
+      "loss": 0.9808,
+      "step": 17364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.494937797925462e-06,
+      "loss": 0.9693,
+      "step": 17365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.493898505904705e-06,
+      "loss": 0.9443,
+      "step": 17366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4928593357240725e-06,
+      "loss": 0.9454,
+      "step": 17367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.491820287403015e-06,
+      "loss": 0.9801,
+      "step": 17368
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.490781360960995e-06,
+      "loss": 0.9419,
+      "step": 17369
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.489742556417465e-06,
+      "loss": 0.9757,
+      "step": 17370
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.488703873791884e-06,
+      "loss": 0.9281,
+      "step": 17371
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.487665313103704e-06,
+      "loss": 0.9859,
+      "step": 17372
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4866268743723675e-06,
+      "loss": 0.971,
+      "step": 17373
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.485588557617325e-06,
+      "loss": 0.9554,
+      "step": 17374
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.484550362858022e-06,
+      "loss": 0.9969,
+      "step": 17375
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.483512290113898e-06,
+      "loss": 0.9204,
+      "step": 17376
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.482474339404395e-06,
+      "loss": 0.938,
+      "step": 17377
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.481436510748956e-06,
+      "loss": 0.9789,
+      "step": 17378
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.480398804167004e-06,
+      "loss": 0.9958,
+      "step": 17379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.479361219677985e-06,
+      "loss": 0.9232,
+      "step": 17380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.478323757301323e-06,
+      "loss": 1.0161,
+      "step": 17381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.477286417056447e-06,
+      "loss": 0.955,
+      "step": 17382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.476249198962788e-06,
+      "loss": 0.9846,
+      "step": 17383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.475212103039759e-06,
+      "loss": 0.9075,
+      "step": 17384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.474175129306795e-06,
+      "loss": 0.9444,
+      "step": 17385
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.473138277783307e-06,
+      "loss": 0.9872,
+      "step": 17386
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.472101548488715e-06,
+      "loss": 0.9508,
+      "step": 17387
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.471064941442434e-06,
+      "loss": 0.9966,
+      "step": 17388
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.470028456663875e-06,
+      "loss": 0.9715,
+      "step": 17389
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.468992094172451e-06,
+      "loss": 0.9351,
+      "step": 17390
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.46795585398757e-06,
+      "loss": 0.9349,
+      "step": 17391
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.466919736128634e-06,
+      "loss": 1.0146,
+      "step": 17392
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.46588374061505e-06,
+      "loss": 0.9617,
+      "step": 17393
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.464847867466215e-06,
+      "loss": 0.9701,
+      "step": 17394
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.463812116701533e-06,
+      "loss": 0.9157,
+      "step": 17395
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.4627764883404e-06,
+      "loss": 0.987,
+      "step": 17396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.461740982402203e-06,
+      "loss": 0.9279,
+      "step": 17397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.460705598906345e-06,
+      "loss": 0.9758,
+      "step": 17398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4596703378722075e-06,
+      "loss": 0.9225,
+      "step": 17399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.45863519931918e-06,
+      "loss": 0.8961,
+      "step": 17400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.457600183266649e-06,
+      "loss": 1.0503,
+      "step": 17401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4565652897339966e-06,
+      "loss": 0.9705,
+      "step": 17402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.455530518740606e-06,
+      "loss": 0.9993,
+      "step": 17403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.454495870305846e-06,
+      "loss": 0.9538,
+      "step": 17404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.453461344449106e-06,
+      "loss": 0.9562,
+      "step": 17405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.45242694118975e-06,
+      "loss": 0.9734,
+      "step": 17406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.45139266054715e-06,
+      "loss": 1.0127,
+      "step": 17407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.450358502540678e-06,
+      "loss": 1.0324,
+      "step": 17408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4493244671897e-06,
+      "loss": 0.9451,
+      "step": 17409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4482905545135834e-06,
+      "loss": 0.9402,
+      "step": 17410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.447256764531685e-06,
+      "loss": 0.9758,
+      "step": 17411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.446223097263365e-06,
+      "loss": 0.9145,
+      "step": 17412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.445189552727984e-06,
+      "loss": 0.9769,
+      "step": 17413
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.444156130944896e-06,
+      "loss": 0.9373,
+      "step": 17414
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.443122831933453e-06,
+      "loss": 0.9922,
+      "step": 17415
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4420896557130105e-06,
+      "loss": 0.9853,
+      "step": 17416
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.441056602302907e-06,
+      "loss": 0.9291,
+      "step": 17417
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.440023671722501e-06,
+      "loss": 0.969,
+      "step": 17418
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4389908639911255e-06,
+      "loss": 1.0075,
+      "step": 17419
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.437958179128127e-06,
+      "loss": 1.0211,
+      "step": 17420
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.436925617152846e-06,
+      "loss": 0.9634,
+      "step": 17421
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.435893178084612e-06,
+      "loss": 0.9519,
+      "step": 17422
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.43486086194277e-06,
+      "loss": 0.9569,
+      "step": 17423
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4338286687466415e-06,
+      "loss": 0.9778,
+      "step": 17424
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.432796598515568e-06,
+      "loss": 0.9321,
+      "step": 17425
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.431764651268869e-06,
+      "loss": 0.9118,
+      "step": 17426
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.430732827025869e-06,
+      "loss": 1.0678,
+      "step": 17427
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4297011258059e-06,
+      "loss": 0.9195,
+      "step": 17428
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.428669547628268e-06,
+      "loss": 1.0099,
+      "step": 17429
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4276380925123084e-06,
+      "loss": 1.0071,
+      "step": 17430
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.426606760477325e-06,
+      "loss": 0.9875,
+      "step": 17431
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.425575551542635e-06,
+      "loss": 0.9682,
+      "step": 17432
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4245444657275515e-06,
+      "loss": 0.9901,
+      "step": 17433
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.423513503051382e-06,
+      "loss": 0.9378,
+      "step": 17434
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.422482663533434e-06,
+      "loss": 0.988,
+      "step": 17435
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.421451947193015e-06,
+      "loss": 1.0351,
+      "step": 17436
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.420421354049422e-06,
+      "loss": 0.9974,
+      "step": 17437
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.419390884121958e-06,
+      "loss": 0.9767,
+      "step": 17438
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.418360537429919e-06,
+      "loss": 0.9929,
+      "step": 17439
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4173303139926014e-06,
+      "loss": 0.9413,
+      "step": 17440
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.416300213829302e-06,
+      "loss": 0.9984,
+      "step": 17441
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.415270236959302e-06,
+      "loss": 0.988,
+      "step": 17442
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4142403834019004e-06,
+      "loss": 0.9916,
+      "step": 17443
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.413210653176374e-06,
+      "loss": 0.9981,
+      "step": 17444
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.412181046302016e-06,
+      "loss": 1.0071,
+      "step": 17445
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.411151562798102e-06,
+      "loss": 1.0152,
+      "step": 17446
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.410122202683911e-06,
+      "loss": 0.9604,
+      "step": 17447
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.409092965978724e-06,
+      "loss": 0.9883,
+      "step": 17448
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.408063852701809e-06,
+      "loss": 0.9692,
+      "step": 17449
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.407034862872444e-06,
+      "loss": 0.9605,
+      "step": 17450
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.406005996509896e-06,
+      "loss": 0.9333,
+      "step": 17451
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4049772536334336e-06,
+      "loss": 1.014,
+      "step": 17452
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.403948634262322e-06,
+      "loss": 0.9762,
+      "step": 17453
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4029201384158275e-06,
+      "loss": 0.9949,
+      "step": 17454
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.401891766113205e-06,
+      "loss": 0.9677,
+      "step": 17455
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.400863517373714e-06,
+      "loss": 1.029,
+      "step": 17456
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3998353922166134e-06,
+      "loss": 1.0363,
+      "step": 17457
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.398807390661156e-06,
+      "loss": 0.961,
+      "step": 17458
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3977795127265955e-06,
+      "loss": 0.9634,
+      "step": 17459
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.396751758432173e-06,
+      "loss": 0.9764,
+      "step": 17460
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.395724127797145e-06,
+      "loss": 0.9664,
+      "step": 17461
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.394696620840747e-06,
+      "loss": 0.9556,
+      "step": 17462
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3936692375822325e-06,
+      "loss": 0.935,
+      "step": 17463
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.39264197804083e-06,
+      "loss": 1.0156,
+      "step": 17464
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.391614842235783e-06,
+      "loss": 0.9894,
+      "step": 17465
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.390587830186329e-06,
+      "loss": 0.9826,
+      "step": 17466
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.38956094191169e-06,
+      "loss": 0.9608,
+      "step": 17467
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.388534177431112e-06,
+      "loss": 0.845,
+      "step": 17468
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.387507536763811e-06,
+      "loss": 0.9922,
+      "step": 17469
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.386481019929016e-06,
+      "loss": 0.9856,
+      "step": 17470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.385454626945954e-06,
+      "loss": 0.9388,
+      "step": 17471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.384428357833843e-06,
+      "loss": 0.9175,
+      "step": 17472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.383402212611904e-06,
+      "loss": 0.962,
+      "step": 17473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.382376191299357e-06,
+      "loss": 0.9711,
+      "step": 17474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.381350293915407e-06,
+      "loss": 0.8912,
+      "step": 17475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.380324520479272e-06,
+      "loss": 0.9334,
+      "step": 17476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.379298871010162e-06,
+      "loss": 0.9475,
+      "step": 17477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.378273345527282e-06,
+      "loss": 0.9796,
+      "step": 17478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.377247944049843e-06,
+      "loss": 0.953,
+      "step": 17479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3762226665970356e-06,
+      "loss": 1.015,
+      "step": 17480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.375197513188075e-06,
+      "loss": 1.0296,
+      "step": 17481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3741724838421485e-06,
+      "loss": 0.9618,
+      "step": 17482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3731475785784564e-06,
+      "loss": 0.9737,
+      "step": 17483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.372122797416191e-06,
+      "loss": 0.9434,
+      "step": 17484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3710981403745425e-06,
+      "loss": 0.9639,
+      "step": 17485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.370073607472706e-06,
+      "loss": 0.9746,
+      "step": 17486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.369049198729855e-06,
+      "loss": 0.9056,
+      "step": 17487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.368024914165189e-06,
+      "loss": 0.9802,
+      "step": 17488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.367000753797879e-06,
+      "loss": 0.9264,
+      "step": 17489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.365976717647107e-06,
+      "loss": 1.0235,
+      "step": 17490
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.364952805732052e-06,
+      "loss": 0.9349,
+      "step": 17491
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.363929018071887e-06,
+      "loss": 0.9607,
+      "step": 17492
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.36290535468579e-06,
+      "loss": 0.9749,
+      "step": 17493
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.361881815592923e-06,
+      "loss": 0.9838,
+      "step": 17494
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3608584008124576e-06,
+      "loss": 0.9527,
+      "step": 17495
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.35983511036356e-06,
+      "loss": 0.9838,
+      "step": 17496
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.358811944265391e-06,
+      "loss": 0.9433,
+      "step": 17497
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.357788902537113e-06,
+      "loss": 0.9369,
+      "step": 17498
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3567659851978905e-06,
+      "loss": 0.9832,
+      "step": 17499
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.355743192266866e-06,
+      "loss": 0.9815,
+      "step": 17500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.354720523763207e-06,
+      "loss": 0.923,
+      "step": 17501
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3536979797060575e-06,
+      "loss": 0.9409,
+      "step": 17502
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.352675560114569e-06,
+      "loss": 0.9768,
+      "step": 17503
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.35165326500789e-06,
+      "loss": 0.9885,
+      "step": 17504
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.350631094405158e-06,
+      "loss": 0.9708,
+      "step": 17505
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.349609048325526e-06,
+      "loss": 0.9721,
+      "step": 17506
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.348587126788122e-06,
+      "loss": 0.9823,
+      "step": 17507
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.347565329812097e-06,
+      "loss": 0.9971,
+      "step": 17508
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.346543657416575e-06,
+      "loss": 0.9747,
+      "step": 17509
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.345522109620693e-06,
+      "loss": 0.9444,
+      "step": 17510
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.344500686443584e-06,
+      "loss": 0.9919,
+      "step": 17511
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.343479387904371e-06,
+      "loss": 0.944,
+      "step": 17512
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.342458214022187e-06,
+      "loss": 0.987,
+      "step": 17513
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.341437164816149e-06,
+      "loss": 1.0059,
+      "step": 17514
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.340416240305379e-06,
+      "loss": 0.9806,
+      "step": 17515
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3393954405089995e-06,
+      "loss": 0.9237,
+      "step": 17516
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3383747654461274e-06,
+      "loss": 0.9504,
+      "step": 17517
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.337354215135869e-06,
+      "loss": 0.9447,
+      "step": 17518
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.336333789597347e-06,
+      "loss": 0.9475,
+      "step": 17519
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.335313488849663e-06,
+      "loss": 0.9556,
+      "step": 17520
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.334293312911927e-06,
+      "loss": 0.9598,
+      "step": 17521
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.333273261803244e-06,
+      "loss": 0.9546,
+      "step": 17522
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3322533355427164e-06,
+      "loss": 0.9244,
+      "step": 17523
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3312335341494486e-06,
+      "loss": 0.8902,
+      "step": 17524
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.330213857642527e-06,
+      "loss": 0.9483,
+      "step": 17525
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.329194306041061e-06,
+      "loss": 0.9685,
+      "step": 17526
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.328174879364135e-06,
+      "loss": 0.9803,
+      "step": 17527
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.327155577630841e-06,
+      "loss": 0.9381,
+      "step": 17528
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.32613640086027e-06,
+      "loss": 0.957,
+      "step": 17529
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.325117349071506e-06,
+      "loss": 0.9231,
+      "step": 17530
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3240984222836374e-06,
+      "loss": 0.9485,
+      "step": 17531
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.323079620515739e-06,
+      "loss": 0.9849,
+      "step": 17532
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.322060943786892e-06,
+      "loss": 1.007,
+      "step": 17533
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.321042392116175e-06,
+      "loss": 0.9874,
+      "step": 17534
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.320023965522661e-06,
+      "loss": 0.975,
+      "step": 17535
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3190056640254225e-06,
+      "loss": 1.0811,
+      "step": 17536
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.317987487643535e-06,
+      "loss": 0.9039,
+      "step": 17537
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.316969436396052e-06,
+      "loss": 1.0157,
+      "step": 17538
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3159515103020546e-06,
+      "loss": 1.001,
+      "step": 17539
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.314933709380594e-06,
+      "loss": 0.9876,
+      "step": 17540
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.313916033650735e-06,
+      "loss": 0.9662,
+      "step": 17541
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.312898483131539e-06,
+      "loss": 0.9417,
+      "step": 17542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.311881057842051e-06,
+      "loss": 1.015,
+      "step": 17543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.310863757801339e-06,
+      "loss": 0.9635,
+      "step": 17544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.309846583028439e-06,
+      "loss": 0.9611,
+      "step": 17545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.308829533542413e-06,
+      "loss": 0.9542,
+      "step": 17546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3078126093623e-06,
+      "loss": 0.9997,
+      "step": 17547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3067958105071445e-06,
+      "loss": 0.9525,
+      "step": 17548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.30577913699599e-06,
+      "loss": 0.9366,
+      "step": 17549
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3047625888478756e-06,
+      "loss": 0.9311,
+      "step": 17550
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.303746166081839e-06,
+      "loss": 0.9992,
+      "step": 17551
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3027298687169105e-06,
+      "loss": 0.9351,
+      "step": 17552
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.301713696772126e-06,
+      "loss": 0.9651,
+      "step": 17553
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.300697650266514e-06,
+      "loss": 0.9627,
+      "step": 17554
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.299681729219104e-06,
+      "loss": 1.0081,
+      "step": 17555
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2986659336489185e-06,
+      "loss": 0.932,
+      "step": 17556
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.297650263574986e-06,
+      "loss": 0.9804,
+      "step": 17557
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2966347190163194e-06,
+      "loss": 0.996,
+      "step": 17558
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.29561929999194e-06,
+      "loss": 0.9023,
+      "step": 17559
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2946040065208656e-06,
+      "loss": 0.9585,
+      "step": 17560
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.293588838622106e-06,
+      "loss": 1.0453,
+      "step": 17561
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.292573796314678e-06,
+      "loss": 1.0113,
+      "step": 17562
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.29155887961758e-06,
+      "loss": 0.9793,
+      "step": 17563
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2905440885498305e-06,
+      "loss": 1.0283,
+      "step": 17564
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.289529423130423e-06,
+      "loss": 0.9538,
+      "step": 17565
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.288514883378371e-06,
+      "loss": 0.9728,
+      "step": 17566
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.287500469312663e-06,
+      "loss": 0.9834,
+      "step": 17567
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2864861809523e-06,
+      "loss": 0.9723,
+      "step": 17568
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.285472018316282e-06,
+      "loss": 0.9689,
+      "step": 17569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.284457981423587e-06,
+      "loss": 1.0329,
+      "step": 17570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.283444070293221e-06,
+      "loss": 0.998,
+      "step": 17571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.282430284944161e-06,
+      "loss": 0.9128,
+      "step": 17572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.281416625395397e-06,
+      "loss": 0.95,
+      "step": 17573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.280403091665909e-06,
+      "loss": 1.0063,
+      "step": 17574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.279389683774681e-06,
+      "loss": 0.9672,
+      "step": 17575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2783764017406875e-06,
+      "loss": 1.0313,
+      "step": 17576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2773632455829095e-06,
+      "loss": 0.9849,
+      "step": 17577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.276350215320314e-06,
+      "loss": 0.9754,
+      "step": 17578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.275337310971876e-06,
+      "loss": 0.9296,
+      "step": 17579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.274324532556562e-06,
+      "loss": 0.959,
+      "step": 17580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2733118800933395e-06,
+      "loss": 0.9825,
+      "step": 17581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.272299353601176e-06,
+      "loss": 0.911,
+      "step": 17582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.271286953099023e-06,
+      "loss": 0.9183,
+      "step": 17583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.270274678605853e-06,
+      "loss": 1.0125,
+      "step": 17584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.269262530140614e-06,
+      "loss": 0.963,
+      "step": 17585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.268250507722263e-06,
+      "loss": 0.9998,
+      "step": 17586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.267238611369751e-06,
+      "loss": 0.9691,
+      "step": 17587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.266226841102029e-06,
+      "loss": 0.9701,
+      "step": 17588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.265215196938048e-06,
+      "loss": 0.9373,
+      "step": 17589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.264203678896747e-06,
+      "loss": 0.9386,
+      "step": 17590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2631922869970705e-06,
+      "loss": 0.9735,
+      "step": 17591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.262181021257962e-06,
+      "loss": 0.899,
+      "step": 17592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.261169881698355e-06,
+      "loss": 0.9681,
+      "step": 17593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.260158868337187e-06,
+      "loss": 0.9758,
+      "step": 17594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.259147981193397e-06,
+      "loss": 0.9638,
+      "step": 17595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.258137220285906e-06,
+      "loss": 0.9496,
+      "step": 17596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.257126585633647e-06,
+      "loss": 0.9285,
+      "step": 17597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.256116077255548e-06,
+      "loss": 0.9576,
+      "step": 17598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2551056951705305e-06,
+      "loss": 0.9831,
+      "step": 17599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.25409543939752e-06,
+      "loss": 1.0017,
+      "step": 17600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.253085309955426e-06,
+      "loss": 0.9926,
+      "step": 17601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.252075306863178e-06,
+      "loss": 0.9682,
+      "step": 17602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.251065430139677e-06,
+      "loss": 0.944,
+      "step": 17603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.25005567980385e-06,
+      "loss": 0.9787,
+      "step": 17604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.249046055874594e-06,
+      "loss": 1.0131,
+      "step": 17605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.248036558370822e-06,
+      "loss": 0.9972,
+      "step": 17606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.24702718731144e-06,
+      "loss": 0.9046,
+      "step": 17607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2460179427153425e-06,
+      "loss": 0.939,
+      "step": 17608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.245008824601442e-06,
+      "loss": 0.8815,
+      "step": 17609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2439998329886276e-06,
+      "loss": 1.0204,
+      "step": 17610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.242990967895795e-06,
+      "loss": 1.0131,
+      "step": 17611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.241982229341842e-06,
+      "loss": 0.9964,
+      "step": 17612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.240973617345655e-06,
+      "loss": 0.9611,
+      "step": 17613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2399651319261285e-06,
+      "loss": 1.0163,
+      "step": 17614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.238956773102139e-06,
+      "loss": 0.9461,
+      "step": 17615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.237948540892577e-06,
+      "loss": 1.0243,
+      "step": 17616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.23694043531632e-06,
+      "loss": 0.9721,
+      "step": 17617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.235932456392249e-06,
+      "loss": 0.9411,
+      "step": 17618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2349246041392416e-06,
+      "loss": 0.9433,
+      "step": 17619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2339168785761726e-06,
+      "loss": 1.0023,
+      "step": 17620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.232909279721904e-06,
+      "loss": 1.0466,
+      "step": 17621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2319018075953195e-06,
+      "loss": 0.937,
+      "step": 17622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2308944622152774e-06,
+      "loss": 0.945,
+      "step": 17623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.229887243600642e-06,
+      "loss": 0.9594,
+      "step": 17624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.228880151770283e-06,
+      "loss": 0.9635,
+      "step": 17625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.227873186743047e-06,
+      "loss": 0.9821,
+      "step": 17626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.226866348537807e-06,
+      "loss": 0.952,
+      "step": 17627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.225859637173404e-06,
+      "loss": 0.8695,
+      "step": 17628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.224853052668703e-06,
+      "loss": 0.9977,
+      "step": 17629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.223846595042546e-06,
+      "loss": 0.9695,
+      "step": 17630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.222840264313783e-06,
+      "loss": 0.9796,
+      "step": 17631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.2218340605012606e-06,
+      "loss": 0.9765,
+      "step": 17632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.220827983623823e-06,
+      "loss": 0.9594,
+      "step": 17633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.219822033700312e-06,
+      "loss": 1.0254,
+      "step": 17634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.21881621074956e-06,
+      "loss": 0.973,
+      "step": 17635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.217810514790408e-06,
+      "loss": 1.009,
+      "step": 17636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.216804945841688e-06,
+      "loss": 1.0139,
+      "step": 17637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.215799503922232e-06,
+      "loss": 0.9427,
+      "step": 17638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.214794189050869e-06,
+      "loss": 0.9135,
+      "step": 17639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.213789001246428e-06,
+      "loss": 0.9713,
+      "step": 17640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.21278394052773e-06,
+      "loss": 0.9686,
+      "step": 17641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2117790069135954e-06,
+      "loss": 0.9347,
+      "step": 17642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.210774200422847e-06,
+      "loss": 0.9928,
+      "step": 17643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2097695210743e-06,
+      "loss": 1.0086,
+      "step": 17644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.208764968886774e-06,
+      "loss": 0.9439,
+      "step": 17645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2077605438790695e-06,
+      "loss": 0.9211,
+      "step": 17646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.20675624607001e-06,
+      "loss": 1.0505,
+      "step": 17647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.205752075478393e-06,
+      "loss": 0.8701,
+      "step": 17648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2047480321230325e-06,
+      "loss": 1.0289,
+      "step": 17649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2037441160227235e-06,
+      "loss": 1.0243,
+      "step": 17650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.20274032719627e-06,
+      "loss": 1.0031,
+      "step": 17651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.201736665662471e-06,
+      "loss": 0.9452,
+      "step": 17652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.200733131440116e-06,
+      "loss": 0.9698,
+      "step": 17653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.199729724548007e-06,
+      "loss": 0.9765,
+      "step": 17654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.19872644500493e-06,
+      "loss": 1.0106,
+      "step": 17655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.197723292829672e-06,
+      "loss": 1.0177,
+      "step": 17656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.196720268041022e-06,
+      "loss": 0.9157,
+      "step": 17657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.195717370657765e-06,
+      "loss": 1.0122,
+      "step": 17658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.194714600698675e-06,
+      "loss": 0.9426,
+      "step": 17659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.193711958182543e-06,
+      "loss": 0.9324,
+      "step": 17660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.192709443128134e-06,
+      "loss": 0.9281,
+      "step": 17661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1917070555542285e-06,
+      "loss": 0.9657,
+      "step": 17662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.190704795479598e-06,
+      "loss": 0.9849,
+      "step": 17663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.189702662923005e-06,
+      "loss": 0.976,
+      "step": 17664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.188700657903229e-06,
+      "loss": 0.984,
+      "step": 17665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.18769878043902e-06,
+      "loss": 0.9257,
+      "step": 17666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.186697030549154e-06,
+      "loss": 0.9521,
+      "step": 17667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1856954082523815e-06,
+      "loss": 1.0031,
+      "step": 17668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.184693913567464e-06,
+      "loss": 0.967,
+      "step": 17669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1836925465131545e-06,
+      "loss": 0.9641,
+      "step": 17670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.182691307108208e-06,
+      "loss": 1.0074,
+      "step": 17671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.181690195371376e-06,
+      "loss": 0.996,
+      "step": 17672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.180689211321401e-06,
+      "loss": 0.9567,
+      "step": 17673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17968835497703e-06,
+      "loss": 0.9725,
+      "step": 17674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.178687626357008e-06,
+      "loss": 1.0412,
+      "step": 17675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1776870254800765e-06,
+      "loss": 0.96,
+      "step": 17676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17668655236497e-06,
+      "loss": 0.9505,
+      "step": 17677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.175686207030431e-06,
+      "loss": 0.9591,
+      "step": 17678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.174685989495187e-06,
+      "loss": 0.9321,
+      "step": 17679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17368589977797e-06,
+      "loss": 0.9815,
+      "step": 17680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.17268593789751e-06,
+      "loss": 0.9588,
+      "step": 17681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.171686103872533e-06,
+      "loss": 0.9821,
+      "step": 17682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.170686397721766e-06,
+      "loss": 0.9724,
+      "step": 17683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.169686819463921e-06,
+      "loss": 0.9323,
+      "step": 17684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.168687369117731e-06,
+      "loss": 1.0146,
+      "step": 17685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.167688046701899e-06,
+      "loss": 0.9435,
+      "step": 17686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.166688852235152e-06,
+      "loss": 0.935,
+      "step": 17687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.165689785736192e-06,
+      "loss": 0.9971,
+      "step": 17688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.164690847223734e-06,
+      "loss": 1.0209,
+      "step": 17689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1636920367164845e-06,
+      "loss": 0.9281,
+      "step": 17690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.162693354233142e-06,
+      "loss": 1.0079,
+      "step": 17691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.161694799792422e-06,
+      "loss": 1.0016,
+      "step": 17692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1606963734130114e-06,
+      "loss": 0.9959,
+      "step": 17693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.159698075113614e-06,
+      "loss": 1.0168,
+      "step": 17694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1586999049129236e-06,
+      "loss": 0.9165,
+      "step": 17695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.157701862829635e-06,
+      "loss": 0.9259,
+      "step": 17696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.156703948882436e-06,
+      "loss": 1.0398,
+      "step": 17697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1557061630900185e-06,
+      "loss": 0.9567,
+      "step": 17698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.154708505471064e-06,
+      "loss": 0.9181,
+      "step": 17699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.153710976044254e-06,
+      "loss": 0.9159,
+      "step": 17700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.152713574828274e-06,
+      "loss": 0.9863,
+      "step": 17701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.151716301841801e-06,
+      "loss": 0.9686,
+      "step": 17702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.150719157103512e-06,
+      "loss": 0.899,
+      "step": 17703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.149722140632075e-06,
+      "loss": 0.9701,
+      "step": 17704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.14872525244617e-06,
+      "loss": 0.9309,
+      "step": 17705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.147728492564459e-06,
+      "loss": 1.018,
+      "step": 17706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1467318610056095e-06,
+      "loss": 0.9624,
+      "step": 17707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.145735357788287e-06,
+      "loss": 0.9429,
+      "step": 17708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.144738982931153e-06,
+      "loss": 0.9579,
+      "step": 17709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.143742736452869e-06,
+      "loss": 0.9888,
+      "step": 17710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.142746618372083e-06,
+      "loss": 0.9907,
+      "step": 17711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.141750628707461e-06,
+      "loss": 0.9637,
+      "step": 17712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.140754767477647e-06,
+      "loss": 0.8738,
+      "step": 17713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.139759034701292e-06,
+      "loss": 0.9316,
+      "step": 17714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.138763430397045e-06,
+      "loss": 0.9068,
+      "step": 17715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1377679545835496e-06,
+      "loss": 0.9607,
+      "step": 17716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.13677260727945e-06,
+      "loss": 0.9502,
+      "step": 17717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.135777388503382e-06,
+      "loss": 0.9771,
+      "step": 17718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.134782298273987e-06,
+      "loss": 0.9973,
+      "step": 17719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1337873366098964e-06,
+      "loss": 0.9483,
+      "step": 17720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.132792503529746e-06,
+      "loss": 1.0049,
+      "step": 17721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.131797799052165e-06,
+      "loss": 0.9906,
+      "step": 17722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.130803223195785e-06,
+      "loss": 1.0295,
+      "step": 17723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1298087759792215e-06,
+      "loss": 0.9359,
+      "step": 17724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.128814457421111e-06,
+      "loss": 0.8687,
+      "step": 17725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1278202675400625e-06,
+      "loss": 1.0352,
+      "step": 17726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1268262063547e-06,
+      "loss": 0.9047,
+      "step": 17727
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1258322738836415e-06,
+      "loss": 0.9646,
+      "step": 17728
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.12483847014549e-06,
+      "loss": 0.9545,
+      "step": 17729
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.123844795158871e-06,
+      "loss": 0.9824,
+      "step": 17730
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.122851248942383e-06,
+      "loss": 0.9661,
+      "step": 17731
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.121857831514634e-06,
+      "loss": 0.9822,
+      "step": 17732
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1208645428942295e-06,
+      "loss": 0.9777,
+      "step": 17733
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.119871383099769e-06,
+      "loss": 0.9486,
+      "step": 17734
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.118878352149854e-06,
+      "loss": 0.9842,
+      "step": 17735
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.117885450063083e-06,
+      "loss": 0.9462,
+      "step": 17736
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.116892676858043e-06,
+      "loss": 0.9506,
+      "step": 17737
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.115900032553329e-06,
+      "loss": 0.9458,
+      "step": 17738
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.114907517167531e-06,
+      "loss": 1.091,
+      "step": 17739
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.113915130719237e-06,
+      "loss": 0.9784,
+      "step": 17740
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.112922873227032e-06,
+      "loss": 0.9744,
+      "step": 17741
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.11193074470949e-06,
+      "loss": 0.9303,
+      "step": 17742
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1109387451852045e-06,
+      "loss": 0.9992,
+      "step": 17743
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.109946874672741e-06,
+      "loss": 1.0025,
+      "step": 17744
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1089551331906784e-06,
+      "loss": 0.9272,
+      "step": 17745
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.107963520757589e-06,
+      "loss": 0.9515,
+      "step": 17746
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.106972037392043e-06,
+      "loss": 0.9735,
+      "step": 17747
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1059806831126116e-06,
+      "loss": 0.9517,
+      "step": 17748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.104989457937849e-06,
+      "loss": 0.9789,
+      "step": 17749
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.103998361886333e-06,
+      "loss": 0.9987,
+      "step": 17750
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.103007394976611e-06,
+      "loss": 0.9725,
+      "step": 17751
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.102016557227247e-06,
+      "loss": 0.9841,
+      "step": 17752
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.101025848656795e-06,
+      "loss": 0.9694,
+      "step": 17753
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.10003526928381e-06,
+      "loss": 0.9378,
+      "step": 17754
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.099044819126843e-06,
+      "loss": 1.0,
+      "step": 17755
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.098054498204437e-06,
+      "loss": 0.9057,
+      "step": 17756
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.097064306535141e-06,
+      "loss": 0.9548,
+      "step": 17757
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.096074244137499e-06,
+      "loss": 1.0045,
+      "step": 17758
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.095084311030051e-06,
+      "loss": 0.9303,
+      "step": 17759
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.094094507231336e-06,
+      "loss": 0.9532,
+      "step": 17760
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.093104832759895e-06,
+      "loss": 0.9727,
+      "step": 17761
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.092115287634248e-06,
+      "loss": 0.9911,
+      "step": 17762
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.091125871872942e-06,
+      "loss": 0.9639,
+      "step": 17763
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.090136585494497e-06,
+      "loss": 0.9608,
+      "step": 17764
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.08914742851744e-06,
+      "loss": 0.959,
+      "step": 17765
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0881584009603004e-06,
+      "loss": 1.0146,
+      "step": 17766
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.087169502841588e-06,
+      "loss": 1.0216,
+      "step": 17767
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.086180734179838e-06,
+      "loss": 1.0272,
+      "step": 17768
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.08519209499355e-06,
+      "loss": 1.0407,
+      "step": 17769
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.084203585301255e-06,
+      "loss": 0.9517,
+      "step": 17770
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.083215205121453e-06,
+      "loss": 0.9886,
+      "step": 17771
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0822269544726575e-06,
+      "loss": 0.959,
+      "step": 17772
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081238833373377e-06,
+      "loss": 0.9228,
+      "step": 17773
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.080250841842113e-06,
+      "loss": 0.9741,
+      "step": 17774
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0792629798973716e-06,
+      "loss": 0.9674,
+      "step": 17775
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.078275247557648e-06,
+      "loss": 1.0198,
+      "step": 17776
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.077287644841441e-06,
+      "loss": 0.9491,
+      "step": 17777
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.076300171767247e-06,
+      "loss": 0.932,
+      "step": 17778
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0753128283535565e-06,
+      "loss": 0.9974,
+      "step": 17779
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074325614618862e-06,
+      "loss": 1.0147,
+      "step": 17780
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.073338530581652e-06,
+      "loss": 0.9561,
+      "step": 17781
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.072351576260407e-06,
+      "loss": 0.9139,
+      "step": 17782
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0713647516736106e-06,
+      "loss": 1.0414,
+      "step": 17783
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0703780568397466e-06,
+      "loss": 0.9946,
+      "step": 17784
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.06939149177729e-06,
+      "loss": 0.9558,
+      "step": 17785
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.068405056504721e-06,
+      "loss": 0.9204,
+      "step": 17786
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.067418751040502e-06,
+      "loss": 0.927,
+      "step": 17787
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.066432575403115e-06,
+      "loss": 0.9991,
+      "step": 17788
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.06544652961102e-06,
+      "loss": 0.9692,
+      "step": 17789
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.064460613682691e-06,
+      "loss": 0.957,
+      "step": 17790
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.063474827636584e-06,
+      "loss": 0.9943,
+      "step": 17791
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.062489171491163e-06,
+      "loss": 0.9065,
+      "step": 17792
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0615036452648885e-06,
+      "loss": 0.9387,
+      "step": 17793
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.060518248976211e-06,
+      "loss": 0.9779,
+      "step": 17794
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.059532982643587e-06,
+      "loss": 0.9649,
+      "step": 17795
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.058547846285468e-06,
+      "loss": 0.9579,
+      "step": 17796
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.057562839920301e-06,
+      "loss": 0.9152,
+      "step": 17797
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.056577963566534e-06,
+      "loss": 0.9751,
+      "step": 17798
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.055593217242616e-06,
+      "loss": 0.9694,
+      "step": 17799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.054608600966977e-06,
+      "loss": 0.9302,
+      "step": 17800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.053624114758063e-06,
+      "loss": 0.9161,
+      "step": 17801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.052639758634309e-06,
+      "loss": 0.9609,
+      "step": 17802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.051655532614149e-06,
+      "loss": 0.9775,
+      "step": 17803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.05067143671602e-06,
+      "loss": 0.8944,
+      "step": 17804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.049687470958338e-06,
+      "loss": 1.0071,
+      "step": 17805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.048703635359545e-06,
+      "loss": 0.9962,
+      "step": 17806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.047719929938052e-06,
+      "loss": 0.951,
+      "step": 17807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.046736354712294e-06,
+      "loss": 0.9202,
+      "step": 17808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.045752909700681e-06,
+      "loss": 0.9689,
+      "step": 17809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.044769594921633e-06,
+      "loss": 0.9445,
+      "step": 17810
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.043786410393567e-06,
+      "loss": 0.9593,
+      "step": 17811
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.042803356134886e-06,
+      "loss": 1.0257,
+      "step": 17812
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.041820432164012e-06,
+      "loss": 0.8826,
+      "step": 17813
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.040837638499344e-06,
+      "loss": 0.9674,
+      "step": 17814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.039854975159289e-06,
+      "loss": 0.9515,
+      "step": 17815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.038872442162252e-06,
+      "loss": 0.9887,
+      "step": 17816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.037890039526629e-06,
+      "loss": 0.9315,
+      "step": 17817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.036907767270819e-06,
+      "loss": 0.9311,
+      "step": 17818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.035925625413221e-06,
+      "loss": 0.9969,
+      "step": 17819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.03494361397222e-06,
+      "loss": 0.9722,
+      "step": 17820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.033961732966211e-06,
+      "loss": 1.0158,
+      "step": 17821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0329799824135794e-06,
+      "loss": 0.9873,
+      "step": 17822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.031998362332714e-06,
+      "loss": 0.9235,
+      "step": 17823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0310168727419975e-06,
+      "loss": 0.9456,
+      "step": 17824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.030035513659802e-06,
+      "loss": 0.9562,
+      "step": 17825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.029054285104519e-06,
+      "loss": 0.9958,
+      "step": 17826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.028073187094513e-06,
+      "loss": 0.9796,
+      "step": 17827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.027092219648161e-06,
+      "loss": 1.0128,
+      "step": 17828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.026111382783834e-06,
+      "loss": 1.0485,
+      "step": 17829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0251306765199e-06,
+      "loss": 0.931,
+      "step": 17830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.024150100874726e-06,
+      "loss": 1.0081,
+      "step": 17831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.023169655866669e-06,
+      "loss": 0.9952,
+      "step": 17832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.022189341514102e-06,
+      "loss": 0.8855,
+      "step": 17833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0212091578353705e-06,
+      "loss": 0.9478,
+      "step": 17834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0202291048488375e-06,
+      "loss": 0.9401,
+      "step": 17835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.019249182572854e-06,
+      "loss": 0.9932,
+      "step": 17836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.018269391025773e-06,
+      "loss": 0.9213,
+      "step": 17837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.017289730225945e-06,
+      "loss": 0.9784,
+      "step": 17838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.016310200191709e-06,
+      "loss": 0.9817,
+      "step": 17839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.015330800941416e-06,
+      "loss": 0.9947,
+      "step": 17840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.014351532493402e-06,
+      "loss": 0.9512,
+      "step": 17841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.013372394866009e-06,
+      "loss": 0.9798,
+      "step": 17842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.012393388077573e-06,
+      "loss": 0.9741,
+      "step": 17843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.011414512146429e-06,
+      "loss": 0.9194,
+      "step": 17844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.010435767090901e-06,
+      "loss": 0.9665,
+      "step": 17845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.00945715292933e-06,
+      "loss": 1.0211,
+      "step": 17846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0084786696800345e-06,
+      "loss": 0.9455,
+      "step": 17847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.00750031736134e-06,
+      "loss": 0.9451,
+      "step": 17848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.006522095991572e-06,
+      "loss": 0.9744,
+      "step": 17849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0055440055890395e-06,
+      "loss": 0.9797,
+      "step": 17850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.004566046172072e-06,
+      "loss": 0.8972,
+      "step": 17851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0035882177589735e-06,
+      "loss": 0.9887,
+      "step": 17852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.002610520368065e-06,
+      "loss": 0.9365,
+      "step": 17853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.001632954017649e-06,
+      "loss": 0.9161,
+      "step": 17854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.000655518726034e-06,
+      "loss": 0.9238,
+      "step": 17855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.999678214511526e-06,
+      "loss": 0.9802,
+      "step": 17856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.998701041392426e-06,
+      "loss": 0.9876,
+      "step": 17857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.997723999387036e-06,
+      "loss": 0.9279,
+      "step": 17858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.996747088513648e-06,
+      "loss": 0.983,
+      "step": 17859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.99577030879056e-06,
+      "loss": 0.9773,
+      "step": 17860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.994793660236064e-06,
+      "loss": 0.9742,
+      "step": 17861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.993817142868452e-06,
+      "loss": 0.9863,
+      "step": 17862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.992840756706003e-06,
+      "loss": 0.9616,
+      "step": 17863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.991864501767013e-06,
+      "loss": 0.9236,
+      "step": 17864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.9908883780697565e-06,
+      "loss": 0.9237,
+      "step": 17865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.989912385632516e-06,
+      "loss": 1.0054,
+      "step": 17866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.988936524473567e-06,
+      "loss": 0.9913,
+      "step": 17867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.987960794611189e-06,
+      "loss": 0.884,
+      "step": 17868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.986985196063654e-06,
+      "loss": 0.9668,
+      "step": 17869
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.986009728849224e-06,
+      "loss": 0.9074,
+      "step": 17870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.985034392986179e-06,
+      "loss": 1.0034,
+      "step": 17871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.984059188492774e-06,
+      "loss": 0.997,
+      "step": 17872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.983084115387276e-06,
+      "loss": 1.0105,
+      "step": 17873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.982109173687945e-06,
+      "loss": 1.0207,
+      "step": 17874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.981134363413037e-06,
+      "loss": 0.9236,
+      "step": 17875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.980159684580813e-06,
+      "loss": 0.915,
+      "step": 17876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.979185137209519e-06,
+      "loss": 0.9702,
+      "step": 17877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.978210721317408e-06,
+      "loss": 1.0198,
+      "step": 17878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.977236436922729e-06,
+      "loss": 0.9704,
+      "step": 17879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.976262284043725e-06,
+      "loss": 0.9837,
+      "step": 17880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.975288262698641e-06,
+      "loss": 0.9662,
+      "step": 17881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9743143729057205e-06,
+      "loss": 0.9809,
+      "step": 17882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.973340614683192e-06,
+      "loss": 0.9793,
+      "step": 17883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.972366988049303e-06,
+      "loss": 0.9884,
+      "step": 17884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.971393493022278e-06,
+      "loss": 0.9494,
+      "step": 17885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.970420129620351e-06,
+      "loss": 0.8879,
+      "step": 17886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.969446897861754e-06,
+      "loss": 0.9097,
+      "step": 17887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.968473797764701e-06,
+      "loss": 0.9613,
+      "step": 17888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.967500829347429e-06,
+      "loss": 0.9045,
+      "step": 17889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9665279926281475e-06,
+      "loss": 0.9581,
+      "step": 17890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.965555287625086e-06,
+      "loss": 0.9833,
+      "step": 17891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9645827143564514e-06,
+      "loss": 0.9406,
+      "step": 17892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.963610272840461e-06,
+      "loss": 0.894,
+      "step": 17893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9626379630953246e-06,
+      "loss": 0.9795,
+      "step": 17894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9616657851392515e-06,
+      "loss": 0.9984,
+      "step": 17895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.96069373899045e-06,
+      "loss": 0.9263,
+      "step": 17896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.959721824667118e-06,
+      "loss": 1.0212,
+      "step": 17897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.958750042187459e-06,
+      "loss": 0.9377,
+      "step": 17898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.957778391569673e-06,
+      "loss": 0.9533,
+      "step": 17899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.956806872831956e-06,
+      "loss": 1.0552,
+      "step": 17900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.955835485992501e-06,
+      "loss": 0.9587,
+      "step": 17901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.954864231069502e-06,
+      "loss": 0.9903,
+      "step": 17902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.953893108081142e-06,
+      "loss": 0.871,
+      "step": 17903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9529221170456095e-06,
+      "loss": 0.9792,
+      "step": 17904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.951951257981091e-06,
+      "loss": 0.9637,
+      "step": 17905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9509805309057645e-06,
+      "loss": 0.9584,
+      "step": 17906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.950009935837814e-06,
+      "loss": 0.9677,
+      "step": 17907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.949039472795407e-06,
+      "loss": 0.9416,
+      "step": 17908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.948069141796727e-06,
+      "loss": 0.9831,
+      "step": 17909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.947098942859934e-06,
+      "loss": 1.0379,
+      "step": 17910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.946128876003214e-06,
+      "loss": 0.966,
+      "step": 17911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.945158941244718e-06,
+      "loss": 0.9598,
+      "step": 17912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.944189138602616e-06,
+      "loss": 1.0364,
+      "step": 17913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.943219468095072e-06,
+      "loss": 0.9767,
+      "step": 17914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9422499297402365e-06,
+      "loss": 0.9741,
+      "step": 17915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.941280523556277e-06,
+      "loss": 0.9926,
+      "step": 17916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.940311249561339e-06,
+      "loss": 0.9593,
+      "step": 17917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.939342107773579e-06,
+      "loss": 0.8794,
+      "step": 17918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.938373098211145e-06,
+      "loss": 1.0199,
+      "step": 17919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.937404220892183e-06,
+      "loss": 0.9766,
+      "step": 17920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.936435475834837e-06,
+      "loss": 0.9447,
+      "step": 17921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.935466863057253e-06,
+      "loss": 0.9379,
+      "step": 17922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9344983825775615e-06,
+      "loss": 0.9519,
+      "step": 17923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.933530034413907e-06,
+      "loss": 0.9815,
+      "step": 17924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.932561818584421e-06,
+      "loss": 0.9623,
+      "step": 17925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9315937351072345e-06,
+      "loss": 0.9814,
+      "step": 17926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.930625784000481e-06,
+      "loss": 0.9639,
+      "step": 17927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.929657965282279e-06,
+      "loss": 0.9836,
+      "step": 17928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.928690278970764e-06,
+      "loss": 0.9655,
+      "step": 17929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.927722725084049e-06,
+      "loss": 0.953,
+      "step": 17930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.926755303640257e-06,
+      "loss": 1.0053,
+      "step": 17931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.925788014657504e-06,
+      "loss": 0.9787,
+      "step": 17932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.924820858153905e-06,
+      "loss": 0.9661,
+      "step": 17933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.923853834147576e-06,
+      "loss": 1.0038,
+      "step": 17934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.922886942656618e-06,
+      "loss": 0.98,
+      "step": 17935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.921920183699143e-06,
+      "loss": 0.9448,
+      "step": 17936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.920953557293255e-06,
+      "loss": 0.9602,
+      "step": 17937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.919987063457055e-06,
+      "loss": 0.9582,
+      "step": 17938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.919020702208646e-06,
+      "loss": 0.9264,
+      "step": 17939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9180544735661245e-06,
+      "loss": 0.9822,
+      "step": 17940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.91708837754758e-06,
+      "loss": 0.9628,
+      "step": 17941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.916122414171109e-06,
+      "loss": 0.9729,
+      "step": 17942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9151565834548e-06,
+      "loss": 0.9627,
+      "step": 17943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.91419088541674e-06,
+      "loss": 0.8893,
+      "step": 17944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.913225320075017e-06,
+      "loss": 0.9007,
+      "step": 17945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.912259887447705e-06,
+      "loss": 0.9588,
+      "step": 17946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.911294587552895e-06,
+      "loss": 0.9732,
+      "step": 17947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.910329420408651e-06,
+      "loss": 0.9785,
+      "step": 17948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.909364386033061e-06,
+      "loss": 0.9721,
+      "step": 17949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.908399484444188e-06,
+      "loss": 1.0151,
+      "step": 17950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.907434715660105e-06,
+      "loss": 0.9732,
+      "step": 17951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.906470079698883e-06,
+      "loss": 0.9316,
+      "step": 17952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.905505576578577e-06,
+      "loss": 1.0064,
+      "step": 17953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.904541206317262e-06,
+      "loss": 0.9675,
+      "step": 17954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.903576968932987e-06,
+      "loss": 0.9396,
+      "step": 17955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9026128644438134e-06,
+      "loss": 0.9247,
+      "step": 17956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.901648892867796e-06,
+      "loss": 0.9975,
+      "step": 17957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.900685054222987e-06,
+      "loss": 0.8776,
+      "step": 17958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.899721348527436e-06,
+      "loss": 0.9434,
+      "step": 17959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.898757775799196e-06,
+      "loss": 0.9277,
+      "step": 17960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.897794336056302e-06,
+      "loss": 0.9824,
+      "step": 17961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.896831029316802e-06,
+      "loss": 0.9951,
+      "step": 17962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.895867855598734e-06,
+      "loss": 0.9739,
+      "step": 17963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.894904814920138e-06,
+      "loss": 0.9885,
+      "step": 17964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8939419072990496e-06,
+      "loss": 0.8963,
+      "step": 17965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.892979132753491e-06,
+      "loss": 0.9855,
+      "step": 17966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.892016491301509e-06,
+      "loss": 0.9522,
+      "step": 17967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.891053982961117e-06,
+      "loss": 0.942,
+      "step": 17968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8900916077503455e-06,
+      "loss": 0.9341,
+      "step": 17969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.889129365687221e-06,
+      "loss": 0.9354,
+      "step": 17970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.888167256789752e-06,
+      "loss": 0.9353,
+      "step": 17971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.887205281075969e-06,
+      "loss": 0.9286,
+      "step": 17972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.886243438563875e-06,
+      "loss": 1.0416,
+      "step": 17973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.885281729271494e-06,
+      "loss": 0.9777,
+      "step": 17974
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8843201532168274e-06,
+      "loss": 0.9001,
+      "step": 17975
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.883358710417886e-06,
+      "loss": 0.9562,
+      "step": 17976
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.882397400892675e-06,
+      "loss": 1.0022,
+      "step": 17977
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8814362246591954e-06,
+      "loss": 0.9188,
+      "step": 17978
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.880475181735451e-06,
+      "loss": 0.9354,
+      "step": 17979
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.879514272139434e-06,
+      "loss": 0.9789,
+      "step": 17980
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.878553495889143e-06,
+      "loss": 0.8923,
+      "step": 17981
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.877592853002568e-06,
+      "loss": 0.9628,
+      "step": 17982
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.876632343497701e-06,
+      "loss": 0.9777,
+      "step": 17983
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.875671967392528e-06,
+      "loss": 0.9864,
+      "step": 17984
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.874711724705038e-06,
+      "loss": 0.9236,
+      "step": 17985
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8737516154532085e-06,
+      "loss": 1.0891,
+      "step": 17986
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87279163965502e-06,
+      "loss": 0.9146,
+      "step": 17987
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.871831797328452e-06,
+      "loss": 1.0161,
+      "step": 17988
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.87087208849148e-06,
+      "loss": 1.0025,
+      "step": 17989
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.869912513162077e-06,
+      "loss": 0.9779,
+      "step": 17990
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.868953071358207e-06,
+      "loss": 0.9727,
+      "step": 17991
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.867993763097848e-06,
+      "loss": 0.9678,
+      "step": 17992
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.867034588398952e-06,
+      "loss": 0.9041,
+      "step": 17993
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.866075547279496e-06,
+      "loss": 0.9959,
+      "step": 17994
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.865116639757428e-06,
+      "loss": 0.9725,
+      "step": 17995
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.864157865850711e-06,
+      "loss": 0.9822,
+      "step": 17996
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.863199225577302e-06,
+      "loss": 0.927,
+      "step": 17997
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.862240718955147e-06,
+      "loss": 0.9472,
+      "step": 17998
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.861282346002199e-06,
+      "loss": 0.9774,
+      "step": 17999
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.860324106736408e-06,
+      "loss": 0.9319,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "eval_text_loss": 0.9330910444259644,
+      "eval_text_runtime": 1.5461,
+      "eval_text_samples_per_second": 47.863,
+      "eval_text_steps_per_second": 1.294,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "eval_svit_loss": 1.0810155868530273,
+      "eval_svit_runtime": 4.6316,
+      "eval_svit_samples_per_second": 147.034,
+      "eval_svit_steps_per_second": 2.375,
+      "step": 18000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.859366001175717e-06,
+      "loss": 0.95,
+      "step": 18001
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.858408029338068e-06,
+      "loss": 0.9105,
+      "step": 18002
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.857450191241405e-06,
+      "loss": 0.9443,
+      "step": 18003
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8564924869036555e-06,
+      "loss": 0.9593,
+      "step": 18004
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8555349163427675e-06,
+      "loss": 0.9226,
+      "step": 18005
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8545774795766645e-06,
+      "loss": 0.9002,
+      "step": 18006
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.853620176623278e-06,
+      "loss": 1.0148,
+      "step": 18007
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.85266300750054e-06,
+      "loss": 0.9957,
+      "step": 18008
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.851705972226367e-06,
+      "loss": 0.91,
+      "step": 18009
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8507490708186916e-06,
+      "loss": 0.9435,
+      "step": 18010
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8497923032954225e-06,
+      "loss": 0.9181,
+      "step": 18011
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.848835669674489e-06,
+      "loss": 1.0096,
+      "step": 18012
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.847879169973797e-06,
+      "loss": 0.9173,
+      "step": 18013
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.846922804211262e-06,
+      "loss": 0.9617,
+      "step": 18014
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.845966572404795e-06,
+      "loss": 1.0066,
+      "step": 18015
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.845010474572302e-06,
+      "loss": 0.9842,
+      "step": 18016
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.844054510731692e-06,
+      "loss": 0.9667,
+      "step": 18017
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.84309868090086e-06,
+      "loss": 0.9962,
+      "step": 18018
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.842142985097709e-06,
+      "loss": 0.9897,
+      "step": 18019
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.841187423340138e-06,
+      "loss": 0.9104,
+      "step": 18020
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.840231995646041e-06,
+      "loss": 0.9836,
+      "step": 18021
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8392767020333095e-06,
+      "loss": 0.9308,
+      "step": 18022
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.838321542519837e-06,
+      "loss": 0.9082,
+      "step": 18023
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.837366517123505e-06,
+      "loss": 0.9997,
+      "step": 18024
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8364116258622026e-06,
+      "loss": 0.9254,
+      "step": 18025
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.83545686875381e-06,
+      "loss": 0.9256,
+      "step": 18026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8345022458162065e-06,
+      "loss": 0.9837,
+      "step": 18027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.833547757067275e-06,
+      "loss": 0.9455,
+      "step": 18028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.832593402524878e-06,
+      "loss": 0.9083,
+      "step": 18029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8316391822069035e-06,
+      "loss": 1.008,
+      "step": 18030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.830685096131206e-06,
+      "loss": 0.8901,
+      "step": 18031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.829731144315666e-06,
+      "loss": 0.9658,
+      "step": 18032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.828777326778139e-06,
+      "loss": 0.9711,
+      "step": 18033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.827823643536491e-06,
+      "loss": 0.9494,
+      "step": 18034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.826870094608585e-06,
+      "loss": 0.9764,
+      "step": 18035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.825916680012268e-06,
+      "loss": 0.9563,
+      "step": 18036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8249633997654055e-06,
+      "loss": 1.0259,
+      "step": 18037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.824010253885843e-06,
+      "loss": 1.026,
+      "step": 18038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.823057242391431e-06,
+      "loss": 0.8618,
+      "step": 18039
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8221043653000195e-06,
+      "loss": 0.8947,
+      "step": 18040
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.821151622629451e-06,
+      "loss": 1.0361,
+      "step": 18041
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.820199014397567e-06,
+      "loss": 0.9699,
+      "step": 18042
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.819246540622211e-06,
+      "loss": 0.9652,
+      "step": 18043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.818294201321215e-06,
+      "loss": 0.9914,
+      "step": 18044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.817341996512415e-06,
+      "loss": 0.9182,
+      "step": 18045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.816389926213643e-06,
+      "loss": 0.9782,
+      "step": 18046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8154379904427285e-06,
+      "loss": 0.9463,
+      "step": 18047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.814486189217504e-06,
+      "loss": 0.9825,
+      "step": 18048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.813534522555781e-06,
+      "loss": 0.9923,
+      "step": 18049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.812582990475395e-06,
+      "loss": 0.9482,
+      "step": 18050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8116315929941575e-06,
+      "loss": 0.9536,
+      "step": 18051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.810680330129888e-06,
+      "loss": 0.9994,
+      "step": 18052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8097292019004e-06,
+      "loss": 0.9503,
+      "step": 18053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.808778208323505e-06,
+      "loss": 1.0474,
+      "step": 18054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8078273494170154e-06,
+      "loss": 0.9459,
+      "step": 18055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.806876625198731e-06,
+      "loss": 0.9367,
+      "step": 18056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.805926035686466e-06,
+      "loss": 0.9363,
+      "step": 18057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804975580898014e-06,
+      "loss": 0.9794,
+      "step": 18058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.804025260851177e-06,
+      "loss": 0.9541,
+      "step": 18059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.803075075563751e-06,
+      "loss": 0.9872,
+      "step": 18060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.802125025053533e-06,
+      "loss": 0.9723,
+      "step": 18061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8011751093383135e-06,
+      "loss": 0.9665,
+      "step": 18062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.800225328435879e-06,
+      "loss": 0.9624,
+      "step": 18063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.799275682364016e-06,
+      "loss": 0.9268,
+      "step": 18064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.798326171140512e-06,
+      "loss": 0.9767,
+      "step": 18065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.797376794783146e-06,
+      "loss": 0.9703,
+      "step": 18066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.796427553309697e-06,
+      "loss": 0.9957,
+      "step": 18067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.795478446737947e-06,
+      "loss": 0.9757,
+      "step": 18068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.794529475085659e-06,
+      "loss": 0.9276,
+      "step": 18069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.793580638370616e-06,
+      "loss": 0.9202,
+      "step": 18070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.79263193661058e-06,
+      "loss": 0.9362,
+      "step": 18071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.791683369823318e-06,
+      "loss": 0.9523,
+      "step": 18072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.790734938026599e-06,
+      "loss": 0.9934,
+      "step": 18073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.789786641238173e-06,
+      "loss": 0.9627,
+      "step": 18074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.788838479475813e-06,
+      "loss": 0.9683,
+      "step": 18075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.787890452757265e-06,
+      "loss": 1.0246,
+      "step": 18076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7869425611002866e-06,
+      "loss": 0.9568,
+      "step": 18077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7859948045226284e-06,
+      "loss": 0.9374,
+      "step": 18078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.785047183042039e-06,
+      "loss": 0.9456,
+      "step": 18079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.784099696676265e-06,
+      "loss": 1.0517,
+      "step": 18080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.783152345443053e-06,
+      "loss": 0.9842,
+      "step": 18081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.782205129360138e-06,
+      "loss": 0.9301,
+      "step": 18082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.781258048445261e-06,
+      "loss": 0.9379,
+      "step": 18083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.780311102716159e-06,
+      "loss": 0.9343,
+      "step": 18084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.779364292190564e-06,
+      "loss": 0.9956,
+      "step": 18085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.778417616886212e-06,
+      "loss": 0.9626,
+      "step": 18086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.777471076820822e-06,
+      "loss": 0.9387,
+      "step": 18087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.77652467201213e-06,
+      "loss": 0.981,
+      "step": 18088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.775578402477852e-06,
+      "loss": 1.018,
+      "step": 18089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.774632268235711e-06,
+      "loss": 0.9033,
+      "step": 18090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.773686269303428e-06,
+      "loss": 0.9753,
+      "step": 18091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.772740405698715e-06,
+      "loss": 0.9522,
+      "step": 18092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7717946774392915e-06,
+      "loss": 0.9518,
+      "step": 18093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.770849084542856e-06,
+      "loss": 0.9654,
+      "step": 18094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.76990362702713e-06,
+      "loss": 0.9238,
+      "step": 18095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.768958304909811e-06,
+      "loss": 1.0181,
+      "step": 18096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.768013118208604e-06,
+      "loss": 0.9466,
+      "step": 18097
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.767068066941212e-06,
+      "loss": 1.01,
+      "step": 18098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.766123151125329e-06,
+      "loss": 0.9504,
+      "step": 18099
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.765178370778656e-06,
+      "loss": 0.9851,
+      "step": 18100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.76423372591888e-06,
+      "loss": 0.9484,
+      "step": 18101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.763289216563694e-06,
+      "loss": 0.9236,
+      "step": 18102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.762344842730785e-06,
+      "loss": 0.9936,
+      "step": 18103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.7614006044378406e-06,
+      "loss": 0.9243,
+      "step": 18104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.76045650170254e-06,
+      "loss": 0.9632,
+      "step": 18105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.759512534542571e-06,
+      "loss": 0.9516,
+      "step": 18106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.758568702975599e-06,
+      "loss": 0.9543,
+      "step": 18107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.757625007019312e-06,
+      "loss": 0.9076,
+      "step": 18108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.756681446691374e-06,
+      "loss": 0.9706,
+      "step": 18109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.755738022009458e-06,
+      "loss": 0.9579,
+      "step": 18110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.754794732991234e-06,
+      "loss": 1.0131,
+      "step": 18111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7538515796543585e-06,
+      "loss": 0.9742,
+      "step": 18112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.752908562016506e-06,
+      "loss": 0.9928,
+      "step": 18113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7519656800953255e-06,
+      "loss": 1.0433,
+      "step": 18114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.751022933908484e-06,
+      "loss": 0.9264,
+      "step": 18115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.75008032347363e-06,
+      "loss": 0.9398,
+      "step": 18116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.749137848808416e-06,
+      "loss": 0.948,
+      "step": 18117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.748195509930493e-06,
+      "loss": 0.8826,
+      "step": 18118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.74725330685751e-06,
+      "loss": 0.9097,
+      "step": 18119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.746311239607112e-06,
+      "loss": 0.9522,
+      "step": 18120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.745369308196938e-06,
+      "loss": 0.9937,
+      "step": 18121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.744427512644627e-06,
+      "loss": 0.9522,
+      "step": 18122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7434858529678185e-06,
+      "loss": 0.9504,
+      "step": 18123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.742544329184147e-06,
+      "loss": 0.9408,
+      "step": 18124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.741602941311244e-06,
+      "loss": 0.8956,
+      "step": 18125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.740661689366741e-06,
+      "loss": 0.9968,
+      "step": 18126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.739720573368261e-06,
+      "loss": 0.9647,
+      "step": 18127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.738779593333429e-06,
+      "loss": 1.0172,
+      "step": 18128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.737838749279869e-06,
+      "loss": 1.0508,
+      "step": 18129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7368980412251984e-06,
+      "loss": 1.0281,
+      "step": 18130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.735957469187038e-06,
+      "loss": 0.9833,
+      "step": 18131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.735017033182991e-06,
+      "loss": 0.9325,
+      "step": 18132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.734076733230685e-06,
+      "loss": 0.9228,
+      "step": 18133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7331365693477126e-06,
+      "loss": 0.9546,
+      "step": 18134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7321965415516965e-06,
+      "loss": 0.9951,
+      "step": 18135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.731256649860228e-06,
+      "loss": 1.0219,
+      "step": 18136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.730316894290912e-06,
+      "loss": 0.9592,
+      "step": 18137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.729377274861352e-06,
+      "loss": 0.8487,
+      "step": 18138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.728437791589139e-06,
+      "loss": 1.0284,
+      "step": 18139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.727498444491867e-06,
+      "loss": 0.9574,
+      "step": 18140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.726559233587128e-06,
+      "loss": 1.0194,
+      "step": 18141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7256201588925125e-06,
+      "loss": 0.9277,
+      "step": 18142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.724681220425605e-06,
+      "loss": 1.0126,
+      "step": 18143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.723742418203993e-06,
+      "loss": 0.9468,
+      "step": 18144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7228037522452465e-06,
+      "loss": 1.0563,
+      "step": 18145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.721865222566959e-06,
+      "loss": 1.0135,
+      "step": 18146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7209268291866935e-06,
+      "loss": 0.9795,
+      "step": 18147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.71998857212203e-06,
+      "loss": 0.954,
+      "step": 18148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.719050451390541e-06,
+      "loss": 0.9724,
+      "step": 18149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.718112467009785e-06,
+      "loss": 0.9641,
+      "step": 18150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.71717461899734e-06,
+      "loss": 0.991,
+      "step": 18151
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.716236907370758e-06,
+      "loss": 0.9611,
+      "step": 18152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.715299332147612e-06,
+      "loss": 0.9673,
+      "step": 18153
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.714361893345448e-06,
+      "loss": 1.0175,
+      "step": 18154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.713424590981828e-06,
+      "loss": 0.9301,
+      "step": 18155
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.712487425074307e-06,
+      "loss": 0.9255,
+      "step": 18156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7115503956404235e-06,
+      "loss": 0.899,
+      "step": 18157
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7106135026977395e-06,
+      "loss": 0.9973,
+      "step": 18158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.709676746263793e-06,
+      "loss": 0.8814,
+      "step": 18159
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.708740126356127e-06,
+      "loss": 0.9398,
+      "step": 18160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.707803642992284e-06,
+      "loss": 0.9713,
+      "step": 18161
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.706867296189799e-06,
+      "loss": 0.9106,
+      "step": 18162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.705931085966208e-06,
+      "loss": 1.0253,
+      "step": 18163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.704995012339048e-06,
+      "loss": 0.913,
+      "step": 18164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7040590753258395e-06,
+      "loss": 1.0057,
+      "step": 18165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.703123274944116e-06,
+      "loss": 0.9162,
+      "step": 18166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.702187611211402e-06,
+      "loss": 0.9797,
+      "step": 18167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.701252084145218e-06,
+      "loss": 0.8754,
+      "step": 18168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.700316693763088e-06,
+      "loss": 0.9723,
+      "step": 18169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6993814400825195e-06,
+      "loss": 0.9229,
+      "step": 18170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.698446323121039e-06,
+      "loss": 0.98,
+      "step": 18171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.697511342896149e-06,
+      "loss": 0.9739,
+      "step": 18172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.696576499425365e-06,
+      "loss": 0.9847,
+      "step": 18173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6956417927261905e-06,
+      "loss": 0.9335,
+      "step": 18174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.69470722281613e-06,
+      "loss": 0.9164,
+      "step": 18175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.69377278971269e-06,
+      "loss": 0.9387,
+      "step": 18176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.692838493433359e-06,
+      "loss": 0.9676,
+      "step": 18177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.691904333995646e-06,
+      "loss": 0.9948,
+      "step": 18178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6909703114170376e-06,
+      "loss": 0.9794,
+      "step": 18179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6900364257150266e-06,
+      "loss": 0.925,
+      "step": 18180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6891026769071025e-06,
+      "loss": 0.9692,
+      "step": 18181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6881690650107516e-06,
+      "loss": 0.9524,
+      "step": 18182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.68723559004346e-06,
+      "loss": 0.992,
+      "step": 18183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.686302252022704e-06,
+      "loss": 0.9841,
+      "step": 18184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.685369050965965e-06,
+      "loss": 0.9428,
+      "step": 18185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.68443598689072e-06,
+      "loss": 0.9083,
+      "step": 18186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.683503059814441e-06,
+      "loss": 0.9335,
+      "step": 18187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.682570269754599e-06,
+      "loss": 0.9376,
+      "step": 18188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6816376167286655e-06,
+      "loss": 0.9725,
+      "step": 18189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.680705100754099e-06,
+      "loss": 0.9962,
+      "step": 18190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.679772721848373e-06,
+      "loss": 0.9619,
+      "step": 18191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.67884048002894e-06,
+      "loss": 1.0001,
+      "step": 18192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.677908375313262e-06,
+      "loss": 0.9632,
+      "step": 18193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6769764077187955e-06,
+      "loss": 0.9837,
+      "step": 18194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.676044577262985e-06,
+      "loss": 0.9275,
+      "step": 18195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.675112883963294e-06,
+      "loss": 0.9139,
+      "step": 18196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.674181327837158e-06,
+      "loss": 0.9403,
+      "step": 18197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.673249908902032e-06,
+      "loss": 0.9793,
+      "step": 18198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6723186271753545e-06,
+      "loss": 0.9726,
+      "step": 18199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.671387482674566e-06,
+      "loss": 0.977,
+      "step": 18200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.670456475417103e-06,
+      "loss": 0.9411,
+      "step": 18201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.669525605420401e-06,
+      "loss": 0.9528,
+      "step": 18202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.668594872701898e-06,
+      "loss": 0.9996,
+      "step": 18203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6676642772790145e-06,
+      "loss": 0.922,
+      "step": 18204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.666733819169183e-06,
+      "loss": 0.952,
+      "step": 18205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6658034983898275e-06,
+      "loss": 0.9599,
+      "step": 18206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6648733149583724e-06,
+      "loss": 0.9859,
+      "step": 18207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.663943268892229e-06,
+      "loss": 0.9664,
+      "step": 18208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.663013360208828e-06,
+      "loss": 0.979,
+      "step": 18209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.662083588925572e-06,
+      "loss": 0.9904,
+      "step": 18210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.661153955059877e-06,
+      "loss": 0.9897,
+      "step": 18211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6602244586291535e-06,
+      "loss": 0.9325,
+      "step": 18212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6592950996508065e-06,
+      "loss": 0.9639,
+      "step": 18213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.658365878142243e-06,
+      "loss": 0.9421,
+      "step": 18214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.657436794120858e-06,
+      "loss": 0.9689,
+      "step": 18215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.65650784760406e-06,
+      "loss": 0.9841,
+      "step": 18216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.655579038609238e-06,
+      "loss": 0.9116,
+      "step": 18217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.654650367153788e-06,
+      "loss": 0.9432,
+      "step": 18218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.653721833255102e-06,
+      "loss": 0.9344,
+      "step": 18219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6527934369305666e-06,
+      "loss": 0.9375,
+      "step": 18220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.651865178197573e-06,
+      "loss": 1.003,
+      "step": 18221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6509370570734995e-06,
+      "loss": 1.0139,
+      "step": 18222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.650009073575727e-06,
+      "loss": 0.969,
+      "step": 18223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.649081227721638e-06,
+      "loss": 1.0327,
+      "step": 18224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.648153519528604e-06,
+      "loss": 0.9936,
+      "step": 18225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.647225949014002e-06,
+      "loss": 0.9282,
+      "step": 18226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.646298516195203e-06,
+      "loss": 1.0049,
+      "step": 18227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.645371221089567e-06,
+      "loss": 0.9623,
+      "step": 18228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.644444063714472e-06,
+      "loss": 0.954,
+      "step": 18229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.643517044087272e-06,
+      "loss": 0.942,
+      "step": 18230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6425901622253285e-06,
+      "loss": 0.9711,
+      "step": 18231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.641663418146004e-06,
+      "loss": 0.9707,
+      "step": 18232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.640736811866644e-06,
+      "loss": 0.923,
+      "step": 18233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.639810343404614e-06,
+      "loss": 0.9984,
+      "step": 18234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.638884012777252e-06,
+      "loss": 0.9214,
+      "step": 18235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.637957820001916e-06,
+      "loss": 0.9793,
+      "step": 18236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.637031765095943e-06,
+      "loss": 1.015,
+      "step": 18237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.636105848076678e-06,
+      "loss": 0.9589,
+      "step": 18238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.63518006896146e-06,
+      "loss": 0.9139,
+      "step": 18239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6342544277676285e-06,
+      "loss": 0.9746,
+      "step": 18240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.633328924512518e-06,
+      "loss": 0.9545,
+      "step": 18241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6324035592134555e-06,
+      "loss": 0.9527,
+      "step": 18242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.631478331887774e-06,
+      "loss": 0.9593,
+      "step": 18243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.630553242552801e-06,
+      "loss": 0.9594,
+      "step": 18244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6296282912258596e-06,
+      "loss": 0.9406,
+      "step": 18245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.62870347792427e-06,
+      "loss": 0.9133,
+      "step": 18246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.627778802665357e-06,
+      "loss": 0.9354,
+      "step": 18247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.62685426546643e-06,
+      "loss": 0.946,
+      "step": 18248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625929866344804e-06,
+      "loss": 0.9244,
+      "step": 18249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.625005605317793e-06,
+      "loss": 0.9603,
+      "step": 18250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.624081482402704e-06,
+      "loss": 0.9646,
+      "step": 18251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.623157497616847e-06,
+      "loss": 0.9988,
+      "step": 18252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.622233650977515e-06,
+      "loss": 0.9572,
+      "step": 18253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6213099425020245e-06,
+      "loss": 0.9643,
+      "step": 18254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6203863722076586e-06,
+      "loss": 1.0116,
+      "step": 18255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6194629401117256e-06,
+      "loss": 0.9469,
+      "step": 18256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.618539646231509e-06,
+      "loss": 0.9334,
+      "step": 18257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.617616490584305e-06,
+      "loss": 1.0376,
+      "step": 18258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6166934731874015e-06,
+      "loss": 0.9326,
+      "step": 18259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.615770594058077e-06,
+      "loss": 0.9204,
+      "step": 18260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.614847853213625e-06,
+      "loss": 0.908,
+      "step": 18261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613925250671318e-06,
+      "loss": 0.8989,
+      "step": 18262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6130027864484365e-06,
+      "loss": 1.006,
+      "step": 18263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.612080460562254e-06,
+      "loss": 0.941,
+      "step": 18264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.611158273030045e-06,
+      "loss": 1.0327,
+      "step": 18265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.610236223869077e-06,
+      "loss": 0.9689,
+      "step": 18266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6093143130966225e-06,
+      "loss": 0.9196,
+      "step": 18267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.608392540729939e-06,
+      "loss": 0.9041,
+      "step": 18268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.607470906786291e-06,
+      "loss": 1.0103,
+      "step": 18269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.606549411282941e-06,
+      "loss": 1.0129,
+      "step": 18270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.605628054237141e-06,
+      "loss": 0.966,
+      "step": 18271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.604706835666153e-06,
+      "loss": 0.9476,
+      "step": 18272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.6037857555872166e-06,
+      "loss": 1.0091,
+      "step": 18273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.602864814017595e-06,
+      "loss": 0.9612,
+      "step": 18274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.601944010974524e-06,
+      "loss": 0.9929,
+      "step": 18275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.601023346475252e-06,
+      "loss": 0.8992,
+      "step": 18276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.600102820537019e-06,
+      "loss": 0.9626,
+      "step": 18277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.599182433177065e-06,
+      "loss": 0.957,
+      "step": 18278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5982621844126276e-06,
+      "loss": 0.9917,
+      "step": 18279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.597342074260936e-06,
+      "loss": 0.9659,
+      "step": 18280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.596422102739223e-06,
+      "loss": 0.9034,
+      "step": 18281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.595502269864718e-06,
+      "loss": 0.9364,
+      "step": 18282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5945825756546455e-06,
+      "loss": 0.9265,
+      "step": 18283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.593663020126229e-06,
+      "loss": 0.99,
+      "step": 18284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.592743603296694e-06,
+      "loss": 0.9799,
+      "step": 18285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5918243251832495e-06,
+      "loss": 0.9535,
+      "step": 18286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.590905185803115e-06,
+      "loss": 0.949,
+      "step": 18287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.589986185173504e-06,
+      "loss": 1.0039,
+      "step": 18288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.589067323311626e-06,
+      "loss": 0.8641,
+      "step": 18289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.588148600234693e-06,
+      "loss": 0.9154,
+      "step": 18290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.587230015959898e-06,
+      "loss": 0.9519,
+      "step": 18291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.586311570504457e-06,
+      "loss": 0.9339,
+      "step": 18292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5853932638855585e-06,
+      "loss": 0.9596,
+      "step": 18293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.584475096120411e-06,
+      "loss": 0.9133,
+      "step": 18294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5835570672262e-06,
+      "loss": 0.9511,
+      "step": 18295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.58263917722012e-06,
+      "loss": 0.9081,
+      "step": 18296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.581721426119366e-06,
+      "loss": 1.0138,
+      "step": 18297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.580803813941112e-06,
+      "loss": 0.9579,
+      "step": 18298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.579886340702556e-06,
+      "loss": 0.9512,
+      "step": 18299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.578969006420872e-06,
+      "loss": 0.942,
+      "step": 18300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.578051811113239e-06,
+      "loss": 0.9788,
+      "step": 18301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.577134754796835e-06,
+      "loss": 0.9336,
+      "step": 18302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5762178374888334e-06,
+      "loss": 1.0091,
+      "step": 18303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5753010592064064e-06,
+      "loss": 1.008,
+      "step": 18304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.574384419966724e-06,
+      "loss": 1.0036,
+      "step": 18305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5734679197869475e-06,
+      "loss": 0.9797,
+      "step": 18306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5725515586842416e-06,
+      "loss": 0.9871,
+      "step": 18307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.571635336675769e-06,
+      "loss": 0.9888,
+      "step": 18308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.570719253778686e-06,
+      "loss": 0.9541,
+      "step": 18309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.569803310010153e-06,
+      "loss": 0.9169,
+      "step": 18310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.568887505387311e-06,
+      "loss": 1.0067,
+      "step": 18311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.567971839927327e-06,
+      "loss": 0.9382,
+      "step": 18312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.567056313647336e-06,
+      "loss": 0.9847,
+      "step": 18313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.566140926564486e-06,
+      "loss": 0.9498,
+      "step": 18314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.565225678695923e-06,
+      "loss": 0.9568,
+      "step": 18315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.564310570058782e-06,
+      "loss": 0.9625,
+      "step": 18316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5633956006702075e-06,
+      "loss": 0.9723,
+      "step": 18317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.562480770547323e-06,
+      "loss": 0.9212,
+      "step": 18318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.561566079707273e-06,
+      "loss": 0.9334,
+      "step": 18319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.560651528167177e-06,
+      "loss": 0.9385,
+      "step": 18320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.559737115944168e-06,
+      "loss": 0.9982,
+      "step": 18321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5588228430553666e-06,
+      "loss": 1.0077,
+      "step": 18322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.557908709517896e-06,
+      "loss": 0.9858,
+      "step": 18323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.556994715348878e-06,
+      "loss": 0.9536,
+      "step": 18324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.556080860565424e-06,
+      "loss": 0.9038,
+      "step": 18325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.55516714518465e-06,
+      "loss": 0.9805,
+      "step": 18326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.554253569223667e-06,
+      "loss": 0.9518,
+      "step": 18327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5533401326995836e-06,
+      "loss": 0.9806,
+      "step": 18328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.552426835629507e-06,
+      "loss": 0.9427,
+      "step": 18329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.551513678030541e-06,
+      "loss": 0.9212,
+      "step": 18330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.55060065991978e-06,
+      "loss": 0.9332,
+      "step": 18331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.549687781314332e-06,
+      "loss": 0.9242,
+      "step": 18332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.548775042231285e-06,
+      "loss": 0.9197,
+      "step": 18333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.547862442687735e-06,
+      "loss": 0.9985,
+      "step": 18334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.546949982700775e-06,
+      "loss": 0.9045,
+      "step": 18335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.546037662287483e-06,
+      "loss": 0.9617,
+      "step": 18336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.545125481464956e-06,
+      "loss": 0.975,
+      "step": 18337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.544213440250265e-06,
+      "loss": 0.9908,
+      "step": 18338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.5433015386605015e-06,
+      "loss": 0.9434,
+      "step": 18339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.542389776712736e-06,
+      "loss": 0.9506,
+      "step": 18340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.541478154424042e-06,
+      "loss": 0.8766,
+      "step": 18341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.540566671811498e-06,
+      "loss": 0.9068,
+      "step": 18342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.539655328892167e-06,
+      "loss": 0.9872,
+      "step": 18343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.538744125683116e-06,
+      "loss": 0.9622,
+      "step": 18344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5378330622014124e-06,
+      "loss": 0.9478,
+      "step": 18345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.536922138464115e-06,
+      "loss": 0.9216,
+      "step": 18346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.536011354488285e-06,
+      "loss": 0.955,
+      "step": 18347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.535100710290981e-06,
+      "loss": 0.9752,
+      "step": 18348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.534190205889246e-06,
+      "loss": 1.0014,
+      "step": 18349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.533279841300146e-06,
+      "loss": 0.9659,
+      "step": 18350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.532369616540718e-06,
+      "loss": 0.9276,
+      "step": 18351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.531459531628012e-06,
+      "loss": 0.9434,
+      "step": 18352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.530549586579074e-06,
+      "loss": 1.007,
+      "step": 18353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.529639781410936e-06,
+      "loss": 0.9769,
+      "step": 18354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.528730116140647e-06,
+      "loss": 0.9629,
+      "step": 18355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.527820590785231e-06,
+      "loss": 0.9196,
+      "step": 18356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.526911205361733e-06,
+      "loss": 0.9542,
+      "step": 18357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.526001959887173e-06,
+      "loss": 0.9535,
+      "step": 18358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.525092854378583e-06,
+      "loss": 0.9521,
+      "step": 18359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5241838888529866e-06,
+      "loss": 0.9075,
+      "step": 18360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.523275063327407e-06,
+      "loss": 0.9062,
+      "step": 18361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.522366377818866e-06,
+      "loss": 0.9638,
+      "step": 18362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.521457832344376e-06,
+      "loss": 0.968,
+      "step": 18363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.520549426920952e-06,
+      "loss": 1.0219,
+      "step": 18364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5196411615656075e-06,
+      "loss": 0.9855,
+      "step": 18365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5187330362953515e-06,
+      "loss": 0.9618,
+      "step": 18366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.517825051127191e-06,
+      "loss": 0.9923,
+      "step": 18367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.516917206078131e-06,
+      "loss": 0.9449,
+      "step": 18368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.516009501165168e-06,
+      "loss": 0.9689,
+      "step": 18369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515101936405304e-06,
+      "loss": 0.9729,
+      "step": 18370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5141945118155335e-06,
+      "loss": 1.0069,
+      "step": 18371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.513287227412853e-06,
+      "loss": 0.9783,
+      "step": 18372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.512380083214253e-06,
+      "loss": 0.9763,
+      "step": 18373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.511473079236714e-06,
+      "loss": 0.9834,
+      "step": 18374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.510566215497233e-06,
+      "loss": 0.9886,
+      "step": 18375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.509659492012782e-06,
+      "loss": 0.9516,
+      "step": 18376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.508752908800352e-06,
+      "loss": 0.9253,
+      "step": 18377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.507846465876912e-06,
+      "loss": 0.9314,
+      "step": 18378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.50694016325944e-06,
+      "loss": 0.9416,
+      "step": 18379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.506034000964911e-06,
+      "loss": 0.9677,
+      "step": 18380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.505127979010288e-06,
+      "loss": 1.0016,
+      "step": 18381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.504222097412547e-06,
+      "loss": 0.8819,
+      "step": 18382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.503316356188646e-06,
+      "loss": 0.9607,
+      "step": 18383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.502410755355549e-06,
+      "loss": 0.9424,
+      "step": 18384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.501505294930214e-06,
+      "loss": 0.9506,
+      "step": 18385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.500599974929599e-06,
+      "loss": 0.9532,
+      "step": 18386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.499694795370657e-06,
+      "loss": 0.9688,
+      "step": 18387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.498789756270345e-06,
+      "loss": 0.9585,
+      "step": 18388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.497884857645602e-06,
+      "loss": 0.9554,
+      "step": 18389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4969800995133784e-06,
+      "loss": 0.9465,
+      "step": 18390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.496075481890619e-06,
+      "loss": 0.9211,
+      "step": 18391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.495171004794262e-06,
+      "loss": 0.9655,
+      "step": 18392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.494266668241252e-06,
+      "loss": 1.0016,
+      "step": 18393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.493362472248513e-06,
+      "loss": 0.9455,
+      "step": 18394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4924584168329905e-06,
+      "loss": 0.9068,
+      "step": 18395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.491554502011606e-06,
+      "loss": 0.9761,
+      "step": 18396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.49065072780129e-06,
+      "loss": 0.9484,
+      "step": 18397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.489747094218969e-06,
+      "loss": 0.9543,
+      "step": 18398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.488843601281563e-06,
+      "loss": 0.9401,
+      "step": 18399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.487940249005997e-06,
+      "loss": 0.9297,
+      "step": 18400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.487037037409177e-06,
+      "loss": 0.9725,
+      "step": 18401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48613396650803e-06,
+      "loss": 0.9225,
+      "step": 18402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.48523103631946e-06,
+      "loss": 0.9304,
+      "step": 18403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.484328246860378e-06,
+      "loss": 0.9236,
+      "step": 18404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.483425598147691e-06,
+      "loss": 0.9718,
+      "step": 18405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.482523090198304e-06,
+      "loss": 0.9883,
+      "step": 18406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4816207230291205e-06,
+      "loss": 0.9888,
+      "step": 18407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.480718496657032e-06,
+      "loss": 0.9881,
+      "step": 18408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47981641109894e-06,
+      "loss": 0.8881,
+      "step": 18409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478914466371735e-06,
+      "loss": 0.957,
+      "step": 18410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.478012662492313e-06,
+      "loss": 0.9791,
+      "step": 18411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.477110999477552e-06,
+      "loss": 0.9397,
+      "step": 18412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.476209477344351e-06,
+      "loss": 0.8747,
+      "step": 18413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.475308096109579e-06,
+      "loss": 0.9428,
+      "step": 18414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.474406855790129e-06,
+      "loss": 0.9723,
+      "step": 18415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47350575640287e-06,
+      "loss": 0.9564,
+      "step": 18416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.472604797964679e-06,
+      "loss": 1.0106,
+      "step": 18417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4717039804924314e-06,
+      "loss": 0.9251,
+      "step": 18418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.47080330400299e-06,
+      "loss": 0.9794,
+      "step": 18419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46990276851323e-06,
+      "loss": 0.9436,
+      "step": 18420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46900237404001e-06,
+      "loss": 0.9374,
+      "step": 18421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468102120600193e-06,
+      "loss": 1.0177,
+      "step": 18422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.46720200821064e-06,
+      "loss": 0.908,
+      "step": 18423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.466302036888206e-06,
+      "loss": 0.968,
+      "step": 18424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.465402206649742e-06,
+      "loss": 0.9852,
+      "step": 18425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.464502517512108e-06,
+      "loss": 0.967,
+      "step": 18426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.463602969492142e-06,
+      "loss": 1.0066,
+      "step": 18427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.462703562606696e-06,
+      "loss": 0.9585,
+      "step": 18428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.461804296872611e-06,
+      "loss": 0.9217,
+      "step": 18429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.460905172306729e-06,
+      "loss": 1.0156,
+      "step": 18430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4600061889258905e-06,
+      "loss": 0.9782,
+      "step": 18431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.45910734674692e-06,
+      "loss": 1.0128,
+      "step": 18432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4582086457866665e-06,
+      "loss": 0.9671,
+      "step": 18433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.457310086061945e-06,
+      "loss": 0.988,
+      "step": 18434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.456411667589592e-06,
+      "loss": 0.9772,
+      "step": 18435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.455513390386428e-06,
+      "loss": 0.9246,
+      "step": 18436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.454615254469277e-06,
+      "loss": 0.9991,
+      "step": 18437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.453717259854961e-06,
+      "loss": 0.9802,
+      "step": 18438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.452819406560287e-06,
+      "loss": 0.943,
+      "step": 18439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.451921694602083e-06,
+      "loss": 1.0278,
+      "step": 18440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.45102412399715e-06,
+      "loss": 0.98,
+      "step": 18441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4501266947623005e-06,
+      "loss": 0.9198,
+      "step": 18442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.44922940691434e-06,
+      "loss": 0.8902,
+      "step": 18443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.448332260470072e-06,
+      "loss": 0.9418,
+      "step": 18444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.447435255446302e-06,
+      "loss": 0.9018,
+      "step": 18445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.446538391859821e-06,
+      "loss": 0.9584,
+      "step": 18446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.445641669727427e-06,
+      "loss": 0.9837,
+      "step": 18447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.444745089065916e-06,
+      "loss": 1.0753,
+      "step": 18448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4438486498920725e-06,
+      "loss": 1.0031,
+      "step": 18449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4429523522226905e-06,
+      "loss": 0.9381,
+      "step": 18450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.442056196074555e-06,
+      "loss": 0.9755,
+      "step": 18451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4411601814644386e-06,
+      "loss": 0.9499,
+      "step": 18452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.440264308409135e-06,
+      "loss": 0.9541,
+      "step": 18453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.43936857692541e-06,
+      "loss": 0.9687,
+      "step": 18454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.438472987030044e-06,
+      "loss": 0.9307,
+      "step": 18455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.437577538739809e-06,
+      "loss": 0.902,
+      "step": 18456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.436682232071466e-06,
+      "loss": 1.0007,
+      "step": 18457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.435787067041795e-06,
+      "loss": 1.0289,
+      "step": 18458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4348920436675455e-06,
+      "loss": 0.9086,
+      "step": 18459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.433997161965492e-06,
+      "loss": 0.9469,
+      "step": 18460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.433102421952384e-06,
+      "loss": 0.9119,
+      "step": 18461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.432207823644979e-06,
+      "loss": 0.9129,
+      "step": 18462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.43131336706003e-06,
+      "loss": 1.0135,
+      "step": 18463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.430419052214291e-06,
+      "loss": 0.9772,
+      "step": 18464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4295248791245094e-06,
+      "loss": 1.0,
+      "step": 18465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.428630847807426e-06,
+      "loss": 0.9148,
+      "step": 18466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.427736958279786e-06,
+      "loss": 0.9778,
+      "step": 18467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.426843210558329e-06,
+      "loss": 0.9211,
+      "step": 18468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.425949604659792e-06,
+      "loss": 0.9853,
+      "step": 18469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.425056140600912e-06,
+      "loss": 0.9753,
+      "step": 18470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.42416281839842e-06,
+      "loss": 0.9368,
+      "step": 18471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.423269638069042e-06,
+      "loss": 0.9535,
+      "step": 18472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.422376599629508e-06,
+      "loss": 0.9468,
+      "step": 18473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.42148370309654e-06,
+      "loss": 0.9512,
+      "step": 18474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.420590948486861e-06,
+      "loss": 0.936,
+      "step": 18475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.419698335817192e-06,
+      "loss": 0.9238,
+      "step": 18476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4188058651042405e-06,
+      "loss": 0.9636,
+      "step": 18477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4179135363647295e-06,
+      "loss": 0.9434,
+      "step": 18478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.417021349615364e-06,
+      "loss": 0.9278,
+      "step": 18479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.416129304872853e-06,
+      "loss": 0.9607,
+      "step": 18480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.415237402153902e-06,
+      "loss": 0.9376,
+      "step": 18481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.414345641475215e-06,
+      "loss": 0.9367,
+      "step": 18482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.413454022853495e-06,
+      "loss": 0.9337,
+      "step": 18483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.41256254630543e-06,
+      "loss": 0.9418,
+      "step": 18484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.411671211847722e-06,
+      "loss": 0.9455,
+      "step": 18485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4107800194970625e-06,
+      "loss": 0.9981,
+      "step": 18486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4098889692701375e-06,
+      "loss": 0.945,
+      "step": 18487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.408998061183637e-06,
+      "loss": 0.9531,
+      "step": 18488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.408107295254249e-06,
+      "loss": 0.919,
+      "step": 18489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4072166714986415e-06,
+      "loss": 0.9078,
+      "step": 18490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.406326189933509e-06,
+      "loss": 0.9159,
+      "step": 18491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.405435850575516e-06,
+      "loss": 0.9726,
+      "step": 18492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.404545653441341e-06,
+      "loss": 1.0049,
+      "step": 18493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.403655598547658e-06,
+      "loss": 0.9471,
+      "step": 18494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.402765685911125e-06,
+      "loss": 0.958,
+      "step": 18495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.401875915548419e-06,
+      "loss": 0.9256,
+      "step": 18496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.400986287476192e-06,
+      "loss": 0.9761,
+      "step": 18497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.400096801711116e-06,
+      "loss": 0.999,
+      "step": 18498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3992074582698375e-06,
+      "loss": 0.9762,
+      "step": 18499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.398318257169017e-06,
+      "loss": 0.9759,
+      "step": 18500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.397429198425305e-06,
+      "loss": 0.9644,
+      "step": 18501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3965402820553515e-06,
+      "loss": 0.9559,
+      "step": 18502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.395651508075805e-06,
+      "loss": 0.9701,
+      "step": 18503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.394762876503307e-06,
+      "loss": 0.9578,
+      "step": 18504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3938743873544985e-06,
+      "loss": 0.9341,
+      "step": 18505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3929860406460205e-06,
+      "loss": 0.9639,
+      "step": 18506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3920978363945065e-06,
+      "loss": 0.8973,
+      "step": 18507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.391209774616591e-06,
+      "loss": 1.0277,
+      "step": 18508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.390321855328911e-06,
+      "loss": 1.0156,
+      "step": 18509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3894340785480845e-06,
+      "loss": 0.9377,
+      "step": 18510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3885464442907414e-06,
+      "loss": 0.9914,
+      "step": 18511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.387658952573505e-06,
+      "loss": 0.9292,
+      "step": 18512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.386771603412994e-06,
+      "loss": 0.9087,
+      "step": 18513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3858843968258295e-06,
+      "loss": 0.9294,
+      "step": 18514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.384997332828619e-06,
+      "loss": 0.9989,
+      "step": 18515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.384110411437984e-06,
+      "loss": 0.9456,
+      "step": 18516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.383223632670523e-06,
+      "loss": 0.9496,
+      "step": 18517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.382336996542854e-06,
+      "loss": 0.9981,
+      "step": 18518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.381450503071575e-06,
+      "loss": 0.9853,
+      "step": 18519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.380564152273286e-06,
+      "loss": 1.0045,
+      "step": 18520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.379677944164592e-06,
+      "loss": 0.8617,
+      "step": 18521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.37879187876208e-06,
+      "loss": 0.9671,
+      "step": 18522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.377905956082354e-06,
+      "loss": 0.9468,
+      "step": 18523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.377020176141995e-06,
+      "loss": 0.9824,
+      "step": 18524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.376134538957596e-06,
+      "loss": 1.0177,
+      "step": 18525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.375249044545742e-06,
+      "loss": 0.8938,
+      "step": 18526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.374363692923015e-06,
+      "loss": 0.9232,
+      "step": 18527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.373478484105998e-06,
+      "loss": 0.9718,
+      "step": 18528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.372593418111264e-06,
+      "loss": 0.9628,
+      "step": 18529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.371708494955387e-06,
+      "loss": 1.0369,
+      "step": 18530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.370823714654943e-06,
+      "loss": 0.9561,
+      "step": 18531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3699390772265e-06,
+      "loss": 0.9287,
+      "step": 18532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.369054582686623e-06,
+      "loss": 0.8992,
+      "step": 18533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.368170231051882e-06,
+      "loss": 0.9298,
+      "step": 18534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.367286022338826e-06,
+      "loss": 0.944,
+      "step": 18535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.366401956564028e-06,
+      "loss": 0.9924,
+      "step": 18536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.365518033744034e-06,
+      "loss": 0.9508,
+      "step": 18537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3646342538954e-06,
+      "loss": 0.9661,
+      "step": 18538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.363750617034681e-06,
+      "loss": 0.9679,
+      "step": 18539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.362867123178414e-06,
+      "loss": 0.9131,
+      "step": 18540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.361983772343159e-06,
+      "loss": 1.0008,
+      "step": 18541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.361100564545445e-06,
+      "loss": 0.9583,
+      "step": 18542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.360217499801821e-06,
+      "loss": 0.8825,
+      "step": 18543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.35933457812882e-06,
+      "loss": 0.8937,
+      "step": 18544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.358451799542976e-06,
+      "loss": 0.9669,
+      "step": 18545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.357569164060823e-06,
+      "loss": 0.942,
+      "step": 18546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.35668667169889e-06,
+      "loss": 0.9433,
+      "step": 18547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.355804322473704e-06,
+      "loss": 0.9705,
+      "step": 18548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3549221164017865e-06,
+      "loss": 0.9983,
+      "step": 18549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3540400534996585e-06,
+      "loss": 0.9714,
+      "step": 18550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3531581337838415e-06,
+      "loss": 0.9915,
+      "step": 18551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.35227635727085e-06,
+      "loss": 0.9751,
+      "step": 18552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.351394723977192e-06,
+      "loss": 0.9372,
+      "step": 18553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.350513233919389e-06,
+      "loss": 0.9939,
+      "step": 18554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3496318871139395e-06,
+      "loss": 0.9213,
+      "step": 18555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.34875068357735e-06,
+      "loss": 0.946,
+      "step": 18556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3478696233261246e-06,
+      "loss": 0.9573,
+      "step": 18557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3469887063767615e-06,
+      "loss": 0.9503,
+      "step": 18558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.346107932745763e-06,
+      "loss": 0.9471,
+      "step": 18559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3452273024496115e-06,
+      "loss": 0.9509,
+      "step": 18560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.344346815504813e-06,
+      "loss": 0.9635,
+      "step": 18561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.343466471927847e-06,
+      "loss": 0.9326,
+      "step": 18562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.342586271735202e-06,
+      "loss": 0.9363,
+      "step": 18563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.341706214943361e-06,
+      "loss": 1.0477,
+      "step": 18564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.340826301568807e-06,
+      "loss": 0.989,
+      "step": 18565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.339946531628019e-06,
+      "loss": 0.9641,
+      "step": 18566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.339066905137467e-06,
+      "loss": 1.0268,
+      "step": 18567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3381874221136275e-06,
+      "loss": 1.0064,
+      "step": 18568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3373080825729715e-06,
+      "loss": 0.9508,
+      "step": 18569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3364288865319634e-06,
+      "loss": 0.9736,
+      "step": 18570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.335549834007071e-06,
+      "loss": 0.9355,
+      "step": 18571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.334670925014759e-06,
+      "loss": 0.9213,
+      "step": 18572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.333792159571476e-06,
+      "loss": 0.9947,
+      "step": 18573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.332913537693692e-06,
+      "loss": 0.9608,
+      "step": 18574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.332035059397851e-06,
+      "loss": 0.9085,
+      "step": 18575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3311567247004084e-06,
+      "loss": 0.8735,
+      "step": 18576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.330278533617815e-06,
+      "loss": 0.9563,
+      "step": 18577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.329400486166509e-06,
+      "loss": 0.9462,
+      "step": 18578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.3285225823629436e-06,
+      "loss": 0.9481,
+      "step": 18579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.327644822223549e-06,
+      "loss": 0.9702,
+      "step": 18580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.326767205764774e-06,
+      "loss": 0.9049,
+      "step": 18581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.325889733003047e-06,
+      "loss": 0.9902,
+      "step": 18582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3250124039548e-06,
+      "loss": 1.0049,
+      "step": 18583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.324135218636467e-06,
+      "loss": 0.9853,
+      "step": 18584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.32325817706447e-06,
+      "loss": 0.9829,
+      "step": 18585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3223812792552405e-06,
+      "loss": 0.9593,
+      "step": 18586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.321504525225192e-06,
+      "loss": 0.8996,
+      "step": 18587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.320627914990748e-06,
+      "loss": 0.9559,
+      "step": 18588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.319751448568325e-06,
+      "loss": 0.8495,
+      "step": 18589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.318875125974335e-06,
+      "loss": 1.0164,
+      "step": 18590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.317998947225189e-06,
+      "loss": 0.9615,
+      "step": 18591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3171229123373e-06,
+      "loss": 0.9737,
+      "step": 18592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.316247021327066e-06,
+      "loss": 0.9366,
+      "step": 18593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.315371274210894e-06,
+      "loss": 0.9755,
+      "step": 18594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.314495671005183e-06,
+      "loss": 0.9213,
+      "step": 18595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.313620211726332e-06,
+      "loss": 0.9895,
+      "step": 18596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.312744896390737e-06,
+      "loss": 0.9747,
+      "step": 18597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.311869725014782e-06,
+      "loss": 0.9692,
+      "step": 18598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3109946976148684e-06,
+      "loss": 0.9853,
+      "step": 18599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.310119814207371e-06,
+      "loss": 0.9715,
+      "step": 18600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.309245074808685e-06,
+      "loss": 1.0297,
+      "step": 18601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.308370479435184e-06,
+      "loss": 0.9836,
+      "step": 18602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3074960281032485e-06,
+      "loss": 0.9273,
+      "step": 18603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3066217208292575e-06,
+      "loss": 0.9697,
+      "step": 18604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.305747557629576e-06,
+      "loss": 1.0276,
+      "step": 18605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.304873538520584e-06,
+      "loss": 0.9226,
+      "step": 18606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.303999663518645e-06,
+      "loss": 0.9613,
+      "step": 18607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.303125932640123e-06,
+      "loss": 0.9137,
+      "step": 18608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3022523459013825e-06,
+      "loss": 1.007,
+      "step": 18609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.301378903318782e-06,
+      "loss": 0.9162,
+      "step": 18610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3005056049086785e-06,
+      "loss": 0.973,
+      "step": 18611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.29963245068743e-06,
+      "loss": 0.9455,
+      "step": 18612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2987594406713825e-06,
+      "loss": 0.9536,
+      "step": 18613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.297886574876886e-06,
+      "loss": 0.9905,
+      "step": 18614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2970138533202884e-06,
+      "loss": 0.856,
+      "step": 18615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.296141276017934e-06,
+      "loss": 0.9403,
+      "step": 18616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2952688429861645e-06,
+      "loss": 0.9923,
+      "step": 18617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.29439655424131e-06,
+      "loss": 0.9542,
+      "step": 18618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.293524409799718e-06,
+      "loss": 0.969,
+      "step": 18619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.292652409677711e-06,
+      "loss": 0.9042,
+      "step": 18620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.291780553891624e-06,
+      "loss": 0.913,
+      "step": 18621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.290908842457784e-06,
+      "loss": 0.9717,
+      "step": 18622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.290037275392514e-06,
+      "loss": 0.8898,
+      "step": 18623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.289165852712139e-06,
+      "loss": 0.9569,
+      "step": 18624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.288294574432974e-06,
+      "loss": 0.9741,
+      "step": 18625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.287423440571338e-06,
+      "loss": 0.995,
+      "step": 18626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2865524511435435e-06,
+      "loss": 0.943,
+      "step": 18627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2856816061659014e-06,
+      "loss": 1.0169,
+      "step": 18628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.284810905654722e-06,
+      "loss": 0.9937,
+      "step": 18629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.283940349626313e-06,
+      "loss": 0.9658,
+      "step": 18630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28306993809697e-06,
+      "loss": 0.876,
+      "step": 18631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.282199671082999e-06,
+      "loss": 0.918,
+      "step": 18632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.281329548600695e-06,
+      "loss": 1.0391,
+      "step": 18633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.280459570666353e-06,
+      "loss": 0.9573,
+      "step": 18634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.27958973729627e-06,
+      "loss": 0.9208,
+      "step": 18635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.278720048506725e-06,
+      "loss": 0.8941,
+      "step": 18636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.277850504314017e-06,
+      "loss": 0.9286,
+      "step": 18637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.276981104734419e-06,
+      "loss": 0.9302,
+      "step": 18638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2761118497842235e-06,
+      "loss": 0.9653,
+      "step": 18639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2752427394797e-06,
+      "loss": 0.9479,
+      "step": 18640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.274373773837126e-06,
+      "loss": 0.9058,
+      "step": 18641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.273504952872779e-06,
+      "loss": 0.9219,
+      "step": 18642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.272636276602922e-06,
+      "loss": 0.9379,
+      "step": 18643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.271767745043833e-06,
+      "loss": 0.8615,
+      "step": 18644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.270899358211767e-06,
+      "loss": 0.9926,
+      "step": 18645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.270031116122991e-06,
+      "loss": 0.9481,
+      "step": 18646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.269163018793766e-06,
+      "loss": 0.9738,
+      "step": 18647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.268295066240345e-06,
+      "loss": 0.9678,
+      "step": 18648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.267427258478983e-06,
+      "loss": 0.9306,
+      "step": 18649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.266559595525939e-06,
+      "loss": 0.9435,
+      "step": 18650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.265692077397451e-06,
+      "loss": 1.0303,
+      "step": 18651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.264824704109769e-06,
+      "loss": 0.9907,
+      "step": 18652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2639574756791385e-06,
+      "loss": 0.9126,
+      "step": 18653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.263090392121797e-06,
+      "loss": 0.9108,
+      "step": 18654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.262223453453988e-06,
+      "loss": 0.9304,
+      "step": 18655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.261356659691937e-06,
+      "loss": 0.9314,
+      "step": 18656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.260490010851888e-06,
+      "loss": 0.9597,
+      "step": 18657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.259623506950061e-06,
+      "loss": 0.9593,
+      "step": 18658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.258757148002688e-06,
+      "loss": 0.9824,
+      "step": 18659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.257890934025992e-06,
+      "loss": 0.9545,
+      "step": 18660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.257024865036196e-06,
+      "loss": 0.989,
+      "step": 18661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25615894104952e-06,
+      "loss": 0.9589,
+      "step": 18662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2552931620821725e-06,
+      "loss": 0.9702,
+      "step": 18663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.254427528150379e-06,
+      "loss": 0.8998,
+      "step": 18664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.25356203927034e-06,
+      "loss": 0.9325,
+      "step": 18665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.252696695458268e-06,
+      "loss": 0.8992,
+      "step": 18666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.251831496730369e-06,
+      "loss": 1.023,
+      "step": 18667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.250966443102843e-06,
+      "loss": 0.9433,
+      "step": 18668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.250101534591894e-06,
+      "loss": 0.9246,
+      "step": 18669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.249236771213715e-06,
+      "loss": 0.9914,
+      "step": 18670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.248372152984501e-06,
+      "loss": 0.9422,
+      "step": 18671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.247507679920445e-06,
+      "loss": 0.974,
+      "step": 18672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.246643352037736e-06,
+      "loss": 0.8906,
+      "step": 18673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.245779169352561e-06,
+      "loss": 0.8958,
+      "step": 18674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.244915131881104e-06,
+      "loss": 0.9868,
+      "step": 18675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.244051239639539e-06,
+      "loss": 1.0334,
+      "step": 18676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.243187492644057e-06,
+      "loss": 0.9797,
+      "step": 18677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.242323890910822e-06,
+      "loss": 0.9903,
+      "step": 18678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.241460434456011e-06,
+      "loss": 0.9974,
+      "step": 18679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.240597123295798e-06,
+      "loss": 0.9124,
+      "step": 18680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2397339574463394e-06,
+      "loss": 0.9846,
+      "step": 18681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.238870936923811e-06,
+      "loss": 0.9465,
+      "step": 18682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.238008061744366e-06,
+      "loss": 0.8792,
+      "step": 18683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.237145331924174e-06,
+      "loss": 0.9188,
+      "step": 18684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2362827474793815e-06,
+      "loss": 0.8854,
+      "step": 18685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.235420308426146e-06,
+      "loss": 0.9614,
+      "step": 18686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2345580147806175e-06,
+      "loss": 0.9849,
+      "step": 18687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.233695866558948e-06,
+      "loss": 0.9207,
+      "step": 18688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.232833863777277e-06,
+      "loss": 0.9752,
+      "step": 18689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.231972006451751e-06,
+      "loss": 0.9149,
+      "step": 18690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.231110294598508e-06,
+      "loss": 0.9859,
+      "step": 18691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2302487282336866e-06,
+      "loss": 0.9441,
+      "step": 18692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.229387307373425e-06,
+      "loss": 0.9588,
+      "step": 18693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.228526032033846e-06,
+      "loss": 0.9449,
+      "step": 18694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.227664902231089e-06,
+      "loss": 1.0162,
+      "step": 18695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.226803917981274e-06,
+      "loss": 0.8831,
+      "step": 18696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.225943079300525e-06,
+      "loss": 0.9762,
+      "step": 18697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.225082386204967e-06,
+      "loss": 0.9446,
+      "step": 18698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2242218387107106e-06,
+      "loss": 0.9372,
+      "step": 18699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.223361436833882e-06,
+      "loss": 0.9649,
+      "step": 18700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.222501180590582e-06,
+      "loss": 0.9933,
+      "step": 18701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2216410699969325e-06,
+      "loss": 0.9445,
+      "step": 18702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.220781105069033e-06,
+      "loss": 1.0479,
+      "step": 18703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21992128582299e-06,
+      "loss": 0.9456,
+      "step": 18704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2190616122749054e-06,
+      "loss": 1.0007,
+      "step": 18705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.218202084440879e-06,
+      "loss": 0.9546,
+      "step": 18706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21734270233701e-06,
+      "loss": 0.9474,
+      "step": 18707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.216483465979385e-06,
+      "loss": 0.9647,
+      "step": 18708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2156243753841e-06,
+      "loss": 1.0162,
+      "step": 18709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.214765430567243e-06,
+      "loss": 0.9439,
+      "step": 18710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213906631544898e-06,
+      "loss": 0.9926,
+      "step": 18711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.213047978333148e-06,
+      "loss": 0.9557,
+      "step": 18712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.212189470948076e-06,
+      "loss": 0.9743,
+      "step": 18713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.211331109405754e-06,
+      "loss": 1.0107,
+      "step": 18714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.21047289372226e-06,
+      "loss": 0.8834,
+      "step": 18715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.209614823913666e-06,
+      "loss": 0.9366,
+      "step": 18716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2087568999960395e-06,
+      "loss": 0.9529,
+      "step": 18717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.207899121985451e-06,
+      "loss": 0.9687,
+      "step": 18718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2070414898979555e-06,
+      "loss": 0.9912,
+      "step": 18719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2061840037496235e-06,
+      "loss": 0.9577,
+      "step": 18720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.205326663556505e-06,
+      "loss": 0.9409,
+      "step": 18721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2044694693346655e-06,
+      "loss": 0.9502,
+      "step": 18722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.203612421100147e-06,
+      "loss": 0.9587,
+      "step": 18723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.202755518869006e-06,
+      "loss": 0.977,
+      "step": 18724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.201898762657291e-06,
+      "loss": 0.9603,
+      "step": 18725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.2010421524810365e-06,
+      "loss": 0.9327,
+      "step": 18726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.200185688356298e-06,
+      "loss": 0.9528,
+      "step": 18727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.199329370299104e-06,
+      "loss": 0.9978,
+      "step": 18728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1984731983254945e-06,
+      "loss": 0.9628,
+      "step": 18729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.197617172451504e-06,
+      "loss": 0.8993,
+      "step": 18730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.196761292693163e-06,
+      "loss": 0.9375,
+      "step": 18731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1959055590664975e-06,
+      "loss": 0.9362,
+      "step": 18732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.19504997158754e-06,
+      "loss": 0.9191,
+      "step": 18733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.194194530272302e-06,
+      "loss": 0.9994,
+      "step": 18734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.193339235136811e-06,
+      "loss": 0.9644,
+      "step": 18735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.192484086197082e-06,
+      "loss": 0.9483,
+      "step": 18736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.191629083469129e-06,
+      "loss": 0.8752,
+      "step": 18737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.190774226968967e-06,
+      "loss": 0.9948,
+      "step": 18738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.189919516712597e-06,
+      "loss": 1.0319,
+      "step": 18739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.189064952716037e-06,
+      "loss": 0.9006,
+      "step": 18740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.18821053499528e-06,
+      "loss": 0.9294,
+      "step": 18741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1873562635663315e-06,
+      "loss": 0.946,
+      "step": 18742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.186502138445187e-06,
+      "loss": 0.9975,
+      "step": 18743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.185648159647846e-06,
+      "loss": 0.9881,
+      "step": 18744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.184794327190299e-06,
+      "loss": 0.9239,
+      "step": 18745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.183940641088531e-06,
+      "loss": 0.9547,
+      "step": 18746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.183087101358538e-06,
+      "loss": 0.9972,
+      "step": 18747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.182233708016296e-06,
+      "loss": 0.9381,
+      "step": 18748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.181380461077792e-06,
+      "loss": 0.9419,
+      "step": 18749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.180527360559002e-06,
+      "loss": 0.9246,
+      "step": 18750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1796744064759034e-06,
+      "loss": 0.9686,
+      "step": 18751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.178821598844471e-06,
+      "loss": 0.9172,
+      "step": 18752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1779689376806705e-06,
+      "loss": 0.9448,
+      "step": 18753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.177116423000474e-06,
+      "loss": 0.962,
+      "step": 18754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.176264054819843e-06,
+      "loss": 0.9929,
+      "step": 18755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.175411833154747e-06,
+      "loss": 0.9351,
+      "step": 18756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.174559758021134e-06,
+      "loss": 0.8817,
+      "step": 18757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.173707829434973e-06,
+      "loss": 0.95,
+      "step": 18758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.172856047412207e-06,
+      "loss": 0.9624,
+      "step": 18759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1720044119688e-06,
+      "loss": 1.0049,
+      "step": 18760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.171152923120689e-06,
+      "loss": 0.9315,
+      "step": 18761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1703015808838254e-06,
+      "loss": 0.9608,
+      "step": 18762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1694503852741545e-06,
+      "loss": 1.0148,
+      "step": 18763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.168599336307608e-06,
+      "loss": 0.9347,
+      "step": 18764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.167748434000135e-06,
+      "loss": 0.9882,
+      "step": 18765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.16689767836766e-06,
+      "loss": 0.9165,
+      "step": 18766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.166047069426121e-06,
+      "loss": 0.9778,
+      "step": 18767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.165196607191446e-06,
+      "loss": 0.8853,
+      "step": 18768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.164346291679562e-06,
+      "loss": 0.9475,
+      "step": 18769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.163496122906392e-06,
+      "loss": 0.9433,
+      "step": 18770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.162646100887862e-06,
+      "loss": 0.966,
+      "step": 18771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1617962256398824e-06,
+      "loss": 0.9647,
+      "step": 18772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.160946497178373e-06,
+      "loss": 0.9064,
+      "step": 18773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.160096915519246e-06,
+      "loss": 1.0029,
+      "step": 18774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.159247480678413e-06,
+      "loss": 0.935,
+      "step": 18775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.158398192671783e-06,
+      "loss": 0.969,
+      "step": 18776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.157549051515251e-06,
+      "loss": 1.0139,
+      "step": 18777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1567000572247314e-06,
+      "loss": 0.903,
+      "step": 18778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.155851209816117e-06,
+      "loss": 0.9905,
+      "step": 18779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.155002509305304e-06,
+      "loss": 0.8487,
+      "step": 18780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.154153955708188e-06,
+      "loss": 0.9666,
+      "step": 18781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.153305549040658e-06,
+      "loss": 0.9107,
+      "step": 18782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.152457289318606e-06,
+      "loss": 0.9581,
+      "step": 18783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1516091765579106e-06,
+      "loss": 0.9542,
+      "step": 18784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.150761210774463e-06,
+      "loss": 1.009,
+      "step": 18785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.149913391984137e-06,
+      "loss": 0.9735,
+      "step": 18786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.149065720202809e-06,
+      "loss": 0.9284,
+      "step": 18787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.148218195446359e-06,
+      "loss": 0.9503,
+      "step": 18788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.147370817730653e-06,
+      "loss": 0.9697,
+      "step": 18789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.146523587071566e-06,
+      "loss": 0.9075,
+      "step": 18790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.145676503484957e-06,
+      "loss": 0.9496,
+      "step": 18791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.144829566986693e-06,
+      "loss": 0.9624,
+      "step": 18792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.143982777592634e-06,
+      "loss": 0.9303,
+      "step": 18793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.143136135318639e-06,
+      "loss": 0.9249,
+      "step": 18794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1422896401805635e-06,
+      "loss": 0.9359,
+      "step": 18795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.141443292194261e-06,
+      "loss": 0.9872,
+      "step": 18796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.140597091375572e-06,
+      "loss": 0.9448,
+      "step": 18797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.139751037740356e-06,
+      "loss": 0.9448,
+      "step": 18798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1389051313044494e-06,
+      "loss": 1.0031,
+      "step": 18799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.138059372083696e-06,
+      "loss": 0.9539,
+      "step": 18800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.137213760093936e-06,
+      "loss": 0.9778,
+      "step": 18801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.136368295350997e-06,
+      "loss": 0.9932,
+      "step": 18802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.135522977870725e-06,
+      "loss": 0.9301,
+      "step": 18803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.134677807668936e-06,
+      "loss": 0.9212,
+      "step": 18804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.133832784761471e-06,
+      "loss": 0.9945,
+      "step": 18805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1329879091641475e-06,
+      "loss": 0.9919,
+      "step": 18806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.132143180892787e-06,
+      "loss": 0.9634,
+      "step": 18807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.131298599963211e-06,
+      "loss": 0.9882,
+      "step": 18808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.1304541663912365e-06,
+      "loss": 0.9497,
+      "step": 18809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.129609880192679e-06,
+      "loss": 0.9831,
+      "step": 18810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.128765741383344e-06,
+      "loss": 0.9849,
+      "step": 18811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.127921749979042e-06,
+      "loss": 0.9621,
+      "step": 18812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.127077905995579e-06,
+      "loss": 0.9066,
+      "step": 18813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.126234209448758e-06,
+      "loss": 0.9183,
+      "step": 18814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.125390660354379e-06,
+      "loss": 0.9726,
+      "step": 18815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.124547258728243e-06,
+      "loss": 0.9905,
+      "step": 18816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.123704004586135e-06,
+      "loss": 1.0543,
+      "step": 18817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122860897943852e-06,
+      "loss": 1.0024,
+      "step": 18818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1220179388171845e-06,
+      "loss": 0.9749,
+      "step": 18819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.121175127221917e-06,
+      "loss": 0.9369,
+      "step": 18820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.120332463173836e-06,
+      "loss": 0.8994,
+      "step": 18821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.119489946688711e-06,
+      "loss": 0.9436,
+      "step": 18822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.118647577782336e-06,
+      "loss": 0.9552,
+      "step": 18823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.117805356470475e-06,
+      "loss": 0.9524,
+      "step": 18824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.116963282768902e-06,
+      "loss": 0.883,
+      "step": 18825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.116121356693389e-06,
+      "loss": 0.9519,
+      "step": 18826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.115279578259702e-06,
+      "loss": 0.9377,
+      "step": 18827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.114437947483605e-06,
+      "loss": 0.8936,
+      "step": 18828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.113596464380858e-06,
+      "loss": 0.9254,
+      "step": 18829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1127551289672215e-06,
+      "loss": 0.9986,
+      "step": 18830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1119139412584476e-06,
+      "loss": 0.8843,
+      "step": 18831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1110729012702935e-06,
+      "loss": 0.9403,
+      "step": 18832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.110232009018508e-06,
+      "loss": 0.9253,
+      "step": 18833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10939126451884e-06,
+      "loss": 0.9118,
+      "step": 18834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.108550667787028e-06,
+      "loss": 0.9875,
+      "step": 18835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.107710218838823e-06,
+      "loss": 1.0494,
+      "step": 18836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.106869917689957e-06,
+      "loss": 1.0235,
+      "step": 18837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10602976435617e-06,
+      "loss": 0.9801,
+      "step": 18838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1051897588531955e-06,
+      "loss": 0.9389,
+      "step": 18839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.104349901196759e-06,
+      "loss": 0.9745,
+      "step": 18840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.103510191402598e-06,
+      "loss": 0.8754,
+      "step": 18841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.102670629486428e-06,
+      "loss": 0.9405,
+      "step": 18842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10183121546398e-06,
+      "loss": 0.9791,
+      "step": 18843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.100991949350968e-06,
+      "loss": 1.0269,
+      "step": 18844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.10015283116311e-06,
+      "loss": 0.9266,
+      "step": 18845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.099313860916123e-06,
+      "loss": 0.9246,
+      "step": 18846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.098475038625715e-06,
+      "loss": 0.9708,
+      "step": 18847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0976363643076e-06,
+      "loss": 0.932,
+      "step": 18848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.096797837977475e-06,
+      "loss": 0.9791,
+      "step": 18849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.095959459651051e-06,
+      "loss": 1.0217,
+      "step": 18850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.095121229344025e-06,
+      "loss": 0.9682,
+      "step": 18851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.094283147072095e-06,
+      "loss": 0.9564,
+      "step": 18852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.093445212850955e-06,
+      "loss": 0.9351,
+      "step": 18853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.092607426696301e-06,
+      "loss": 0.9872,
+      "step": 18854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.091769788623819e-06,
+      "loss": 0.9301,
+      "step": 18855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.090932298649193e-06,
+      "loss": 0.9574,
+      "step": 18856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.09009495678811e-06,
+      "loss": 0.9961,
+      "step": 18857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.089257763056252e-06,
+      "loss": 0.9596,
+      "step": 18858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.088420717469297e-06,
+      "loss": 0.9835,
+      "step": 18859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.087583820042915e-06,
+      "loss": 0.9762,
+      "step": 18860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.086747070792786e-06,
+      "loss": 0.9503,
+      "step": 18861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.085910469734572e-06,
+      "loss": 0.9444,
+      "step": 18862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.08507401688395e-06,
+      "loss": 0.9783,
+      "step": 18863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.084237712256577e-06,
+      "loss": 0.9099,
+      "step": 18864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.083401555868115e-06,
+      "loss": 0.934,
+      "step": 18865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.082565547734229e-06,
+      "loss": 0.9111,
+      "step": 18866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.081729687870563e-06,
+      "loss": 0.9809,
+      "step": 18867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.080893976292782e-06,
+      "loss": 0.96,
+      "step": 18868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0800584130165315e-06,
+      "loss": 0.9241,
+      "step": 18869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.079222998057457e-06,
+      "loss": 1.0017,
+      "step": 18870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.078387731431208e-06,
+      "loss": 0.9559,
+      "step": 18871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.077552613153423e-06,
+      "loss": 0.9498,
+      "step": 18872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.076717643239742e-06,
+      "loss": 1.0363,
+      "step": 18873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075882821705805e-06,
+      "loss": 0.9425,
+      "step": 18874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0750481485672416e-06,
+      "loss": 0.9468,
+      "step": 18875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.074213623839682e-06,
+      "loss": 0.9454,
+      "step": 18876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.073379247538759e-06,
+      "loss": 0.8822,
+      "step": 18877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.072545019680094e-06,
+      "loss": 0.9325,
+      "step": 18878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.071710940279314e-06,
+      "loss": 0.9753,
+      "step": 18879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0708770093520305e-06,
+      "loss": 0.96,
+      "step": 18880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.070043226913871e-06,
+      "loss": 0.8954,
+      "step": 18881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.069209592980443e-06,
+      "loss": 1.0128,
+      "step": 18882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.068376107567359e-06,
+      "loss": 0.9439,
+      "step": 18883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0675427706902314e-06,
+      "loss": 0.9527,
+      "step": 18884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.066709582364659e-06,
+      "loss": 0.9666,
+      "step": 18885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.065876542606253e-06,
+      "loss": 0.9974,
+      "step": 18886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.065043651430605e-06,
+      "loss": 0.9633,
+      "step": 18887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.064210908853325e-06,
+      "loss": 0.9965,
+      "step": 18888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.063378314889996e-06,
+      "loss": 0.9411,
+      "step": 18889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.062545869556215e-06,
+      "loss": 0.9764,
+      "step": 18890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.06171357286757e-06,
+      "loss": 1.0057,
+      "step": 18891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.060881424839651e-06,
+      "loss": 0.9457,
+      "step": 18892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.060049425488037e-06,
+      "loss": 0.9957,
+      "step": 18893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.059217574828311e-06,
+      "loss": 0.9564,
+      "step": 18894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05838587287605e-06,
+      "loss": 1.0167,
+      "step": 18895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.05755431964683e-06,
+      "loss": 0.9888,
+      "step": 18896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0567229151562275e-06,
+      "loss": 1.0087,
+      "step": 18897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.055891659419801e-06,
+      "loss": 0.9286,
+      "step": 18898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.055060552453133e-06,
+      "loss": 0.9215,
+      "step": 18899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.054229594271776e-06,
+      "loss": 0.9878,
+      "step": 18900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.053398784891295e-06,
+      "loss": 0.9783,
+      "step": 18901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.052568124327248e-06,
+      "loss": 0.9636,
+      "step": 18902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0517376125951924e-06,
+      "loss": 0.9283,
+      "step": 18903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.050907249710683e-06,
+      "loss": 1.0136,
+      "step": 18904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.050077035689261e-06,
+      "loss": 0.9571,
+      "step": 18905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.049246970546486e-06,
+      "loss": 0.8782,
+      "step": 18906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.048417054297895e-06,
+      "loss": 0.9818,
+      "step": 18907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.047587286959032e-06,
+      "loss": 0.9629,
+      "step": 18908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.046757668545436e-06,
+      "loss": 0.9967,
+      "step": 18909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045928199072642e-06,
+      "loss": 0.9098,
+      "step": 18910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.045098878556189e-06,
+      "loss": 0.8859,
+      "step": 18911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0442697070116e-06,
+      "loss": 0.9053,
+      "step": 18912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.043440684454406e-06,
+      "loss": 0.9909,
+      "step": 18913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.042611810900132e-06,
+      "loss": 0.9305,
+      "step": 18914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.041783086364303e-06,
+      "loss": 1.0007,
+      "step": 18915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.040954510862435e-06,
+      "loss": 0.9179,
+      "step": 18916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.04012608441005e-06,
+      "loss": 0.9384,
+      "step": 18917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.039297807022652e-06,
+      "loss": 0.9215,
+      "step": 18918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.038469678715765e-06,
+      "loss": 0.9515,
+      "step": 18919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0376416995048865e-06,
+      "loss": 0.9902,
+      "step": 18920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0368138694055275e-06,
+      "loss": 0.8991,
+      "step": 18921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.035986188433192e-06,
+      "loss": 0.9341,
+      "step": 18922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.035158656603373e-06,
+      "loss": 0.9116,
+      "step": 18923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.034331273931578e-06,
+      "loss": 0.8917,
+      "step": 18924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.033504040433289e-06,
+      "loss": 0.9837,
+      "step": 18925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.032676956124012e-06,
+      "loss": 0.994,
+      "step": 18926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.031850021019226e-06,
+      "loss": 0.9509,
+      "step": 18927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.031023235134419e-06,
+      "loss": 1.0029,
+      "step": 18928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0301965984850734e-06,
+      "loss": 0.9832,
+      "step": 18929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.029370111086672e-06,
+      "loss": 0.9388,
+      "step": 18930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.028543772954695e-06,
+      "loss": 0.9455,
+      "step": 18931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02771758410461e-06,
+      "loss": 0.9444,
+      "step": 18932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.026891544551894e-06,
+      "loss": 0.98,
+      "step": 18933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.026065654312014e-06,
+      "loss": 0.9392,
+      "step": 18934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02523991340044e-06,
+      "loss": 0.9976,
+      "step": 18935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.024414321832631e-06,
+      "loss": 0.9472,
+      "step": 18936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.023588879624055e-06,
+      "loss": 0.9604,
+      "step": 18937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.022763586790162e-06,
+      "loss": 0.8804,
+      "step": 18938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.02193844334641e-06,
+      "loss": 1.0092,
+      "step": 18939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0211134493082526e-06,
+      "loss": 1.0,
+      "step": 18940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.020288604691139e-06,
+      "loss": 0.9996,
+      "step": 18941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.019463909510519e-06,
+      "loss": 0.9735,
+      "step": 18942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.018639363781829e-06,
+      "loss": 0.9219,
+      "step": 18943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.01781496752052e-06,
+      "loss": 0.8931,
+      "step": 18944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.016990720742021e-06,
+      "loss": 0.9469,
+      "step": 18945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.016166623461778e-06,
+      "loss": 0.9278,
+      "step": 18946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.015342675695215e-06,
+      "loss": 0.9633,
+      "step": 18947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.014518877457766e-06,
+      "loss": 0.9102,
+      "step": 18948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.013695228764861e-06,
+      "loss": 0.9801,
+      "step": 18949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.012871729631915e-06,
+      "loss": 0.9537,
+      "step": 18950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0120483800743615e-06,
+      "loss": 0.9499,
+      "step": 18951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0112251801076106e-06,
+      "loss": 0.9688,
+      "step": 18952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.010402129747083e-06,
+      "loss": 0.9398,
+      "step": 18953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.009579229008192e-06,
+      "loss": 1.0637,
+      "step": 18954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.008756477906346e-06,
+      "loss": 0.9512,
+      "step": 18955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007933876456952e-06,
+      "loss": 0.9438,
+      "step": 18956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.007111424675419e-06,
+      "loss": 0.9734,
+      "step": 18957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0062891225771464e-06,
+      "loss": 1.0043,
+      "step": 18958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0054669701775324e-06,
+      "loss": 0.9507,
+      "step": 18959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.004644967491977e-06,
+      "loss": 0.9272,
+      "step": 18960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003823114535868e-06,
+      "loss": 0.89,
+      "step": 18961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.003001411324603e-06,
+      "loss": 0.9219,
+      "step": 18962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.002179857873564e-06,
+      "loss": 1.013,
+      "step": 18963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.001358454198144e-06,
+      "loss": 0.9129,
+      "step": 18964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.000537200313717e-06,
+      "loss": 0.9379,
+      "step": 18965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.999716096235667e-06,
+      "loss": 0.9303,
+      "step": 18966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.998895141979371e-06,
+      "loss": 0.9343,
+      "step": 18967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.998074337560202e-06,
+      "loss": 0.9634,
+      "step": 18968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.997253682993534e-06,
+      "loss": 0.8823,
+      "step": 18969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.99643317829473e-06,
+      "loss": 0.9865,
+      "step": 18970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.995612823479159e-06,
+      "loss": 0.9759,
+      "step": 18971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.994792618562185e-06,
+      "loss": 1.042,
+      "step": 18972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.993972563559165e-06,
+      "loss": 1.0259,
+      "step": 18973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.993152658485459e-06,
+      "loss": 0.9431,
+      "step": 18974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.992332903356421e-06,
+      "loss": 0.9771,
+      "step": 18975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9915132981874005e-06,
+      "loss": 0.9681,
+      "step": 18976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.990693842993747e-06,
+      "loss": 0.9868,
+      "step": 18977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.989874537790808e-06,
+      "loss": 0.9302,
+      "step": 18978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.989055382593925e-06,
+      "loss": 0.9821,
+      "step": 18979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.988236377418442e-06,
+      "loss": 0.9365,
+      "step": 18980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.987417522279688e-06,
+      "loss": 0.9583,
+      "step": 18981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.986598817193008e-06,
+      "loss": 0.9097,
+      "step": 18982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.985780262173725e-06,
+      "loss": 0.9657,
+      "step": 18983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9849618572371785e-06,
+      "loss": 1.0274,
+      "step": 18984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.984143602398685e-06,
+      "loss": 0.9053,
+      "step": 18985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.983325497673572e-06,
+      "loss": 0.9555,
+      "step": 18986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.982507543077162e-06,
+      "loss": 0.9668,
+      "step": 18987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.981689738624767e-06,
+      "loss": 0.9769,
+      "step": 18988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.980872084331709e-06,
+      "loss": 0.9211,
+      "step": 18989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9800545802132975e-06,
+      "loss": 0.9324,
+      "step": 18990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.979237226284839e-06,
+      "loss": 0.9323,
+      "step": 18991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9784200225616446e-06,
+      "loss": 0.9595,
+      "step": 18992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.977602969059015e-06,
+      "loss": 0.8695,
+      "step": 18993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.976786065792253e-06,
+      "loss": 1.0287,
+      "step": 18994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.975969312776659e-06,
+      "loss": 0.9016,
+      "step": 18995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.975152710027522e-06,
+      "loss": 0.9541,
+      "step": 18996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.974336257560139e-06,
+      "loss": 0.8996,
+      "step": 18997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9735199553898e-06,
+      "loss": 1.0036,
+      "step": 18998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.972703803531789e-06,
+      "loss": 0.9729,
+      "step": 18999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.971887802001396e-06,
+      "loss": 0.9787,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "eval_text_loss": 0.9290885925292969,
+      "eval_text_runtime": 1.5852,
+      "eval_text_samples_per_second": 46.682,
+      "eval_text_steps_per_second": 1.262,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "eval_svit_loss": 1.0777496099472046,
+      "eval_svit_runtime": 3.613,
+      "eval_svit_samples_per_second": 188.484,
+      "eval_svit_steps_per_second": 3.045,
+      "step": 19000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.971071950813892e-06,
+      "loss": 0.9875,
+      "step": 19001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.970256249984567e-06,
+      "loss": 0.9705,
+      "step": 19002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.969440699528691e-06,
+      "loss": 0.9618,
+      "step": 19003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9686252994615355e-06,
+      "loss": 1.0151,
+      "step": 19004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.967810049798372e-06,
+      "loss": 0.9544,
+      "step": 19005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9669949505544694e-06,
+      "loss": 1.0165,
+      "step": 19006
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9661800017450925e-06,
+      "loss": 0.8808,
+      "step": 19007
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.965365203385496e-06,
+      "loss": 0.941,
+      "step": 19008
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.964550555490949e-06,
+      "loss": 0.9124,
+      "step": 19009
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9637360580767005e-06,
+      "loss": 0.9545,
+      "step": 19010
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.962921711158005e-06,
+      "loss": 0.9309,
+      "step": 19011
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.962107514750113e-06,
+      "loss": 0.9401,
+      "step": 19012
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.961293468868274e-06,
+      "loss": 0.9269,
+      "step": 19013
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.960479573527731e-06,
+      "loss": 0.9412,
+      "step": 19014
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.959665828743725e-06,
+      "loss": 1.0115,
+      "step": 19015
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.958852234531495e-06,
+      "loss": 0.9013,
+      "step": 19016
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.958038790906279e-06,
+      "loss": 0.976,
+      "step": 19017
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9572254978833085e-06,
+      "loss": 0.9337,
+      "step": 19018
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.956412355477816e-06,
+      "loss": 1.0616,
+      "step": 19019
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.95559936370503e-06,
+      "loss": 0.9669,
+      "step": 19020
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.954786522580168e-06,
+      "loss": 0.9206,
+      "step": 19021
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953973832118465e-06,
+      "loss": 1.0098,
+      "step": 19022
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.953161292335129e-06,
+      "loss": 0.9432,
+      "step": 19023
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.952348903245382e-06,
+      "loss": 1.0008,
+      "step": 19024
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.951536664864439e-06,
+      "loss": 0.9487,
+      "step": 19025
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.950724577207502e-06,
+      "loss": 0.9121,
+      "step": 19026
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9499126402897905e-06,
+      "loss": 0.9501,
+      "step": 19027
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9491008541265e-06,
+      "loss": 0.9377,
+      "step": 19028
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9482892187328415e-06,
+      "loss": 0.9323,
+      "step": 19029
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.947477734124008e-06,
+      "loss": 0.95,
+      "step": 19030
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.946666400315198e-06,
+      "loss": 0.9193,
+      "step": 19031
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.945855217321608e-06,
+      "loss": 0.9808,
+      "step": 19032
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9450441851584265e-06,
+      "loss": 0.9084,
+      "step": 19033
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.944233303840841e-06,
+      "loss": 0.8936,
+      "step": 19034
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.943422573384038e-06,
+      "loss": 0.9208,
+      "step": 19035
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.942611993803201e-06,
+      "loss": 0.9663,
+      "step": 19036
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.941801565113508e-06,
+      "loss": 0.979,
+      "step": 19037
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.940991287330139e-06,
+      "loss": 0.9263,
+      "step": 19038
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.940181160468262e-06,
+      "loss": 0.9373,
+      "step": 19039
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.939371184543057e-06,
+      "loss": 0.9969,
+      "step": 19040
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9385613595696855e-06,
+      "loss": 1.0147,
+      "step": 19041
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.937751685563315e-06,
+      "loss": 0.9073,
+      "step": 19042
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936942162539112e-06,
+      "loss": 0.928,
+      "step": 19043
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.936132790512227e-06,
+      "loss": 0.9548,
+      "step": 19044
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.935323569497831e-06,
+      "loss": 0.9663,
+      "step": 19045
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.934514499511062e-06,
+      "loss": 1.0048,
+      "step": 19046
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9337055805670865e-06,
+      "loss": 0.9691,
+      "step": 19047
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.932896812681045e-06,
+      "loss": 1.0127,
+      "step": 19048
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.9320881958680864e-06,
+      "loss": 1.0187,
+      "step": 19049
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.931279730143351e-06,
+      "loss": 1.0275,
+      "step": 19050
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.93047141552198e-06,
+      "loss": 0.9395,
+      "step": 19051
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.929663252019113e-06,
+      "loss": 0.9217,
+      "step": 19052
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.928855239649882e-06,
+      "loss": 1.0037,
+      "step": 19053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9280473784294195e-06,
+      "loss": 0.9573,
+      "step": 19054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.927239668372854e-06,
+      "loss": 0.9463,
+      "step": 19055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9264321094953115e-06,
+      "loss": 0.9805,
+      "step": 19056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.925624701811916e-06,
+      "loss": 0.9784,
+      "step": 19057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.92481744533779e-06,
+      "loss": 0.9425,
+      "step": 19058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.924010340088042e-06,
+      "loss": 0.9761,
+      "step": 19059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9232033860778e-06,
+      "loss": 1.0017,
+      "step": 19060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9223965833221655e-06,
+      "loss": 0.9344,
+      "step": 19061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.921589931836252e-06,
+      "loss": 0.9484,
+      "step": 19062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9207834316351675e-06,
+      "loss": 1.0037,
+      "step": 19063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.919977082734006e-06,
+      "loss": 0.9161,
+      "step": 19064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.919170885147881e-06,
+      "loss": 0.9561,
+      "step": 19065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.918364838891878e-06,
+      "loss": 0.9599,
+      "step": 19066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.917558943981102e-06,
+      "loss": 0.9621,
+      "step": 19067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.916753200430641e-06,
+      "loss": 0.9755,
+      "step": 19068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.915947608255582e-06,
+      "loss": 0.996,
+      "step": 19069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9151421674710166e-06,
+      "loss": 0.9409,
+      "step": 19070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.914336878092018e-06,
+      "loss": 0.9641,
+      "step": 19071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.913531740133683e-06,
+      "loss": 0.932,
+      "step": 19072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.912726753611075e-06,
+      "loss": 0.9376,
+      "step": 19073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.911921918539275e-06,
+      "loss": 0.9748,
+      "step": 19074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.911117234933355e-06,
+      "loss": 0.9454,
+      "step": 19075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.910312702808385e-06,
+      "loss": 1.0434,
+      "step": 19076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9095083221794294e-06,
+      "loss": 0.9576,
+      "step": 19077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.908704093061555e-06,
+      "loss": 0.9735,
+      "step": 19078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9079000154698204e-06,
+      "loss": 0.9125,
+      "step": 19079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.907096089419281e-06,
+      "loss": 0.982,
+      "step": 19080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9062923149249965e-06,
+      "loss": 1.0197,
+      "step": 19081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.905488692002018e-06,
+      "loss": 0.9303,
+      "step": 19082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.904685220665396e-06,
+      "loss": 0.9448,
+      "step": 19083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.903881900930171e-06,
+      "loss": 0.9358,
+      "step": 19084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.903078732811398e-06,
+      "loss": 0.9727,
+      "step": 19085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9022757163241065e-06,
+      "loss": 1.0098,
+      "step": 19086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.901472851483341e-06,
+      "loss": 0.9604,
+      "step": 19087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9006701383041345e-06,
+      "loss": 0.9011,
+      "step": 19088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.899867576801522e-06,
+      "loss": 0.9509,
+      "step": 19089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.899065166990534e-06,
+      "loss": 0.9608,
+      "step": 19090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.898262908886188e-06,
+      "loss": 0.9676,
+      "step": 19091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.897460802503521e-06,
+      "loss": 0.909,
+      "step": 19092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.896658847857544e-06,
+      "loss": 0.98,
+      "step": 19093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8958570449632805e-06,
+      "loss": 0.9068,
+      "step": 19094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.895055393835744e-06,
+      "loss": 0.9344,
+      "step": 19095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.894253894489947e-06,
+      "loss": 1.0124,
+      "step": 19096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.893452546940903e-06,
+      "loss": 0.9297,
+      "step": 19097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8926513512036126e-06,
+      "loss": 0.9862,
+      "step": 19098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.891850307293084e-06,
+      "loss": 0.9311,
+      "step": 19099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.891049415224315e-06,
+      "loss": 0.9624,
+      "step": 19100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.890248675012311e-06,
+      "loss": 0.9472,
+      "step": 19101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.889448086672057e-06,
+      "loss": 0.9252,
+      "step": 19102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.888647650218556e-06,
+      "loss": 1.0001,
+      "step": 19103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.887847365666787e-06,
+      "loss": 0.9129,
+      "step": 19104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88704723303175e-06,
+      "loss": 0.9074,
+      "step": 19105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.886247252328419e-06,
+      "loss": 0.919,
+      "step": 19106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8854474235717784e-06,
+      "loss": 0.9729,
+      "step": 19107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88464774677681e-06,
+      "loss": 0.9337,
+      "step": 19108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.883848221958481e-06,
+      "loss": 0.9247,
+      "step": 19109
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.883048849131774e-06,
+      "loss": 0.9618,
+      "step": 19110
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.882249628311652e-06,
+      "loss": 0.9777,
+      "step": 19111
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.881450559513083e-06,
+      "loss": 1.0168,
+      "step": 19112
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.880651642751034e-06,
+      "loss": 0.9566,
+      "step": 19113
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.879852878040466e-06,
+      "loss": 0.9137,
+      "step": 19114
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.879054265396334e-06,
+      "loss": 0.9629,
+      "step": 19115
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8782558048336e-06,
+      "loss": 0.9844,
+      "step": 19116
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.87745749636721e-06,
+      "loss": 1.0052,
+      "step": 19117
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.876659340012116e-06,
+      "loss": 0.9079,
+      "step": 19118
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.875861335783267e-06,
+      "loss": 0.9236,
+      "step": 19119
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.875063483695605e-06,
+      "loss": 0.8833,
+      "step": 19120
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.874265783764076e-06,
+      "loss": 0.9515,
+      "step": 19121
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.873468236003609e-06,
+      "loss": 0.9572,
+      "step": 19122
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.872670840429152e-06,
+      "loss": 0.9869,
+      "step": 19123
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.87187359705563e-06,
+      "loss": 0.9974,
+      "step": 19124
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.871076505897973e-06,
+      "loss": 0.8985,
+      "step": 19125
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.870279566971111e-06,
+      "loss": 0.9383,
+      "step": 19126
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.869482780289967e-06,
+      "loss": 0.9913,
+      "step": 19127
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.868686145869466e-06,
+      "loss": 0.926,
+      "step": 19128
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8678896637245166e-06,
+      "loss": 0.9742,
+      "step": 19129
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.867093333870048e-06,
+      "loss": 0.9193,
+      "step": 19130
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8662971563209645e-06,
+      "loss": 0.9702,
+      "step": 19131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.865501131092176e-06,
+      "loss": 0.8816,
+      "step": 19132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.864705258198593e-06,
+      "loss": 0.8915,
+      "step": 19133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.863909537655119e-06,
+      "loss": 0.9236,
+      "step": 19134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8631139694766564e-06,
+      "loss": 0.9981,
+      "step": 19135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.862318553678101e-06,
+      "loss": 0.8914,
+      "step": 19136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8615232902743495e-06,
+      "loss": 0.9255,
+      "step": 19137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.860728179280297e-06,
+      "loss": 0.9367,
+      "step": 19138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.859933220710831e-06,
+      "loss": 0.9312,
+      "step": 19139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85913841458084e-06,
+      "loss": 0.9701,
+      "step": 19140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85834376090521e-06,
+      "loss": 0.9164,
+      "step": 19141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.857549259698817e-06,
+      "loss": 0.9378,
+      "step": 19142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.856754910976548e-06,
+      "loss": 0.9541,
+      "step": 19143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.855960714753272e-06,
+      "loss": 0.9857,
+      "step": 19144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.855166671043864e-06,
+      "loss": 0.9266,
+      "step": 19145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.854372779863197e-06,
+      "loss": 0.9666,
+      "step": 19146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.85357904122613e-06,
+      "loss": 0.9765,
+      "step": 19147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.852785455147539e-06,
+      "loss": 0.9699,
+      "step": 19148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.851992021642272e-06,
+      "loss": 0.9466,
+      "step": 19149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.851198740725203e-06,
+      "loss": 0.9692,
+      "step": 19150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.850405612411177e-06,
+      "loss": 0.9029,
+      "step": 19151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.849612636715049e-06,
+      "loss": 0.9564,
+      "step": 19152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84881981365167e-06,
+      "loss": 0.9633,
+      "step": 19153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.848027143235888e-06,
+      "loss": 1.0128,
+      "step": 19154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.847234625482548e-06,
+      "loss": 1.0238,
+      "step": 19155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.846442260406487e-06,
+      "loss": 1.0197,
+      "step": 19156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.845650048022549e-06,
+      "loss": 0.9184,
+      "step": 19157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844857988345567e-06,
+      "loss": 0.9651,
+      "step": 19158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.844066081390374e-06,
+      "loss": 0.9383,
+      "step": 19159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.843274327171801e-06,
+      "loss": 0.9818,
+      "step": 19160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.842482725704677e-06,
+      "loss": 0.9223,
+      "step": 19161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.841691277003823e-06,
+      "loss": 0.9475,
+      "step": 19162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84089998108406e-06,
+      "loss": 0.9504,
+      "step": 19163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.840108837960209e-06,
+      "loss": 0.9757,
+      "step": 19164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8393178476470864e-06,
+      "loss": 0.949,
+      "step": 19165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.838527010159506e-06,
+      "loss": 1.0753,
+      "step": 19166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8377363255122695e-06,
+      "loss": 0.964,
+      "step": 19167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.836945793720195e-06,
+      "loss": 0.9116,
+      "step": 19168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.836155414798081e-06,
+      "loss": 0.947,
+      "step": 19169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.83536518876073e-06,
+      "loss": 0.9655,
+      "step": 19170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.83457511562294e-06,
+      "loss": 0.9373,
+      "step": 19171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8337851953995076e-06,
+      "loss": 0.9653,
+      "step": 19172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.832995428105229e-06,
+      "loss": 0.8982,
+      "step": 19173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.832205813754888e-06,
+      "loss": 0.9879,
+      "step": 19174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.831416352363274e-06,
+      "loss": 0.9181,
+      "step": 19175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.830627043945172e-06,
+      "loss": 0.9435,
+      "step": 19176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.829837888515363e-06,
+      "loss": 0.8859,
+      "step": 19177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8290488860886265e-06,
+      "loss": 0.9427,
+      "step": 19178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.828260036679739e-06,
+      "loss": 0.9665,
+      "step": 19179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.827471340303468e-06,
+      "loss": 0.9646,
+      "step": 19180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.826682796974592e-06,
+      "loss": 0.9348,
+      "step": 19181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82589440670787e-06,
+      "loss": 0.9279,
+      "step": 19182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.825106169518071e-06,
+      "loss": 1.0285,
+      "step": 19183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.824318085419956e-06,
+      "loss": 0.9586,
+      "step": 19184
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8235301544282774e-06,
+      "loss": 0.9398,
+      "step": 19185
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.822742376557802e-06,
+      "loss": 0.9773,
+      "step": 19186
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82195475182327e-06,
+      "loss": 0.9317,
+      "step": 19187
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.821167280239444e-06,
+      "loss": 0.9656,
+      "step": 19188
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.820379961821062e-06,
+      "loss": 1.0205,
+      "step": 19189
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.819592796582871e-06,
+      "loss": 0.924,
+      "step": 19190
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.818805784539612e-06,
+      "loss": 0.9443,
+      "step": 19191
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8180189257060245e-06,
+      "loss": 0.9994,
+      "step": 19192
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.817232220096844e-06,
+      "loss": 0.9769,
+      "step": 19193
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8164456677268e-06,
+      "loss": 1.0286,
+      "step": 19194
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.815659268610627e-06,
+      "loss": 0.9506,
+      "step": 19195
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814873022763047e-06,
+      "loss": 0.9876,
+      "step": 19196
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.814086930198787e-06,
+      "loss": 0.9483,
+      "step": 19197
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8133009909325693e-06,
+      "loss": 0.9369,
+      "step": 19198
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8125152049791117e-06,
+      "loss": 0.908,
+      "step": 19199
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8117295723531267e-06,
+      "loss": 0.9666,
+      "step": 19200
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.81094409306933e-06,
+      "loss": 0.9984,
+      "step": 19201
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.810158767142429e-06,
+      "loss": 0.9494,
+      "step": 19202
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8093735945871315e-06,
+      "loss": 0.9409,
+      "step": 19203
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8085885754181454e-06,
+      "loss": 0.983,
+      "step": 19204
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.807803709650162e-06,
+      "loss": 0.927,
+      "step": 19205
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8070189972978925e-06,
+      "loss": 0.997,
+      "step": 19206
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.806234438376019e-06,
+      "loss": 0.9838,
+      "step": 19207
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.805450032899245e-06,
+      "loss": 1.0178,
+      "step": 19208
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8046657808822544e-06,
+      "loss": 1.0556,
+      "step": 19209
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.803881682339733e-06,
+      "loss": 0.9395,
+      "step": 19210
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.803097737286371e-06,
+      "loss": 1.0312,
+      "step": 19211
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.80231394573684e-06,
+      "loss": 1.017,
+      "step": 19212
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8015303077058273e-06,
+      "loss": 0.9288,
+      "step": 19213
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.8007468232080017e-06,
+      "loss": 0.9527,
+      "step": 19214
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7999634922580377e-06,
+      "loss": 0.8964,
+      "step": 19215
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7991803148706054e-06,
+      "loss": 1.0044,
+      "step": 19216
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.798397291060371e-06,
+      "loss": 0.9626,
+      "step": 19217
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7976144208419974e-06,
+      "loss": 0.9964,
+      "step": 19218
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7968317042301495e-06,
+      "loss": 0.9381,
+      "step": 19219
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.79604914123948e-06,
+      "loss": 0.9387,
+      "step": 19220
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.795266731884645e-06,
+      "loss": 0.9637,
+      "step": 19221
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.794484476180298e-06,
+      "loss": 0.9111,
+      "step": 19222
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7937023741410887e-06,
+      "loss": 0.9886,
+      "step": 19223
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7929204257816645e-06,
+      "loss": 0.9224,
+      "step": 19224
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7921386311166633e-06,
+      "loss": 0.9688,
+      "step": 19225
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.791356990160734e-06,
+      "loss": 0.8774,
+      "step": 19226
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7905755029285087e-06,
+      "loss": 0.9117,
+      "step": 19227
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7897941694346245e-06,
+      "loss": 1.0037,
+      "step": 19228
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7890129896937145e-06,
+      "loss": 0.9533,
+      "step": 19229
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7882319637204015e-06,
+      "loss": 0.9866,
+      "step": 19230
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7874510915293218e-06,
+      "loss": 0.9298,
+      "step": 19231
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7866703731350887e-06,
+      "loss": 0.9943,
+      "step": 19232
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7858898085523323e-06,
+      "loss": 0.9704,
+      "step": 19233
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7851093977956643e-06,
+      "loss": 1.0034,
+      "step": 19234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7843291408797002e-06,
+      "loss": 0.9562,
+      "step": 19235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.783549037819053e-06,
+      "loss": 0.9788,
+      "step": 19236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7827690886283342e-06,
+      "loss": 0.9408,
+      "step": 19237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.781989293322145e-06,
+      "loss": 0.9348,
+      "step": 19238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.78120965191509e-06,
+      "loss": 0.8797,
+      "step": 19239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7804301644217716e-06,
+      "loss": 0.9567,
+      "step": 19240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.779650830856785e-06,
+      "loss": 0.9852,
+      "step": 19241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.778871651234729e-06,
+      "loss": 0.9702,
+      "step": 19242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.778092625570187e-06,
+      "loss": 0.9594,
+      "step": 19243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7773137538777587e-06,
+      "loss": 0.9388,
+      "step": 19244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.776535036172019e-06,
+      "loss": 0.9454,
+      "step": 19245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7757564724675613e-06,
+      "loss": 0.9116,
+      "step": 19246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.774978062778959e-06,
+      "loss": 0.9721,
+      "step": 19247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.774199807120791e-06,
+      "loss": 1.0093,
+      "step": 19248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.773421705507635e-06,
+      "loss": 0.9453,
+      "step": 19249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.772643757954054e-06,
+      "loss": 0.9212,
+      "step": 19250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7718659644746294e-06,
+      "loss": 0.9733,
+      "step": 19251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.771088325083916e-06,
+      "loss": 0.968,
+      "step": 19252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7703108397964805e-06,
+      "loss": 0.9735,
+      "step": 19253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.769533508626884e-06,
+      "loss": 0.9787,
+      "step": 19254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.768756331589682e-06,
+      "loss": 0.95,
+      "step": 19255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7679793086994325e-06,
+      "loss": 0.9436,
+      "step": 19256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7672024399706813e-06,
+      "loss": 0.9756,
+      "step": 19257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7664257254179782e-06,
+      "loss": 0.9522,
+      "step": 19258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7656491650558707e-06,
+      "loss": 0.9234,
+      "step": 19259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.764872758898901e-06,
+      "loss": 0.9124,
+      "step": 19260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7640965069616076e-06,
+      "loss": 0.8947,
+      "step": 19261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.763320409258532e-06,
+      "loss": 0.9571,
+      "step": 19262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7625444658042e-06,
+      "loss": 0.9463,
+      "step": 19263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7617686766131515e-06,
+      "loss": 1.0022,
+      "step": 19264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.760993041699909e-06,
+      "loss": 0.9469,
+      "step": 19265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.760217561078998e-06,
+      "loss": 0.9675,
+      "step": 19266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7594422347649466e-06,
+      "loss": 0.9658,
+      "step": 19267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7586670627722648e-06,
+      "loss": 0.9965,
+      "step": 19268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7578920451154793e-06,
+      "loss": 0.9622,
+      "step": 19269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.757117181809094e-06,
+      "loss": 0.9579,
+      "step": 19270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75634247286763e-06,
+      "loss": 0.9463,
+      "step": 19271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7555679183055896e-06,
+      "loss": 1.0039,
+      "step": 19272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7547935181374773e-06,
+      "loss": 0.9463,
+      "step": 19273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.754019272377797e-06,
+      "loss": 0.9898,
+      "step": 19274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7532451810410487e-06,
+      "loss": 1.0196,
+      "step": 19275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.75247124414173e-06,
+      "loss": 0.911,
+      "step": 19276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7516974616943307e-06,
+      "loss": 0.9633,
+      "step": 19277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.750923833713342e-06,
+      "loss": 0.9606,
+      "step": 19278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.750150360213253e-06,
+      "loss": 0.9825,
+      "step": 19279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7493770412085485e-06,
+      "loss": 0.936,
+      "step": 19280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7486038767137113e-06,
+      "loss": 0.9232,
+      "step": 19281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.747830866743222e-06,
+      "loss": 0.948,
+      "step": 19282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7470580113115518e-06,
+      "loss": 0.8963,
+      "step": 19283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7462853104331764e-06,
+      "loss": 0.9529,
+      "step": 19284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7455127641225666e-06,
+      "loss": 0.9903,
+      "step": 19285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.74474037239419e-06,
+      "loss": 0.9783,
+      "step": 19286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.7439681352625135e-06,
+      "loss": 0.9352,
+      "step": 19287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.743196052741991e-06,
+      "loss": 0.946,
+      "step": 19288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.742424124847092e-06,
+      "loss": 0.938,
+      "step": 19289
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.741652351592262e-06,
+      "loss": 0.9977,
+      "step": 19290
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.740880732991964e-06,
+      "loss": 0.9513,
+      "step": 19291
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7401092690606413e-06,
+      "loss": 0.9449,
+      "step": 19292
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7393379598127426e-06,
+      "loss": 1.0056,
+      "step": 19293
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7385668052627156e-06,
+      "loss": 1.002,
+      "step": 19294
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7377958054249945e-06,
+      "loss": 0.9674,
+      "step": 19295
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7370249603140263e-06,
+      "loss": 0.9171,
+      "step": 19296
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7362542699442407e-06,
+      "loss": 0.9297,
+      "step": 19297
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7354837343300724e-06,
+      "loss": 0.8896,
+      "step": 19298
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7347133534859505e-06,
+      "loss": 0.9494,
+      "step": 19299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7339431274263043e-06,
+      "loss": 1.0144,
+      "step": 19300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7331730561655558e-06,
+      "loss": 0.9662,
+      "step": 19301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7324031397181286e-06,
+      "loss": 0.8916,
+      "step": 19302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.731633378098436e-06,
+      "loss": 0.932,
+      "step": 19303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7308637713208977e-06,
+      "loss": 0.9051,
+      "step": 19304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7300943193999274e-06,
+      "loss": 0.9647,
+      "step": 19305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7293250223499266e-06,
+      "loss": 1.0164,
+      "step": 19306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7285558801853116e-06,
+      "loss": 0.9598,
+      "step": 19307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7277868929204788e-06,
+      "loss": 0.9941,
+      "step": 19308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.727018060569837e-06,
+      "loss": 0.9631,
+      "step": 19309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.726249383147777e-06,
+      "loss": 0.9741,
+      "step": 19310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.725480860668696e-06,
+      "loss": 0.9578,
+      "step": 19311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.724712493146986e-06,
+      "loss": 0.9228,
+      "step": 19312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7239442805970373e-06,
+      "loss": 0.9371,
+      "step": 19313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7231762230332393e-06,
+      "loss": 0.908,
+      "step": 19314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.722408320469968e-06,
+      "loss": 0.9047,
+      "step": 19315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.721640572921609e-06,
+      "loss": 0.9624,
+      "step": 19316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.720872980402538e-06,
+      "loss": 0.9601,
+      "step": 19317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7201055429271307e-06,
+      "loss": 0.9757,
+      "step": 19318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7193382605097597e-06,
+      "loss": 0.9481,
+      "step": 19319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7185711331647946e-06,
+      "loss": 0.9343,
+      "step": 19320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7178041609065976e-06,
+      "loss": 0.9909,
+      "step": 19321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7170373437495346e-06,
+      "loss": 0.8945,
+      "step": 19322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7162706817079643e-06,
+      "loss": 0.9389,
+      "step": 19323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.715504174796246e-06,
+      "loss": 0.9467,
+      "step": 19324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.714737823028736e-06,
+      "loss": 0.9484,
+      "step": 19325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.713971626419777e-06,
+      "loss": 0.9394,
+      "step": 19326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.713205584983729e-06,
+      "loss": 0.9441,
+      "step": 19327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7124396987349265e-06,
+      "loss": 0.9822,
+      "step": 19328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7116739676877248e-06,
+      "loss": 0.9338,
+      "step": 19329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7109083918564536e-06,
+      "loss": 0.9271,
+      "step": 19330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7101429712554537e-06,
+      "loss": 1.0211,
+      "step": 19331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7093777058990614e-06,
+      "loss": 0.9847,
+      "step": 19332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7086125958015994e-06,
+      "loss": 0.9701,
+      "step": 19333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7078476409774077e-06,
+      "loss": 0.9092,
+      "step": 19334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7070828414408035e-06,
+      "loss": 0.9273,
+      "step": 19335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7063181972061105e-06,
+      "loss": 0.9755,
+      "step": 19336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7055537082876506e-06,
+      "loss": 0.9651,
+      "step": 19337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.704789374699739e-06,
+      "loss": 0.9666,
+      "step": 19338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7040251964566888e-06,
+      "loss": 0.9672,
+      "step": 19339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7032611735728143e-06,
+      "loss": 1.0045,
+      "step": 19340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7024973060624186e-06,
+      "loss": 0.9809,
+      "step": 19341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.701733593939807e-06,
+      "loss": 0.9921,
+      "step": 19342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.700970037219285e-06,
+      "loss": 0.9204,
+      "step": 19343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7002066359151503e-06,
+      "loss": 1.0224,
+      "step": 19344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.699443390041701e-06,
+      "loss": 0.9341,
+      "step": 19345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6986802996132223e-06,
+      "loss": 0.9522,
+      "step": 19346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6979173646440174e-06,
+      "loss": 1.0267,
+      "step": 19347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6971545851483636e-06,
+      "loss": 0.9488,
+      "step": 19348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6963919611405497e-06,
+      "loss": 0.9812,
+      "step": 19349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.695629492634856e-06,
+      "loss": 0.9438,
+      "step": 19350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.694867179645563e-06,
+      "loss": 0.9544,
+      "step": 19351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6941050221869467e-06,
+      "loss": 1.036,
+      "step": 19352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6933430202732758e-06,
+      "loss": 0.9428,
+      "step": 19353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.692581173918827e-06,
+      "loss": 0.9702,
+      "step": 19354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.691819483137862e-06,
+      "loss": 0.9865,
+      "step": 19355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6910579479446477e-06,
+      "loss": 1.0454,
+      "step": 19356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6902965683534443e-06,
+      "loss": 0.9384,
+      "step": 19357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6895353443785097e-06,
+      "loss": 0.9187,
+      "step": 19358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.688774276034104e-06,
+      "loss": 0.9801,
+      "step": 19359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.688013363334473e-06,
+      "loss": 0.9745,
+      "step": 19360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.68725260629387e-06,
+      "loss": 0.9261,
+      "step": 19361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.686492004926541e-06,
+      "loss": 1.0193,
+      "step": 19362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.685731559246729e-06,
+      "loss": 0.9962,
+      "step": 19363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.684971269268677e-06,
+      "loss": 0.9502,
+      "step": 19364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6842111350066234e-06,
+      "loss": 0.9543,
+      "step": 19365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6834511564747978e-06,
+      "loss": 0.9529,
+      "step": 19366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6826913336874406e-06,
+      "loss": 0.9598,
+      "step": 19367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6819316666587745e-06,
+      "loss": 0.9182,
+      "step": 19368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.681172155403028e-06,
+      "loss": 0.9552,
+      "step": 19369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6804127999344276e-06,
+      "loss": 0.9258,
+      "step": 19370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.679653600267186e-06,
+      "loss": 0.8623,
+      "step": 19371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6788945564155287e-06,
+      "loss": 0.9638,
+      "step": 19372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.678135668393667e-06,
+      "loss": 0.9181,
+      "step": 19373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.677376936215812e-06,
+      "loss": 0.9285,
+      "step": 19374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6766183598961726e-06,
+      "loss": 0.9279,
+      "step": 19375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6758599394489557e-06,
+      "loss": 0.9409,
+      "step": 19376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6751016748883653e-06,
+      "loss": 0.9415,
+      "step": 19377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.674343566228601e-06,
+      "loss": 0.9334,
+      "step": 19378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.673585613483858e-06,
+      "loss": 0.9695,
+      "step": 19379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.672827816668331e-06,
+      "loss": 0.914,
+      "step": 19380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.672070175796212e-06,
+      "loss": 0.9769,
+      "step": 19381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.67131269088169e-06,
+      "loss": 0.9156,
+      "step": 19382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6705553619389515e-06,
+      "loss": 0.9491,
+      "step": 19383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6697981889821726e-06,
+      "loss": 0.9406,
+      "step": 19384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.669041172025543e-06,
+      "loss": 0.9736,
+      "step": 19385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6682843110832316e-06,
+      "loss": 0.8914,
+      "step": 19386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6675276061694147e-06,
+      "loss": 0.9366,
+      "step": 19387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.666771057298262e-06,
+      "loss": 0.9592,
+      "step": 19388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6660146644839438e-06,
+      "loss": 0.9177,
+      "step": 19389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6652584277406256e-06,
+      "loss": 0.929,
+      "step": 19390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.664502347082464e-06,
+      "loss": 0.9504,
+      "step": 19391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6637464225236254e-06,
+      "loss": 0.963,
+      "step": 19392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6629906540782604e-06,
+      "loss": 1.0482,
+      "step": 19393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.662235041760525e-06,
+      "loss": 0.9893,
+      "step": 19394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6614795855845687e-06,
+      "loss": 1.021,
+      "step": 19395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.660724285564539e-06,
+      "loss": 0.929,
+      "step": 19396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6599691417145835e-06,
+      "loss": 0.9863,
+      "step": 19397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.659214154048839e-06,
+      "loss": 0.9736,
+      "step": 19398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.658459322581446e-06,
+      "loss": 0.9074,
+      "step": 19399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.657704647326541e-06,
+      "loss": 0.9834,
+      "step": 19400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6569501282982558e-06,
+      "loss": 0.9519,
+      "step": 19401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6561957655107207e-06,
+      "loss": 0.9652,
+      "step": 19402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6554415589780655e-06,
+      "loss": 0.892,
+      "step": 19403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.654687508714408e-06,
+      "loss": 0.8774,
+      "step": 19404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.653933614733877e-06,
+      "loss": 0.914,
+      "step": 19405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6531798770505846e-06,
+      "loss": 1.0017,
+      "step": 19406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.652426295678647e-06,
+      "loss": 0.9909,
+      "step": 19407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6516728706321815e-06,
+      "loss": 0.9888,
+      "step": 19408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.650919601925288e-06,
+      "loss": 0.9156,
+      "step": 19409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6501664895720845e-06,
+      "loss": 0.9815,
+      "step": 19410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6494135335866644e-06,
+      "loss": 0.9795,
+      "step": 19411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.648660733983136e-06,
+      "loss": 1.0087,
+      "step": 19412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647908090775593e-06,
+      "loss": 0.9101,
+      "step": 19413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.647155603978131e-06,
+      "loss": 1.0002,
+      "step": 19414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.646403273604844e-06,
+      "loss": 0.9699,
+      "step": 19415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6456510996698136e-06,
+      "loss": 0.9117,
+      "step": 19416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6448990821871357e-06,
+      "loss": 0.9449,
+      "step": 19417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6441472211708877e-06,
+      "loss": 0.9837,
+      "step": 19418
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.643395516635149e-06,
+      "loss": 0.9572,
+      "step": 19419
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6426439685939995e-06,
+      "loss": 0.9976,
+      "step": 19420
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6418925770615123e-06,
+      "loss": 0.9562,
+      "step": 19421
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6411413420517596e-06,
+      "loss": 0.9549,
+      "step": 19422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6403902635788105e-06,
+      "loss": 0.9133,
+      "step": 19423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6396393416567277e-06,
+      "loss": 0.9404,
+      "step": 19424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6388885762995753e-06,
+      "loss": 0.921,
+      "step": 19425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6381379675214124e-06,
+      "loss": 0.9513,
+      "step": 19426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637387515336296e-06,
+      "loss": 0.9071,
+      "step": 19427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6366372197582834e-06,
+      "loss": 0.9302,
+      "step": 19428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635887080801416e-06,
+      "loss": 1.0142,
+      "step": 19429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635137098479754e-06,
+      "loss": 0.9525,
+      "step": 19430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6343872728073307e-06,
+      "loss": 0.9549,
+      "step": 19431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633637603798198e-06,
+      "loss": 0.997,
+      "step": 19432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632888091466388e-06,
+      "loss": 0.9821,
+      "step": 19433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63213873582594e-06,
+      "loss": 0.9061,
+      "step": 19434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6313895368908883e-06,
+      "loss": 0.9414,
+      "step": 19435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6306404946752574e-06,
+      "loss": 0.9186,
+      "step": 19436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629891609193084e-06,
+      "loss": 0.9354,
+      "step": 19437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629142880458384e-06,
+      "loss": 0.982,
+      "step": 19438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628394308485183e-06,
+      "loss": 0.9381,
+      "step": 19439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627645893287497e-06,
+      "loss": 1.053,
+      "step": 19440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6268976348793473e-06,
+      "loss": 0.931,
+      "step": 19441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62614953327474e-06,
+      "loss": 0.9659,
+      "step": 19442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.625401588487687e-06,
+      "loss": 0.9823,
+      "step": 19443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6246538005321954e-06,
+      "loss": 0.9075,
+      "step": 19444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62390616942227e-06,
+      "loss": 0.9463,
+      "step": 19445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.623158695171912e-06,
+      "loss": 0.9852,
+      "step": 19446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6224113777951152e-06,
+      "loss": 0.9927,
+      "step": 19447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6216642173058815e-06,
+      "loss": 0.9768,
+      "step": 19448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.620917213718195e-06,
+      "loss": 1.0042,
+      "step": 19449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6201703670460535e-06,
+      "loss": 1.0098,
+      "step": 19450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6194236773034367e-06,
+      "loss": 0.99,
+      "step": 19451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.61867714450433e-06,
+      "loss": 0.9247,
+      "step": 19452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6179307686627177e-06,
+      "loss": 0.9594,
+      "step": 19453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6171845497925666e-06,
+      "loss": 1.0294,
+      "step": 19454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6164384879078645e-06,
+      "loss": 0.997,
+      "step": 19455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6156925830225724e-06,
+      "loss": 0.9126,
+      "step": 19456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6149468351506635e-06,
+      "loss": 0.9277,
+      "step": 19457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.614201244306103e-06,
+      "loss": 0.9585,
+      "step": 19458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6134558105028528e-06,
+      "loss": 0.9129,
+      "step": 19459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.612710533754875e-06,
+      "loss": 0.9246,
+      "step": 19460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6119654140761267e-06,
+      "loss": 1.0295,
+      "step": 19461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.611220451480556e-06,
+      "loss": 0.9279,
+      "step": 19462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.610475645982119e-06,
+      "loss": 0.9329,
+      "step": 19463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.609730997594763e-06,
+      "loss": 0.9324,
+      "step": 19464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6089865063324317e-06,
+      "loss": 0.9589,
+      "step": 19465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.608242172209071e-06,
+      "loss": 0.9295,
+      "step": 19466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6074979952386115e-06,
+      "loss": 0.9652,
+      "step": 19467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.606753975435e-06,
+      "loss": 0.9586,
+      "step": 19468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.606010112812164e-06,
+      "loss": 0.9582,
+      "step": 19469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.605266407384034e-06,
+      "loss": 0.9229,
+      "step": 19470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.604522859164538e-06,
+      "loss": 0.9146,
+      "step": 19471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6037794681675996e-06,
+      "loss": 0.869,
+      "step": 19472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.603036234407146e-06,
+      "loss": 0.9777,
+      "step": 19473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.602293157897085e-06,
+      "loss": 0.9621,
+      "step": 19474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6015502386513426e-06,
+      "loss": 0.9473,
+      "step": 19475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6008074766838257e-06,
+      "loss": 0.8662,
+      "step": 19476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6000648720084455e-06,
+      "loss": 0.9323,
+      "step": 19477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5993224246391093e-06,
+      "loss": 0.9402,
+      "step": 19478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59858013458972e-06,
+      "loss": 0.9326,
+      "step": 19479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.597838001874181e-06,
+      "loss": 0.9994,
+      "step": 19480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5970960265063872e-06,
+      "loss": 0.9473,
+      "step": 19481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5963542085002347e-06,
+      "loss": 0.9385,
+      "step": 19482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5956125478696147e-06,
+      "loss": 0.963,
+      "step": 19483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5948710446284176e-06,
+      "loss": 0.922,
+      "step": 19484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59412969879053e-06,
+      "loss": 0.9405,
+      "step": 19485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.593388510369835e-06,
+      "loss": 0.9703,
+      "step": 19486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.592647479380208e-06,
+      "loss": 0.9955,
+      "step": 19487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5919066058355357e-06,
+      "loss": 0.8892,
+      "step": 19488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.591165889749685e-06,
+      "loss": 0.9921,
+      "step": 19489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.59042533113653e-06,
+      "loss": 0.9848,
+      "step": 19490
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5896849300099402e-06,
+      "loss": 1.0363,
+      "step": 19491
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5889446863837764e-06,
+      "loss": 0.8974,
+      "step": 19492
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5882046002719086e-06,
+      "loss": 1.0031,
+      "step": 19493
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.587464671688187e-06,
+      "loss": 0.9196,
+      "step": 19494
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5867249006464793e-06,
+      "loss": 0.9524,
+      "step": 19495
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5859852871606306e-06,
+      "loss": 0.9931,
+      "step": 19496
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.585245831244495e-06,
+      "loss": 0.9055,
+      "step": 19497
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5845065329119205e-06,
+      "loss": 0.9219,
+      "step": 19498
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.58376739217675e-06,
+      "loss": 0.9917,
+      "step": 19499
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.58302840905283e-06,
+      "loss": 1.0078,
+      "step": 19500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.582289583553994e-06,
+      "loss": 1.0023,
+      "step": 19501
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5815509156940813e-06,
+      "loss": 0.9413,
+      "step": 19502
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5808124054869233e-06,
+      "loss": 1.0058,
+      "step": 19503
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.580074052946351e-06,
+      "loss": 0.914,
+      "step": 19504
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5793358580861916e-06,
+      "loss": 0.9295,
+      "step": 19505
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.578597820920271e-06,
+      "loss": 0.9558,
+      "step": 19506
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5778599414624075e-06,
+      "loss": 0.9785,
+      "step": 19507
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.577122219726419e-06,
+      "loss": 0.9681,
+      "step": 19508
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.576384655726123e-06,
+      "loss": 0.9757,
+      "step": 19509
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5756472494753315e-06,
+      "loss": 0.9985,
+      "step": 19510
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.574910000987857e-06,
+      "loss": 0.9419,
+      "step": 19511
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.574172910277497e-06,
+      "loss": 0.9597,
+      "step": 19512
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5734359773580653e-06,
+      "loss": 0.9299,
+      "step": 19513
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.572699202243356e-06,
+      "loss": 0.9161,
+      "step": 19514
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.571962584947169e-06,
+      "loss": 0.9219,
+      "step": 19515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.571226125483298e-06,
+      "loss": 0.8631,
+      "step": 19516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5704898238655364e-06,
+      "loss": 0.9652,
+      "step": 19517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5697536801076738e-06,
+      "loss": 0.9339,
+      "step": 19518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5690176942234914e-06,
+      "loss": 0.9397,
+      "step": 19519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.568281866226775e-06,
+      "loss": 0.9044,
+      "step": 19520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5675461961313054e-06,
+      "loss": 0.9121,
+      "step": 19521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5668106839508565e-06,
+      "loss": 0.9624,
+      "step": 19522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5660753296992056e-06,
+      "loss": 0.9444,
+      "step": 19523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.565340133390124e-06,
+      "loss": 0.9292,
+      "step": 19524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.5646050950373737e-06,
+      "loss": 0.9615,
+      "step": 19525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.563870214654729e-06,
+      "loss": 1.0026,
+      "step": 19526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.563135492255945e-06,
+      "loss": 0.9731,
+      "step": 19527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.562400927854782e-06,
+      "loss": 0.8574,
+      "step": 19528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5616665214650017e-06,
+      "loss": 0.9181,
+      "step": 19529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5609322731003473e-06,
+      "loss": 0.8878,
+      "step": 19530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5601981827745792e-06,
+      "loss": 0.9338,
+      "step": 19531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5594642505014356e-06,
+      "loss": 1.026,
+      "step": 19532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.558730476294671e-06,
+      "loss": 0.9347,
+      "step": 19533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5579968601680197e-06,
+      "loss": 0.9728,
+      "step": 19534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5572634021352206e-06,
+      "loss": 0.9519,
+      "step": 19535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5565301022100107e-06,
+      "loss": 0.9043,
+      "step": 19536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5557969604061225e-06,
+      "loss": 0.8479,
+      "step": 19537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5550639767372886e-06,
+      "loss": 1.039,
+      "step": 19538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.55433115121723e-06,
+      "loss": 0.9684,
+      "step": 19539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.553598483859673e-06,
+      "loss": 0.9925,
+      "step": 19540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.552865974678338e-06,
+      "loss": 0.8683,
+      "step": 19541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5521336236869432e-06,
+      "loss": 0.9913,
+      "step": 19542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.551401430899202e-06,
+      "loss": 0.9381,
+      "step": 19543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.550669396328831e-06,
+      "loss": 0.9634,
+      "step": 19544
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5499375199895326e-06,
+      "loss": 0.9908,
+      "step": 19545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.549205801895015e-06,
+      "loss": 0.9838,
+      "step": 19546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5484742420589827e-06,
+      "loss": 0.9639,
+      "step": 19547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5477428404951334e-06,
+      "loss": 0.9814,
+      "step": 19548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5470115972171684e-06,
+      "loss": 1.006,
+      "step": 19549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5462805122387734e-06,
+      "loss": 0.9683,
+      "step": 19550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5455495855736488e-06,
+      "loss": 0.9715,
+      "step": 19551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5448188172354746e-06,
+      "loss": 0.914,
+      "step": 19552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544088207237945e-06,
+      "loss": 0.9937,
+      "step": 19553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.543357755594734e-06,
+      "loss": 0.945,
+      "step": 19554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5426274623195246e-06,
+      "loss": 0.9403,
+      "step": 19555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.541897327425994e-06,
+      "loss": 0.9515,
+      "step": 19556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.541167350927809e-06,
+      "loss": 0.9921,
+      "step": 19557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.54043753283865e-06,
+      "loss": 0.9249,
+      "step": 19558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.539707873172177e-06,
+      "loss": 0.9838,
+      "step": 19559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5389783719420556e-06,
+      "loss": 0.9667,
+      "step": 19560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5382490291619486e-06,
+      "loss": 0.9609,
+      "step": 19561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.537519844845513e-06,
+      "loss": 0.9314,
+      "step": 19562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.536790819006406e-06,
+      "loss": 0.9398,
+      "step": 19563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.536061951658282e-06,
+      "loss": 0.9778,
+      "step": 19564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5353332428147847e-06,
+      "loss": 0.8979,
+      "step": 19565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.534604692489564e-06,
+      "loss": 1.0043,
+      "step": 19566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.533876300696263e-06,
+      "loss": 0.9707,
+      "step": 19567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5331480674485235e-06,
+      "loss": 0.9258,
+      "step": 19568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5324199927599833e-06,
+      "loss": 1.0106,
+      "step": 19569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.531692076644272e-06,
+      "loss": 0.9538,
+      "step": 19570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.53096431911503e-06,
+      "loss": 0.96,
+      "step": 19571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.530236720185879e-06,
+      "loss": 0.8727,
+      "step": 19572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.529509279870448e-06,
+      "loss": 0.9483,
+      "step": 19573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.528781998182358e-06,
+      "loss": 0.9256,
+      "step": 19574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5280548751352305e-06,
+      "loss": 0.9373,
+      "step": 19575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5273279107426843e-06,
+      "loss": 0.9503,
+      "step": 19576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.526601105018325e-06,
+      "loss": 0.9658,
+      "step": 19577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5258744579757743e-06,
+      "loss": 0.8639,
+      "step": 19578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.525147969628634e-06,
+      "loss": 0.9695,
+      "step": 19579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5244216399905084e-06,
+      "loss": 0.847,
+      "step": 19580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5236954690750024e-06,
+      "loss": 0.959,
+      "step": 19581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5229694568957176e-06,
+      "loss": 0.9409,
+      "step": 19582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5222436034662423e-06,
+      "loss": 0.9679,
+      "step": 19583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.521517908800174e-06,
+      "loss": 0.9391,
+      "step": 19584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5207923729111026e-06,
+      "loss": 0.9385,
+      "step": 19585
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5200669958126157e-06,
+      "loss": 0.9626,
+      "step": 19586
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.519341777518299e-06,
+      "loss": 1.0234,
+      "step": 19587
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5186167180417278e-06,
+      "loss": 0.9495,
+      "step": 19588
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5178918173964886e-06,
+      "loss": 0.9551,
+      "step": 19589
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.517167075596148e-06,
+      "loss": 0.9078,
+      "step": 19590
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.516442492654285e-06,
+      "loss": 0.9331,
+      "step": 19591
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5157180685844663e-06,
+      "loss": 0.9204,
+      "step": 19592
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5149938034002585e-06,
+      "loss": 0.9273,
+      "step": 19593
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.514269697115226e-06,
+      "loss": 0.983,
+      "step": 19594
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.513545749742924e-06,
+      "loss": 1.0094,
+      "step": 19595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5128219612969185e-06,
+      "loss": 0.9514,
+      "step": 19596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5120983317907576e-06,
+      "loss": 0.9936,
+      "step": 19597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5113748612379944e-06,
+      "loss": 0.8677,
+      "step": 19598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.510651549652177e-06,
+      "loss": 0.965,
+      "step": 19599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5099283970468503e-06,
+      "loss": 0.8554,
+      "step": 19600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5092054034355617e-06,
+      "loss": 0.9509,
+      "step": 19601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5084825688318446e-06,
+      "loss": 1.0351,
+      "step": 19602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5077598932492375e-06,
+      "loss": 0.8945,
+      "step": 19603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5070373767012744e-06,
+      "loss": 0.9766,
+      "step": 19604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5063150192014864e-06,
+      "loss": 0.9513,
+      "step": 19605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5055928207633996e-06,
+      "loss": 0.9829,
+      "step": 19606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5048707814005436e-06,
+      "loss": 0.9584,
+      "step": 19607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5041489011264297e-06,
+      "loss": 0.9295,
+      "step": 19608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.503427179954588e-06,
+      "loss": 0.9447,
+      "step": 19609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.502705617898529e-06,
+      "loss": 0.986,
+      "step": 19610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.501984214971763e-06,
+      "loss": 0.9586,
+      "step": 19611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.501262971187807e-06,
+      "loss": 0.9308,
+      "step": 19612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.500541886560157e-06,
+      "loss": 0.9223,
+      "step": 19613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4998209611023284e-06,
+      "loss": 0.9297,
+      "step": 19614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.499100194827812e-06,
+      "loss": 0.9055,
+      "step": 19615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4983795877501137e-06,
+      "loss": 0.9786,
+      "step": 19616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.497659139882723e-06,
+      "loss": 0.9628,
+      "step": 19617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4969388512391347e-06,
+      "loss": 0.9321,
+      "step": 19618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.496218721832836e-06,
+      "loss": 0.9339,
+      "step": 19619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.495498751677313e-06,
+      "loss": 0.9863,
+      "step": 19620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4947789407860523e-06,
+      "loss": 0.8914,
+      "step": 19621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.494059289172528e-06,
+      "loss": 0.9757,
+      "step": 19622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4933397968502195e-06,
+      "loss": 1.0048,
+      "step": 19623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4926204638326012e-06,
+      "loss": 0.9363,
+      "step": 19624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491901290133143e-06,
+      "loss": 0.962,
+      "step": 19625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.491182275765315e-06,
+      "loss": 0.9249,
+      "step": 19626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4904634207425817e-06,
+      "loss": 0.9781,
+      "step": 19627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4897447250784032e-06,
+      "loss": 0.9321,
+      "step": 19628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4890261887862402e-06,
+      "loss": 0.9722,
+      "step": 19629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.488307811879547e-06,
+      "loss": 0.9584,
+      "step": 19630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4875895943717785e-06,
+      "loss": 0.9274,
+      "step": 19631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4868715362763865e-06,
+      "loss": 0.9798,
+      "step": 19632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.486153637606811e-06,
+      "loss": 1.0009,
+      "step": 19633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4854358983765063e-06,
+      "loss": 0.9913,
+      "step": 19634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4847183185989036e-06,
+      "loss": 0.9369,
+      "step": 19635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4840008982874496e-06,
+      "loss": 0.9416,
+      "step": 19636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4832836374555733e-06,
+      "loss": 0.9872,
+      "step": 19637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4825665361167106e-06,
+      "loss": 0.9832,
+      "step": 19638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.48184959428429e-06,
+      "loss": 0.972,
+      "step": 19639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4811328119717324e-06,
+      "loss": 0.9603,
+      "step": 19640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4804161891924716e-06,
+      "loss": 0.9936,
+      "step": 19641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.479699725959919e-06,
+      "loss": 1.029,
+      "step": 19642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4789834222874943e-06,
+      "loss": 0.9175,
+      "step": 19643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4782672781886117e-06,
+      "loss": 0.9881,
+      "step": 19644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4775512936766834e-06,
+      "loss": 0.9286,
+      "step": 19645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4768354687651175e-06,
+      "loss": 0.9159,
+      "step": 19646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4761198034673204e-06,
+      "loss": 0.9997,
+      "step": 19647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4754042977966917e-06,
+      "loss": 0.9668,
+      "step": 19648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.474688951766632e-06,
+      "loss": 0.9457,
+      "step": 19649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4739737653905395e-06,
+      "loss": 0.9445,
+      "step": 19650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.473258738681801e-06,
+      "loss": 0.9255,
+      "step": 19651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4725438716538165e-06,
+      "loss": 0.8833,
+      "step": 19652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.471829164319963e-06,
+      "loss": 0.9792,
+      "step": 19653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.471114616693635e-06,
+      "loss": 0.9145,
+      "step": 19654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4704002287882068e-06,
+      "loss": 0.918,
+      "step": 19655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4696860006170597e-06,
+      "loss": 0.9041,
+      "step": 19656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4689719321935678e-06,
+      "loss": 0.9949,
+      "step": 19657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.468258023531104e-06,
+      "loss": 0.9282,
+      "step": 19658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.467544274643041e-06,
+      "loss": 0.9579,
+      "step": 19659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4668306855427396e-06,
+      "loss": 1.0073,
+      "step": 19660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4661172562435657e-06,
+      "loss": 1.0162,
+      "step": 19661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4654039867588786e-06,
+      "loss": 0.9266,
+      "step": 19662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4646908771020386e-06,
+      "loss": 0.9091,
+      "step": 19663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4639779272863974e-06,
+      "loss": 0.9131,
+      "step": 19664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4632651373253094e-06,
+      "loss": 0.9366,
+      "step": 19665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4625525072321187e-06,
+      "loss": 0.9196,
+      "step": 19666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4618400370201745e-06,
+      "loss": 0.9981,
+      "step": 19667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.461127726702817e-06,
+      "loss": 0.9667,
+      "step": 19668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.460415576293387e-06,
+      "loss": 0.9384,
+      "step": 19669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4597035858052233e-06,
+      "loss": 0.9389,
+      "step": 19670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.458991755251651e-06,
+      "loss": 0.9456,
+      "step": 19671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.458280084646011e-06,
+      "loss": 0.9741,
+      "step": 19672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.457568574001622e-06,
+      "loss": 0.9676,
+      "step": 19673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4568572233318178e-06,
+      "loss": 0.9593,
+      "step": 19674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.456146032649911e-06,
+      "loss": 0.9271,
+      "step": 19675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4554350019692244e-06,
+      "loss": 0.9775,
+      "step": 19676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4547241313030757e-06,
+      "loss": 0.9735,
+      "step": 19677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4540134206647695e-06,
+      "loss": 0.9488,
+      "step": 19678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4533028700676242e-06,
+      "loss": 0.9429,
+      "step": 19679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.452592479524941e-06,
+      "loss": 0.9316,
+      "step": 19680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.451882249050025e-06,
+      "loss": 0.9827,
+      "step": 19681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.451172178656177e-06,
+      "loss": 0.9608,
+      "step": 19682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.450462268356694e-06,
+      "loss": 0.9632,
+      "step": 19683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449752518164871e-06,
+      "loss": 0.8875,
+      "step": 19684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.449042928094001e-06,
+      "loss": 0.9366,
+      "step": 19685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4483334981573692e-06,
+      "loss": 0.9629,
+      "step": 19686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4476242283682616e-06,
+      "loss": 0.9892,
+      "step": 19687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.446915118739963e-06,
+      "loss": 0.9189,
+      "step": 19688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4462061692857508e-06,
+      "loss": 0.9828,
+      "step": 19689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4454973800189055e-06,
+      "loss": 0.9253,
+      "step": 19690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4447887509526936e-06,
+      "loss": 0.9699,
+      "step": 19691
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.444080282100393e-06,
+      "loss": 0.9767,
+      "step": 19692
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.443371973475267e-06,
+      "loss": 0.9455,
+      "step": 19693
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4426638250905796e-06,
+      "loss": 0.9661,
+      "step": 19694
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4419558369595953e-06,
+      "loss": 0.9348,
+      "step": 19695
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4412480090955705e-06,
+      "loss": 1.0369,
+      "step": 19696
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.440540341511764e-06,
+      "loss": 0.9143,
+      "step": 19697
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4398328342214207e-06,
+      "loss": 0.9412,
+      "step": 19698
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.439125487237799e-06,
+      "loss": 0.9617,
+      "step": 19699
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43841830057414e-06,
+      "loss": 0.9484,
+      "step": 19700
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.437711274243689e-06,
+      "loss": 0.9396,
+      "step": 19701
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.437004408259687e-06,
+      "loss": 0.9476,
+      "step": 19702
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4362977026353695e-06,
+      "loss": 0.9809,
+      "step": 19703
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.435591157383975e-06,
+      "loss": 1.0049,
+      "step": 19704
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.43488477251873e-06,
+      "loss": 0.9395,
+      "step": 19705
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4341785480528656e-06,
+      "loss": 1.066,
+      "step": 19706
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.433472483999607e-06,
+      "loss": 0.9151,
+      "step": 19707
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.432766580372177e-06,
+      "loss": 0.9504,
+      "step": 19708
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.432060837183794e-06,
+      "loss": 0.961,
+      "step": 19709
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4313552544476785e-06,
+      "loss": 0.9937,
+      "step": 19710
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4306498321770356e-06,
+      "loss": 0.9145,
+      "step": 19711
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4299445703850867e-06,
+      "loss": 0.9886,
+      "step": 19712
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4292394690850307e-06,
+      "loss": 0.9415,
+      "step": 19713
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4285345282900742e-06,
+      "loss": 0.9591,
+      "step": 19714
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.427829748013423e-06,
+      "loss": 0.9517,
+      "step": 19715
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.427125128268266e-06,
+      "loss": 0.9529,
+      "step": 19716
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4264206690678103e-06,
+      "loss": 0.9274,
+      "step": 19717
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.425716370425241e-06,
+      "loss": 0.8672,
+      "step": 19718
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4250122323537474e-06,
+      "loss": 1.0088,
+      "step": 19719
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.424308254866519e-06,
+      "loss": 0.8846,
+      "step": 19720
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.423604437976738e-06,
+      "loss": 0.9873,
+      "step": 19721
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4229007816975834e-06,
+      "loss": 0.9412,
+      "step": 19722
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4221972860422383e-06,
+      "loss": 0.9755,
+      "step": 19723
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.421493951023869e-06,
+      "loss": 0.9218,
+      "step": 19724
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4207907766556517e-06,
+      "loss": 0.9384,
+      "step": 19725
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4200877629507538e-06,
+      "loss": 0.8983,
+      "step": 19726
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.41938490992234e-06,
+      "loss": 0.9004,
+      "step": 19727
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4186822175835754e-06,
+      "loss": 1.0069,
+      "step": 19728
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4179796859476135e-06,
+      "loss": 0.9386,
+      "step": 19729
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4172773150276186e-06,
+      "loss": 0.9773,
+      "step": 19730
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.416575104836737e-06,
+      "loss": 0.9762,
+      "step": 19731
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4158730553881227e-06,
+      "loss": 0.9686,
+      "step": 19732
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4151711666949217e-06,
+      "loss": 0.943,
+      "step": 19733
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4144694387702795e-06,
+      "loss": 0.9624,
+      "step": 19734
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.413767871627339e-06,
+      "loss": 0.8661,
+      "step": 19735
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4130664652792314e-06,
+      "loss": 0.9197,
+      "step": 19736
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4123652197391013e-06,
+      "loss": 0.9569,
+      "step": 19737
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4116641350200746e-06,
+      "loss": 0.9705,
+      "step": 19738
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.410963211135281e-06,
+      "loss": 0.9703,
+      "step": 19739
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4102624480978496e-06,
+      "loss": 0.9212,
+      "step": 19740
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4095618459209013e-06,
+      "loss": 0.9149,
+      "step": 19741
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4088614046175606e-06,
+      "loss": 0.9242,
+      "step": 19742
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4081611242009385e-06,
+      "loss": 0.9733,
+      "step": 19743
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4074610046841506e-06,
+      "loss": 0.9646,
+      "step": 19744
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4067610460803115e-06,
+      "loss": 0.9613,
+      "step": 19745
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4060612484025256e-06,
+      "loss": 0.9735,
+      "step": 19746
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4053616116639013e-06,
+      "loss": 0.9215,
+      "step": 19747
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4046621358775403e-06,
+      "loss": 0.9436,
+      "step": 19748
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.403962821056537e-06,
+      "loss": 0.9729,
+      "step": 19749
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4032636672139954e-06,
+      "loss": 0.9855,
+      "step": 19750
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4025646743630017e-06,
+      "loss": 0.9198,
+      "step": 19751
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4018658425166486e-06,
+      "loss": 1.0132,
+      "step": 19752
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4011671716880256e-06,
+      "loss": 0.9631,
+      "step": 19753
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4004686618902096e-06,
+      "loss": 0.9535,
+      "step": 19754
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3997703131362912e-06,
+      "loss": 0.9084,
+      "step": 19755
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.399072125439339e-06,
+      "loss": 1.0582,
+      "step": 19756
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.398374098812437e-06,
+      "loss": 1.0206,
+      "step": 19757
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.39767623326865e-06,
+      "loss": 0.9747,
+      "step": 19758
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.39697852882105e-06,
+      "loss": 0.9296,
+      "step": 19759
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.396280985482703e-06,
+      "loss": 0.943,
+      "step": 19760
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3955836032666715e-06,
+      "loss": 0.9705,
+      "step": 19761
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.394886382186018e-06,
+      "loss": 0.9135,
+      "step": 19762
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3941893222537936e-06,
+      "loss": 0.9922,
+      "step": 19763
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.393492423483056e-06,
+      "loss": 0.9713,
+      "step": 19764
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3927956858868565e-06,
+      "loss": 0.8902,
+      "step": 19765
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.392099109478241e-06,
+      "loss": 0.9411,
+      "step": 19766
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3914026942702553e-06,
+      "loss": 1.0021,
+      "step": 19767
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.390706440275944e-06,
+      "loss": 0.8851,
+      "step": 19768
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.39001034750834e-06,
+      "loss": 0.9453,
+      "step": 19769
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3893144159804828e-06,
+      "loss": 0.9019,
+      "step": 19770
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3886186457054038e-06,
+      "loss": 0.9942,
+      "step": 19771
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3879230366961335e-06,
+      "loss": 0.9772,
+      "step": 19772
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3872275889657015e-06,
+      "loss": 0.9516,
+      "step": 19773
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.386532302527125e-06,
+      "loss": 0.9099,
+      "step": 19774
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.385837177393431e-06,
+      "loss": 0.9283,
+      "step": 19775
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3851422135776304e-06,
+      "loss": 0.9619,
+      "step": 19776
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.384447411092746e-06,
+      "loss": 0.9463,
+      "step": 19777
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3837527699517827e-06,
+      "loss": 0.926,
+      "step": 19778
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3830582901677523e-06,
+      "loss": 0.9807,
+      "step": 19779
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.382363971753662e-06,
+      "loss": 0.9057,
+      "step": 19780
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3816698147225074e-06,
+      "loss": 0.9233,
+      "step": 19781
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3809758190872956e-06,
+      "loss": 0.9429,
+      "step": 19782
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3802819848610195e-06,
+      "loss": 0.9307,
+      "step": 19783
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.379588312056672e-06,
+      "loss": 0.9288,
+      "step": 19784
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.378894800687246e-06,
+      "loss": 0.9557,
+      "step": 19785
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3782014507657293e-06,
+      "loss": 0.9354,
+      "step": 19786
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3775082623051017e-06,
+      "loss": 0.8626,
+      "step": 19787
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3768152353183483e-06,
+      "loss": 0.9176,
+      "step": 19788
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3761223698184463e-06,
+      "loss": 0.9489,
+      "step": 19789
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3754296658183717e-06,
+      "loss": 0.9801,
+      "step": 19790
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3747371233310977e-06,
+      "loss": 1.0047,
+      "step": 19791
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.374044742369589e-06,
+      "loss": 0.9558,
+      "step": 19792
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.373352522946819e-06,
+      "loss": 0.9132,
+      "step": 19793
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3726604650757433e-06,
+      "loss": 0.9668,
+      "step": 19794
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3719685687693308e-06,
+      "loss": 0.9282,
+      "step": 19795
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3712768340405304e-06,
+      "loss": 0.907,
+      "step": 19796
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.370585260902301e-06,
+      "loss": 0.9068,
+      "step": 19797
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3698938493675925e-06,
+      "loss": 0.9658,
+      "step": 19798
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3692025994493506e-06,
+      "loss": 0.9779,
+      "step": 19799
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3685115111605263e-06,
+      "loss": 0.904,
+      "step": 19800
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3678205845140557e-06,
+      "loss": 0.9814,
+      "step": 19801
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3671298195228813e-06,
+      "loss": 0.9779,
+      "step": 19802
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3664392161999362e-06,
+      "loss": 0.9792,
+      "step": 19803
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3657487745581565e-06,
+      "loss": 0.9748,
+      "step": 19804
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3650584946104704e-06,
+      "loss": 0.9328,
+      "step": 19805
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.364368376369807e-06,
+      "loss": 0.9273,
+      "step": 19806
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.363678419849086e-06,
+      "loss": 0.8574,
+      "step": 19807
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.362988625061231e-06,
+      "loss": 0.9053,
+      "step": 19808
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3622989920191596e-06,
+      "loss": 0.922,
+      "step": 19809
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.361609520735786e-06,
+      "loss": 0.9117,
+      "step": 19810
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.360920211224025e-06,
+      "loss": 1.0508,
+      "step": 19811
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.360231063496778e-06,
+      "loss": 1.0232,
+      "step": 19812
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.359542077566961e-06,
+      "loss": 0.9894,
+      "step": 19813
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.358853253447467e-06,
+      "loss": 0.8986,
+      "step": 19814
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3581645911512008e-06,
+      "loss": 0.9473,
+      "step": 19815
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3574760906910585e-06,
+      "loss": 0.9646,
+      "step": 19816
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.356787752079932e-06,
+      "loss": 0.9708,
+      "step": 19817
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.356099575330716e-06,
+      "loss": 0.9702,
+      "step": 19818
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3554115604562912e-06,
+      "loss": 0.9653,
+      "step": 19819
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3547237074695493e-06,
+      "loss": 0.9228,
+      "step": 19820
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3540360163833665e-06,
+      "loss": 1.0003,
+      "step": 19821
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3533484872106227e-06,
+      "loss": 0.8871,
+      "step": 19822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3526611199641947e-06,
+      "loss": 0.9156,
+      "step": 19823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.351973914656953e-06,
+      "loss": 0.9045,
+      "step": 19824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.35128687130177e-06,
+      "loss": 0.9765,
+      "step": 19825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3505999899115087e-06,
+      "loss": 0.98,
+      "step": 19826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3499132704990324e-06,
+      "loss": 0.9233,
+      "step": 19827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3492267130772027e-06,
+      "loss": 0.9204,
+      "step": 19828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3485403176588763e-06,
+      "loss": 0.9062,
+      "step": 19829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3478540842569077e-06,
+      "loss": 0.9839,
+      "step": 19830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3471680128841503e-06,
+      "loss": 0.946,
+      "step": 19831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.346482103553446e-06,
+      "loss": 0.9307,
+      "step": 19832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3457963562776465e-06,
+      "loss": 0.9037,
+      "step": 19833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.34511077106959e-06,
+      "loss": 0.9973,
+      "step": 19834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3444253479421165e-06,
+      "loss": 0.961,
+      "step": 19835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.343740086908063e-06,
+      "loss": 0.9229,
+      "step": 19836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3430549879802578e-06,
+      "loss": 0.9332,
+      "step": 19837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.342370051171539e-06,
+      "loss": 0.9646,
+      "step": 19838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.341685276494724e-06,
+      "loss": 0.9281,
+      "step": 19839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.341000663962646e-06,
+      "loss": 0.9249,
+      "step": 19840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3403162135881185e-06,
+      "loss": 0.9019,
+      "step": 19841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.339631925383963e-06,
+      "loss": 1.0253,
+      "step": 19842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3389477993629928e-06,
+      "loss": 0.9664,
+      "step": 19843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3382638355380193e-06,
+      "loss": 0.9687,
+      "step": 19844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.337580033921854e-06,
+      "loss": 0.9044,
+      "step": 19845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3368963945272984e-06,
+      "loss": 1.0294,
+      "step": 19846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3362129173671567e-06,
+      "loss": 1.0121,
+      "step": 19847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3355296024542283e-06,
+      "loss": 0.947,
+      "step": 19848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.33484644980131e-06,
+      "loss": 0.985,
+      "step": 19849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3341634594211946e-06,
+      "loss": 0.9585,
+      "step": 19850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.333480631326677e-06,
+      "loss": 0.9304,
+      "step": 19851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3327979655305363e-06,
+      "loss": 0.9852,
+      "step": 19852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3321154620455634e-06,
+      "loss": 1.021,
+      "step": 19853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3314331208845365e-06,
+      "loss": 0.9499,
+      "step": 19854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.330750942060234e-06,
+      "loss": 0.8998,
+      "step": 19855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.330068925585435e-06,
+      "loss": 0.9346,
+      "step": 19856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3293870714729028e-06,
+      "loss": 1.017,
+      "step": 19857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.328705379735418e-06,
+      "loss": 0.9757,
+      "step": 19858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3280238503857367e-06,
+      "loss": 0.9288,
+      "step": 19859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3273424834366276e-06,
+      "loss": 0.9916,
+      "step": 19860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3266612789008494e-06,
+      "loss": 0.9263,
+      "step": 19861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3259802367911575e-06,
+      "loss": 0.9826,
+      "step": 19862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.325299357120311e-06,
+      "loss": 0.9444,
+      "step": 19863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.324618639901053e-06,
+      "loss": 1.0027,
+      "step": 19864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3239380851461357e-06,
+      "loss": 1.0088,
+      "step": 19865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3232576928683048e-06,
+      "loss": 0.9908,
+      "step": 19866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.322577463080299e-06,
+      "loss": 0.9819,
+      "step": 19867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.32189739579486e-06,
+      "loss": 1.0099,
+      "step": 19868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3212174910247233e-06,
+      "loss": 0.9377,
+      "step": 19869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3205377487826165e-06,
+      "loss": 0.9351,
+      "step": 19870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3198581690812774e-06,
+      "loss": 1.0082,
+      "step": 19871
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3191787519334252e-06,
+      "loss": 0.9328,
+      "step": 19872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.318499497351787e-06,
+      "loss": 1.0167,
+      "step": 19873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.317820405349085e-06,
+      "loss": 0.9711,
+      "step": 19874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3171414759380286e-06,
+      "loss": 1.0291,
+      "step": 19875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.316462709131343e-06,
+      "loss": 0.9158,
+      "step": 19876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3157841049417296e-06,
+      "loss": 0.9311,
+      "step": 19877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3151056633819057e-06,
+      "loss": 0.9631,
+      "step": 19878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3144273844645707e-06,
+      "loss": 0.9382,
+      "step": 19879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3137492682024266e-06,
+      "loss": 0.941,
+      "step": 19880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.313071314608175e-06,
+      "loss": 0.9634,
+      "step": 19881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3123935236945104e-06,
+      "loss": 0.961,
+      "step": 19882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.311715895474129e-06,
+      "loss": 0.9813,
+      "step": 19883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.311038429959717e-06,
+      "loss": 0.9953,
+      "step": 19884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3103611271639617e-06,
+      "loss": 0.9666,
+      "step": 19885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.309683987099548e-06,
+      "loss": 1.0134,
+      "step": 19886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3090070097791563e-06,
+      "loss": 0.9562,
+      "step": 19887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3083301952154645e-06,
+      "loss": 0.9265,
+      "step": 19888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.30765354342115e-06,
+      "loss": 0.913,
+      "step": 19889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.30697705440888e-06,
+      "loss": 0.9375,
+      "step": 19890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3063007281913254e-06,
+      "loss": 0.9706,
+      "step": 19891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3056245647811515e-06,
+      "loss": 0.9521,
+      "step": 19892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.304948564191021e-06,
+      "loss": 0.9487,
+      "step": 19893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.304272726433594e-06,
+      "loss": 0.9515,
+      "step": 19894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3035970515215237e-06,
+      "loss": 0.9283,
+      "step": 19895
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.30292153946747e-06,
+      "loss": 1.0154,
+      "step": 19896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.302246190284074e-06,
+      "loss": 0.9879,
+      "step": 19897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3015710039839946e-06,
+      "loss": 0.9755,
+      "step": 19898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.300895980579866e-06,
+      "loss": 0.9774,
+      "step": 19899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3002211200843337e-06,
+      "loss": 0.9321,
+      "step": 19900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.299546422510037e-06,
+      "loss": 0.9344,
+      "step": 19901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.298871887869606e-06,
+      "loss": 0.9083,
+      "step": 19902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2981975161756796e-06,
+      "loss": 0.9007,
+      "step": 19903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.297523307440881e-06,
+      "loss": 0.9525,
+      "step": 19904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.296849261677839e-06,
+      "loss": 0.974,
+      "step": 19905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2961753788991756e-06,
+      "loss": 0.883,
+      "step": 19906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2955016591175117e-06,
+      "loss": 0.9248,
+      "step": 19907
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2948281023454622e-06,
+      "loss": 0.9197,
+      "step": 19908
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.294154708595645e-06,
+      "loss": 1.003,
+      "step": 19909
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2934814778806647e-06,
+      "loss": 0.8908,
+      "step": 19910
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.292808410213132e-06,
+      "loss": 0.9194,
+      "step": 19911
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.292135505605653e-06,
+      "loss": 0.8996,
+      "step": 19912
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2914627640708265e-06,
+      "loss": 0.8612,
+      "step": 19913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.290790185621253e-06,
+      "loss": 0.9507,
+      "step": 19914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2901177702695254e-06,
+      "loss": 0.9478,
+      "step": 19915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2894455180282405e-06,
+      "loss": 0.898,
+      "step": 19916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.288773428909983e-06,
+      "loss": 0.8889,
+      "step": 19917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.28810150292734e-06,
+      "loss": 0.9685,
+      "step": 19918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.287429740092897e-06,
+      "loss": 0.9894,
+      "step": 19919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.286758140419232e-06,
+      "loss": 0.879,
+      "step": 19920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.286086703918926e-06,
+      "loss": 0.9301,
+      "step": 19921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2854154306045437e-06,
+      "loss": 0.9143,
+      "step": 19922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.284744320488669e-06,
+      "loss": 0.9405,
+      "step": 19923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2840733735838593e-06,
+      "loss": 0.9342,
+      "step": 19924
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2834025899026836e-06,
+      "loss": 1.0076,
+      "step": 19925
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2827319694577033e-06,
+      "loss": 0.9544,
+      "step": 19926
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2820615122614795e-06,
+      "loss": 0.9487,
+      "step": 19927
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.281391218326564e-06,
+      "loss": 0.9746,
+      "step": 19928
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2807210876655105e-06,
+      "loss": 0.9095,
+      "step": 19929
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2800511202908693e-06,
+      "loss": 0.9792,
+      "step": 19930
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2793813162151856e-06,
+      "loss": 0.9448,
+      "step": 19931
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2787116754510075e-06,
+      "loss": 0.964,
+      "step": 19932
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.278042198010867e-06,
+      "loss": 0.9136,
+      "step": 19933
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2773728839073105e-06,
+      "loss": 0.9534,
+      "step": 19934
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2767037331528642e-06,
+      "loss": 0.933,
+      "step": 19935
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2760347457600674e-06,
+      "loss": 0.9784,
+      "step": 19936
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2753659217414414e-06,
+      "loss": 0.9567,
+      "step": 19937
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2746972611095153e-06,
+      "loss": 0.9621,
+      "step": 19938
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.274028763876811e-06,
+      "loss": 0.9332,
+      "step": 19939
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.273360430055842e-06,
+      "loss": 0.915,
+      "step": 19940
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.272692259659133e-06,
+      "loss": 0.9718,
+      "step": 19941
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2720242526991898e-06,
+      "loss": 0.9205,
+      "step": 19942
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2713564091885253e-06,
+      "loss": 0.9461,
+      "step": 19943
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.270688729139646e-06,
+      "loss": 0.9235,
+      "step": 19944
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2700212125650543e-06,
+      "loss": 1.0235,
+      "step": 19945
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2693538594772522e-06,
+      "loss": 0.919,
+      "step": 19946
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.26868666988874e-06,
+      "loss": 0.9353,
+      "step": 19947
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.268019643812006e-06,
+      "loss": 0.9541,
+      "step": 19948
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2673527812595453e-06,
+      "loss": 0.8787,
+      "step": 19949
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2666860822438452e-06,
+      "loss": 0.9891,
+      "step": 19950
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2660195467773917e-06,
+      "loss": 0.9525,
+      "step": 19951
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.26535317487267e-06,
+      "loss": 0.9348,
+      "step": 19952
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.26468696654215e-06,
+      "loss": 0.9509,
+      "step": 19953
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.264020921798319e-06,
+      "loss": 0.9525,
+      "step": 19954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.263355040653643e-06,
+      "loss": 0.8966,
+      "step": 19955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.262689323120595e-06,
+      "loss": 0.9402,
+      "step": 19956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2620237692116418e-06,
+      "loss": 0.9498,
+      "step": 19957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2613583789392425e-06,
+      "loss": 0.9229,
+      "step": 19958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.260693152315866e-06,
+      "loss": 0.9619,
+      "step": 19959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2600280893539618e-06,
+      "loss": 0.8739,
+      "step": 19960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2593631900659917e-06,
+      "loss": 0.8628,
+      "step": 19961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.258698454464403e-06,
+      "loss": 0.9754,
+      "step": 19962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2580338825616454e-06,
+      "loss": 1.0218,
+      "step": 19963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2573694743701648e-06,
+      "loss": 0.9929,
+      "step": 19964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.256705229902403e-06,
+      "loss": 0.9733,
+      "step": 19965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2560411491708024e-06,
+      "loss": 0.9537,
+      "step": 19966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.255377232187793e-06,
+      "loss": 0.9995,
+      "step": 19967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.254713478965812e-06,
+      "loss": 1.0081,
+      "step": 19968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.254049889517289e-06,
+      "loss": 0.9477,
+      "step": 19969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.253386463854651e-06,
+      "loss": 0.9219,
+      "step": 19970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.252723201990322e-06,
+      "loss": 0.9648,
+      "step": 19971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2520601039367252e-06,
+      "loss": 0.9106,
+      "step": 19972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2513971697062757e-06,
+      "loss": 1.0539,
+      "step": 19973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2507343993113863e-06,
+      "loss": 0.9211,
+      "step": 19974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2500717927644733e-06,
+      "loss": 0.8964,
+      "step": 19975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2494093500779423e-06,
+      "loss": 0.9456,
+      "step": 19976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2487470712642024e-06,
+      "loss": 1.0208,
+      "step": 19977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.24808495633565e-06,
+      "loss": 0.9728,
+      "step": 19978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.247423005304692e-06,
+      "loss": 0.8586,
+      "step": 19979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2467612181837173e-06,
+      "loss": 0.9383,
+      "step": 19980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2460995949851274e-06,
+      "loss": 0.9655,
+      "step": 19981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.245438135721305e-06,
+      "loss": 0.9707,
+      "step": 19982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.244776840404641e-06,
+      "loss": 0.9263,
+      "step": 19983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.244115709047522e-06,
+      "loss": 0.9047,
+      "step": 19984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.243454741662321e-06,
+      "loss": 0.9504,
+      "step": 19985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.242793938261425e-06,
+      "loss": 0.9237,
+      "step": 19986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2421332988572034e-06,
+      "loss": 0.958,
+      "step": 19987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.241472823462029e-06,
+      "loss": 0.8282,
+      "step": 19988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.240812512088271e-06,
+      "loss": 0.9442,
+      "step": 19989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.240152364748296e-06,
+      "loss": 0.9586,
+      "step": 19990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.239492381454465e-06,
+      "loss": 0.8975,
+      "step": 19991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2388325622191397e-06,
+      "loss": 0.9814,
+      "step": 19992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.238172907054674e-06,
+      "loss": 0.8766,
+      "step": 19993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.237513415973424e-06,
+      "loss": 0.917,
+      "step": 19994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.236854088987739e-06,
+      "loss": 0.9733,
+      "step": 19995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2361949261099614e-06,
+      "loss": 0.9065,
+      "step": 19996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.235535927352444e-06,
+      "loss": 0.9202,
+      "step": 19997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2348770927275196e-06,
+      "loss": 0.8873,
+      "step": 19998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.234218422247535e-06,
+      "loss": 0.948,
+      "step": 19999
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.233559915924818e-06,
+      "loss": 0.9285,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "eval_text_loss": 0.926559329032898,
+      "eval_text_runtime": 1.5549,
+      "eval_text_samples_per_second": 47.591,
+      "eval_text_steps_per_second": 1.286,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "eval_svit_loss": 1.074912428855896,
+      "eval_svit_runtime": 3.8164,
+      "eval_svit_samples_per_second": 178.439,
+      "eval_svit_steps_per_second": 2.882,
+      "step": 20000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.232901573771703e-06,
+      "loss": 0.9108,
+      "step": 20001
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2322433958005183e-06,
+      "loss": 0.9315,
+      "step": 20002
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.231585382023591e-06,
+      "loss": 0.9998,
+      "step": 20003
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.230927532453244e-06,
+      "loss": 0.935,
+      "step": 20004
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2302698471017937e-06,
+      "loss": 0.9589,
+      "step": 20005
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2296123259815597e-06,
+      "loss": 1.0777,
+      "step": 20006
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.228954969104853e-06,
+      "loss": 0.9258,
+      "step": 20007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2282977764839866e-06,
+      "loss": 0.9633,
+      "step": 20008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2276407481312658e-06,
+      "loss": 0.9259,
+      "step": 20009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2269838840589977e-06,
+      "loss": 0.9099,
+      "step": 20010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.226327184279479e-06,
+      "loss": 0.9305,
+      "step": 20011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2256706488050097e-06,
+      "loss": 0.9364,
+      "step": 20012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.225014277647886e-06,
+      "loss": 0.9357,
+      "step": 20013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2243580708203986e-06,
+      "loss": 0.9684,
+      "step": 20014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2237020283348375e-06,
+      "loss": 0.9632,
+      "step": 20015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.223046150203485e-06,
+      "loss": 0.9321,
+      "step": 20016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2223904364386295e-06,
+      "loss": 0.9501,
+      "step": 20017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2217348870525436e-06,
+      "loss": 0.9603,
+      "step": 20018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2210795020575118e-06,
+      "loss": 1.0074,
+      "step": 20019
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2204242814658004e-06,
+      "loss": 0.9673,
+      "step": 20020
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.219769225289684e-06,
+      "loss": 0.9861,
+      "step": 20021
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.21911433354143e-06,
+      "loss": 0.9069,
+      "step": 20022
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2184596062332966e-06,
+      "loss": 0.9514,
+      "step": 20023
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2178050433775545e-06,
+      "loss": 0.9261,
+      "step": 20024
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.217150644986454e-06,
+      "loss": 0.966,
+      "step": 20025
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.216496411072253e-06,
+      "loss": 0.9688,
+      "step": 20026
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.215842341647203e-06,
+      "loss": 0.9365,
+      "step": 20027
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2151884367235535e-06,
+      "loss": 0.9609,
+      "step": 20028
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2145346963135487e-06,
+      "loss": 0.9327,
+      "step": 20029
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2138811204294342e-06,
+      "loss": 0.9859,
+      "step": 20030
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.213227709083445e-06,
+      "loss": 0.923,
+      "step": 20031
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.212574462287822e-06,
+      "loss": 0.9375,
+      "step": 20032
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2119213800547956e-06,
+      "loss": 0.995,
+      "step": 20033
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2112684623965963e-06,
+      "loss": 1.0288,
+      "step": 20034
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2106157093254554e-06,
+      "loss": 0.9842,
+      "step": 20035
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2099631208535896e-06,
+      "loss": 0.9464,
+      "step": 20036
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2093106969932276e-06,
+      "loss": 0.9397,
+      "step": 20037
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2086584377565836e-06,
+      "loss": 1.007,
+      "step": 20038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.208006343155871e-06,
+      "loss": 0.9708,
+      "step": 20039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2073544132033045e-06,
+      "loss": 0.9097,
+      "step": 20040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2067026479110908e-06,
+      "loss": 0.9036,
+      "step": 20041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.20605104729144e-06,
+      "loss": 0.9457,
+      "step": 20042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.205399611356547e-06,
+      "loss": 1.0194,
+      "step": 20043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.204748340118619e-06,
+      "loss": 1.0042,
+      "step": 20044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.204097233589847e-06,
+      "loss": 0.9563,
+      "step": 20045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.203446291782425e-06,
+      "loss": 0.9907,
+      "step": 20046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.202795514708547e-06,
+      "loss": 0.962,
+      "step": 20047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2021449023803956e-06,
+      "loss": 1.0131,
+      "step": 20048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2014944548101592e-06,
+      "loss": 0.9465,
+      "step": 20049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.200844172010014e-06,
+      "loss": 0.8876,
+      "step": 20050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.2001940539921405e-06,
+      "loss": 1.0285,
+      "step": 20051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1995441007687137e-06,
+      "loss": 0.9609,
+      "step": 20052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1988943123519046e-06,
+      "loss": 0.9034,
+      "step": 20053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1982446887538822e-06,
+      "loss": 0.9512,
+      "step": 20054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.197595229986814e-06,
+      "loss": 0.9051,
+      "step": 20055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1969459360628562e-06,
+      "loss": 0.956,
+      "step": 20056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.196296806994177e-06,
+      "loss": 0.9318,
+      "step": 20057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1956478427929254e-06,
+      "loss": 0.9579,
+      "step": 20058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1949990434712564e-06,
+      "loss": 0.9825,
+      "step": 20059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1943504090413236e-06,
+      "loss": 0.9805,
+      "step": 20060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.193701939515268e-06,
+      "loss": 0.9584,
+      "step": 20061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1930536349052403e-06,
+      "loss": 0.9268,
+      "step": 20062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1924054952233756e-06,
+      "loss": 0.8495,
+      "step": 20063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.191757520481814e-06,
+      "loss": 0.9782,
+      "step": 20064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1911097106926905e-06,
+      "loss": 0.9243,
+      "step": 20065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.190462065868135e-06,
+      "loss": 0.9142,
+      "step": 20066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1898145860202794e-06,
+      "loss": 0.9459,
+      "step": 20067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1891672711612466e-06,
+      "loss": 0.8448,
+      "step": 20068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1885201213031583e-06,
+      "loss": 0.9395,
+      "step": 20069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.187873136458135e-06,
+      "loss": 0.9595,
+      "step": 20070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1872263166382926e-06,
+      "loss": 0.9175,
+      "step": 20071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1865796618557433e-06,
+      "loss": 0.8759,
+      "step": 20072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1859331721225995e-06,
+      "loss": 0.9505,
+      "step": 20073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1852868474509625e-06,
+      "loss": 0.9545,
+      "step": 20074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1846406878529434e-06,
+      "loss": 0.9318,
+      "step": 20075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1839946933406386e-06,
+      "loss": 0.9231,
+      "step": 20076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.183348863926146e-06,
+      "loss": 0.9431,
+      "step": 20077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.18270319962156e-06,
+      "loss": 0.89,
+      "step": 20078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.182057700438972e-06,
+      "loss": 0.9199,
+      "step": 20079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1814123663904728e-06,
+      "loss": 0.852,
+      "step": 20080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.180767197488143e-06,
+      "loss": 0.9429,
+      "step": 20081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.18012219374407e-06,
+      "loss": 0.977,
+      "step": 20082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1794773551703272e-06,
+      "loss": 0.9369,
+      "step": 20083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1788326817789953e-06,
+      "loss": 0.9778,
+      "step": 20084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1781881735821442e-06,
+      "loss": 1.0027,
+      "step": 20085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1775438305918452e-06,
+      "loss": 0.8867,
+      "step": 20086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.176899652820166e-06,
+      "loss": 0.9733,
+      "step": 20087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.176255640279167e-06,
+      "loss": 0.9336,
+      "step": 20088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1756117929809096e-06,
+      "loss": 0.9991,
+      "step": 20089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.174968110937452e-06,
+      "loss": 0.9639,
+      "step": 20090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1743245941608485e-06,
+      "loss": 0.901,
+      "step": 20091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1736812426631507e-06,
+      "loss": 0.954,
+      "step": 20092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1730380564564067e-06,
+      "loss": 0.9654,
+      "step": 20093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.172395035552657e-06,
+      "loss": 0.958,
+      "step": 20094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1717521799639516e-06,
+      "loss": 0.9445,
+      "step": 20095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1711094897023233e-06,
+      "loss": 0.9405,
+      "step": 20096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1704669647798094e-06,
+      "loss": 0.9387,
+      "step": 20097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1698246052084437e-06,
+      "loss": 0.9001,
+      "step": 20098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1691824110002515e-06,
+      "loss": 0.9419,
+      "step": 20099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1685403821672657e-06,
+      "loss": 0.9022,
+      "step": 20100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.167898518721503e-06,
+      "loss": 0.937,
+      "step": 20101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.16725682067499e-06,
+      "loss": 0.8734,
+      "step": 20102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1666152880397393e-06,
+      "loss": 0.9277,
+      "step": 20103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1659739208277664e-06,
+      "loss": 0.9839,
+      "step": 20104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.165332719051081e-06,
+      "loss": 0.9631,
+      "step": 20105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1646916827216925e-06,
+      "loss": 0.9298,
+      "step": 20106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.164050811851607e-06,
+      "loss": 0.8871,
+      "step": 20107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1634101064528222e-06,
+      "loss": 0.9466,
+      "step": 20108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1627695665373394e-06,
+      "loss": 0.9347,
+      "step": 20109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.162129192117153e-06,
+      "loss": 1.0094,
+      "step": 20110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.161488983204254e-06,
+      "loss": 0.9145,
+      "step": 20111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1608489398106347e-06,
+      "loss": 0.9436,
+      "step": 20112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1602090619482817e-06,
+      "loss": 0.941,
+      "step": 20113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1595693496291734e-06,
+      "loss": 0.9787,
+      "step": 20114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.158929802865294e-06,
+      "loss": 0.9231,
+      "step": 20115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.158290421668617e-06,
+      "loss": 0.9609,
+      "step": 20116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.157651206051119e-06,
+      "loss": 0.9576,
+      "step": 20117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1570121560247703e-06,
+      "loss": 0.9076,
+      "step": 20118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.156373271601535e-06,
+      "loss": 0.9782,
+      "step": 20119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1557345527933837e-06,
+      "loss": 1.0373,
+      "step": 20120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.155095999612269e-06,
+      "loss": 0.8794,
+      "step": 20121
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1544576120701598e-06,
+      "loss": 0.8953,
+      "step": 20122
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.153819390179003e-06,
+      "loss": 0.9169,
+      "step": 20123
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1531813339507535e-06,
+      "loss": 0.9453,
+      "step": 20124
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1525434433973614e-06,
+      "loss": 1.0457,
+      "step": 20125
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.151905718530768e-06,
+      "loss": 0.9385,
+      "step": 20126
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1512681593629223e-06,
+      "loss": 0.9782,
+      "step": 20127
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.15063076590576e-06,
+      "loss": 0.9284,
+      "step": 20128
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.149993538171218e-06,
+      "loss": 0.9482,
+      "step": 20129
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1493564761712294e-06,
+      "loss": 0.9043,
+      "step": 20130
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.148719579917727e-06,
+      "loss": 0.9636,
+      "step": 20131
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1480828494226325e-06,
+      "loss": 0.9566,
+      "step": 20132
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1474462846978784e-06,
+      "loss": 0.8922,
+      "step": 20133
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1468098857553775e-06,
+      "loss": 0.9895,
+      "step": 20134
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1461736526070518e-06,
+      "loss": 0.9868,
+      "step": 20135
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1455375852648173e-06,
+      "loss": 0.8835,
+      "step": 20136
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.14490168374058e-06,
+      "loss": 0.955,
+      "step": 20137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.144265948046256e-06,
+      "loss": 1.0192,
+      "step": 20138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.143630378193743e-06,
+      "loss": 0.9679,
+      "step": 20139
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1429949741949506e-06,
+      "loss": 0.939,
+      "step": 20140
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1423597360617735e-06,
+      "loss": 0.9131,
+      "step": 20141
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1417246638061087e-06,
+      "loss": 0.9496,
+      "step": 20142
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1410897574398527e-06,
+      "loss": 0.9221,
+      "step": 20143
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.140455016974887e-06,
+      "loss": 1.0043,
+      "step": 20144
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1398204424231084e-06,
+      "loss": 0.9849,
+      "step": 20145
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.139186033796393e-06,
+      "loss": 1.0498,
+      "step": 20146
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1385517911066254e-06,
+      "loss": 0.927,
+      "step": 20147
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1379177143656825e-06,
+      "loss": 0.9854,
+      "step": 20148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1372838035854372e-06,
+      "loss": 0.9826,
+      "step": 20149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.136650058777763e-06,
+      "loss": 0.9583,
+      "step": 20150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.136016479954528e-06,
+      "loss": 0.9343,
+      "step": 20151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.135383067127593e-06,
+      "loss": 0.9528,
+      "step": 20152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.134749820308826e-06,
+      "loss": 0.9143,
+      "step": 20153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1341167395100803e-06,
+      "loss": 0.8824,
+      "step": 20154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1334838247432154e-06,
+      "loss": 0.962,
+      "step": 20155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1328510760200858e-06,
+      "loss": 0.9707,
+      "step": 20156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1322184933525325e-06,
+      "loss": 0.9527,
+      "step": 20157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.131586076752413e-06,
+      "loss": 0.9956,
+      "step": 20158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.130953826231562e-06,
+      "loss": 0.9685,
+      "step": 20159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.130321741801823e-06,
+      "loss": 1.0724,
+      "step": 20160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.129689823475033e-06,
+      "loss": 0.9822,
+      "step": 20161
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.129058071263026e-06,
+      "loss": 0.9994,
+      "step": 20162
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.128426485177635e-06,
+      "loss": 0.9741,
+      "step": 20163
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1277950652306802e-06,
+      "loss": 0.9158,
+      "step": 20164
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1271638114339964e-06,
+      "loss": 0.8973,
+      "step": 20165
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.126532723799398e-06,
+      "loss": 0.8903,
+      "step": 20166
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.125901802338706e-06,
+      "loss": 0.9303,
+      "step": 20167
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1252710470637342e-06,
+      "loss": 0.9543,
+      "step": 20168
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.124640457986297e-06,
+      "loss": 0.8821,
+      "step": 20169
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1240100351182036e-06,
+      "loss": 0.8919,
+      "step": 20170
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1233797784712575e-06,
+      "loss": 0.8772,
+      "step": 20171
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1227496880572616e-06,
+      "loss": 0.9877,
+      "step": 20172
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1221197638880174e-06,
+      "loss": 0.9298,
+      "step": 20173
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1214900059753218e-06,
+      "loss": 0.9678,
+      "step": 20174
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.120860414330966e-06,
+      "loss": 0.9861,
+      "step": 20175
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.120230988966745e-06,
+      "loss": 0.98,
+      "step": 20176
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1196017298944384e-06,
+      "loss": 0.9734,
+      "step": 20177
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1189726371258393e-06,
+      "loss": 0.992,
+      "step": 20178
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1183437106727223e-06,
+      "loss": 0.9444,
+      "step": 20179
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1177149505468674e-06,
+      "loss": 0.9853,
+      "step": 20180
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.117086356760051e-06,
+      "loss": 0.8755,
+      "step": 20181
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1164579293240395e-06,
+      "loss": 0.9255,
+      "step": 20182
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1158296682506086e-06,
+      "loss": 0.9297,
+      "step": 20183
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1152015735515175e-06,
+      "loss": 0.9659,
+      "step": 20184
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1145736452385344e-06,
+      "loss": 0.9725,
+      "step": 20185
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.113945883323415e-06,
+      "loss": 0.9349,
+      "step": 20186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1133182878179148e-06,
+      "loss": 0.946,
+      "step": 20187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1126908587337892e-06,
+      "loss": 0.9378,
+      "step": 20188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1120635960827864e-06,
+      "loss": 1.0015,
+      "step": 20189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1114364998766545e-06,
+      "loss": 0.9291,
+      "step": 20190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.110809570127136e-06,
+      "loss": 0.9047,
+      "step": 20191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1101828068459716e-06,
+      "loss": 0.9224,
+      "step": 20192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1095562100448987e-06,
+      "loss": 1.0015,
+      "step": 20193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.108929779735652e-06,
+      "loss": 0.9651,
+      "step": 20194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1083035159299627e-06,
+      "loss": 0.9247,
+      "step": 20195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.107677418639561e-06,
+      "loss": 0.9526,
+      "step": 20196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1070514878761672e-06,
+      "loss": 0.9709,
+      "step": 20197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1064257236515054e-06,
+      "loss": 0.9525,
+      "step": 20198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.105800125977295e-06,
+      "loss": 0.9511,
+      "step": 20199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1051746948652508e-06,
+      "loss": 0.9723,
+      "step": 20200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1045494303270873e-06,
+      "loss": 0.9564,
+      "step": 20201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1039243323745086e-06,
+      "loss": 0.9507,
+      "step": 20202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1032994010192288e-06,
+      "loss": 0.9509,
+      "step": 20203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1026746362729444e-06,
+      "loss": 0.9603,
+      "step": 20204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.102050038147357e-06,
+      "loss": 0.9565,
+      "step": 20205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.101425606654165e-06,
+      "loss": 0.9764,
+      "step": 20206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1008013418050613e-06,
+      "loss": 0.9588,
+      "step": 20207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1001772436117383e-06,
+      "loss": 0.895,
+      "step": 20208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0995533120858802e-06,
+      "loss": 0.9367,
+      "step": 20209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0989295472391723e-06,
+      "loss": 0.9757,
+      "step": 20210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.098305949083297e-06,
+      "loss": 0.9334,
+      "step": 20211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.097682517629932e-06,
+      "loss": 0.9851,
+      "step": 20212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.097059252890753e-06,
+      "loss": 0.9134,
+      "step": 20213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0964361548774327e-06,
+      "loss": 1.0731,
+      "step": 20214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.095813223601635e-06,
+      "loss": 0.856,
+      "step": 20215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0951904590750325e-06,
+      "loss": 0.8746,
+      "step": 20216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0945678613092824e-06,
+      "loss": 0.949,
+      "step": 20217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0939454303160458e-06,
+      "loss": 1.0207,
+      "step": 20218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.093323166106982e-06,
+      "loss": 0.9313,
+      "step": 20219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0927010686937366e-06,
+      "loss": 0.9468,
+      "step": 20220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0920791380879694e-06,
+      "loss": 0.9121,
+      "step": 20221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.091457374301318e-06,
+      "loss": 0.912,
+      "step": 20222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0908357773454334e-06,
+      "loss": 0.9836,
+      "step": 20223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0902143472319525e-06,
+      "loss": 0.9563,
+      "step": 20224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0895930839725146e-06,
+      "loss": 0.9973,
+      "step": 20225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.088971987578752e-06,
+      "loss": 0.9825,
+      "step": 20226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0883510580622976e-06,
+      "loss": 0.9676,
+      "step": 20227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0877302954347824e-06,
+      "loss": 0.9687,
+      "step": 20228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.087109699707826e-06,
+      "loss": 0.8731,
+      "step": 20229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0864892708930517e-06,
+      "loss": 0.9661,
+      "step": 20230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0858690090020796e-06,
+      "loss": 0.9989,
+      "step": 20231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0852489140465257e-06,
+      "loss": 0.9523,
+      "step": 20232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.084628986038002e-06,
+      "loss": 0.9191,
+      "step": 20233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0840092249881194e-06,
+      "loss": 0.9227,
+      "step": 20234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0833896309084812e-06,
+      "loss": 0.8741,
+      "step": 20235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.082770203810691e-06,
+      "loss": 0.9945,
+      "step": 20236
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0821509437063513e-06,
+      "loss": 0.9504,
+      "step": 20237
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.081531850607057e-06,
+      "loss": 0.9263,
+      "step": 20238
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0809129245244043e-06,
+      "loss": 0.9224,
+      "step": 20239
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0802941654699782e-06,
+      "loss": 0.9465,
+      "step": 20240
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.079675573455374e-06,
+      "loss": 0.9204,
+      "step": 20241
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0790571484921684e-06,
+      "loss": 0.951,
+      "step": 20242
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0784388905919505e-06,
+      "loss": 0.9365,
+      "step": 20243
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0778207997662922e-06,
+      "loss": 0.9687,
+      "step": 20244
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0772028760267715e-06,
+      "loss": 1.0209,
+      "step": 20245
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.076585119384962e-06,
+      "loss": 0.9726,
+      "step": 20246
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.075967529852424e-06,
+      "loss": 0.9422,
+      "step": 20247
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0753501074407347e-06,
+      "loss": 0.9174,
+      "step": 20248
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0747328521614484e-06,
+      "loss": 0.9553,
+      "step": 20249
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.074115764026128e-06,
+      "loss": 0.9711,
+      "step": 20250
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0734988430463276e-06,
+      "loss": 1.0073,
+      "step": 20251
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.072882089233602e-06,
+      "loss": 0.9711,
+      "step": 20252
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.072265502599501e-06,
+      "loss": 0.9544,
+      "step": 20253
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0716490831555735e-06,
+      "loss": 0.9844,
+      "step": 20254
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.071032830913359e-06,
+      "loss": 0.9964,
+      "step": 20255
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0704167458843993e-06,
+      "loss": 0.8794,
+      "step": 20256
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.069800828080233e-06,
+      "loss": 0.9206,
+      "step": 20257
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0691850775123933e-06,
+      "loss": 0.9695,
+      "step": 20258
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0685694941924148e-06,
+      "loss": 0.9937,
+      "step": 20259
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0679540781318184e-06,
+      "loss": 1.0107,
+      "step": 20260
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.067338829342138e-06,
+      "loss": 0.946,
+      "step": 20261
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0667237478348887e-06,
+      "loss": 0.9626,
+      "step": 20262
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.066108833621592e-06,
+      "loss": 0.9685,
+      "step": 20263
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.065494086713763e-06,
+      "loss": 0.9097,
+      "step": 20264
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0648795071229125e-06,
+      "loss": 0.8886,
+      "step": 20265
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0642650948605535e-06,
+      "loss": 0.9429,
+      "step": 20266
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.063650849938188e-06,
+      "loss": 0.9365,
+      "step": 20267
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.063036772367322e-06,
+      "loss": 0.966,
+      "step": 20268
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.062422862159452e-06,
+      "loss": 0.9311,
+      "step": 20269
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.061809119326078e-06,
+      "loss": 0.9982,
+      "step": 20270
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0611955438786925e-06,
+      "loss": 0.9541,
+      "step": 20271
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.060582135828788e-06,
+      "loss": 0.9639,
+      "step": 20272
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.059968895187847e-06,
+      "loss": 0.905,
+      "step": 20273
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.059355821967357e-06,
+      "loss": 0.961,
+      "step": 20274
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0587429161787988e-06,
+      "loss": 0.9397,
+      "step": 20275
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0581301778336508e-06,
+      "loss": 0.9537,
+      "step": 20276
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0575176069433877e-06,
+      "loss": 0.9528,
+      "step": 20277
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.056905203519477e-06,
+      "loss": 0.9164,
+      "step": 20278
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0562929675733933e-06,
+      "loss": 0.9085,
+      "step": 20279
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0556808991165967e-06,
+      "loss": 0.9005,
+      "step": 20280
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0550689981605565e-06,
+      "loss": 0.9974,
+      "step": 20281
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0544572647167247e-06,
+      "loss": 0.9494,
+      "step": 20282
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0538456987965594e-06,
+      "loss": 0.9624,
+      "step": 20283
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0532343004115176e-06,
+      "loss": 0.9669,
+      "step": 20284
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0526230695730397e-06,
+      "loss": 0.9535,
+      "step": 20285
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0520120062925833e-06,
+      "loss": 0.9811,
+      "step": 20286
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0514011105815835e-06,
+      "loss": 0.9367,
+      "step": 20287
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0507903824514844e-06,
+      "loss": 0.9134,
+      "step": 20288
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0501798219137215e-06,
+      "loss": 0.9618,
+      "step": 20289
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0495694289797296e-06,
+      "loss": 0.9903,
+      "step": 20290
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0489592036609397e-06,
+      "loss": 0.9651,
+      "step": 20291
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.048349145968781e-06,
+      "loss": 0.9932,
+      "step": 20292
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0477392559146747e-06,
+      "loss": 0.9182,
+      "step": 20293
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0471295335100437e-06,
+      "loss": 0.9305,
+      "step": 20294
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.046519978766307e-06,
+      "loss": 0.9339,
+      "step": 20295
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0459105916948795e-06,
+      "loss": 0.9705,
+      "step": 20296
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0453013723071746e-06,
+      "loss": 0.9315,
+      "step": 20297
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0446923206145964e-06,
+      "loss": 0.9538,
+      "step": 20298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0440834366285567e-06,
+      "loss": 0.9401,
+      "step": 20299
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0434747203604544e-06,
+      "loss": 0.9448,
+      "step": 20300
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.042866171821689e-06,
+      "loss": 0.9562,
+      "step": 20301
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.042257791023659e-06,
+      "loss": 0.9616,
+      "step": 20302
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0416495779777547e-06,
+      "loss": 0.9774,
+      "step": 20303
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0410415326953705e-06,
+      "loss": 0.9684,
+      "step": 20304
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0404336551878867e-06,
+      "loss": 1.0004,
+      "step": 20305
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0398259454666945e-06,
+      "loss": 0.8679,
+      "step": 20306
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0392184035431696e-06,
+      "loss": 1.0228,
+      "step": 20307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.03861102942869e-06,
+      "loss": 1.0171,
+      "step": 20308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.03800382313463e-06,
+      "loss": 0.9019,
+      "step": 20309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.037396784672363e-06,
+      "loss": 0.8544,
+      "step": 20310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0367899140532567e-06,
+      "loss": 0.943,
+      "step": 20311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.036183211288674e-06,
+      "loss": 0.882,
+      "step": 20312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0355766763899764e-06,
+      "loss": 0.936,
+      "step": 20313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0349703093685236e-06,
+      "loss": 0.9431,
+      "step": 20314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0343641102356714e-06,
+      "loss": 0.9839,
+      "step": 20315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.033758079002772e-06,
+      "loss": 0.9337,
+      "step": 20316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0331522156811767e-06,
+      "loss": 0.9393,
+      "step": 20317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.032546520282226e-06,
+      "loss": 0.9595,
+      "step": 20318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0319409928172684e-06,
+      "loss": 0.9799,
+      "step": 20319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0313356332976407e-06,
+      "loss": 0.9441,
+      "step": 20320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0307304417346794e-06,
+      "loss": 0.9075,
+      "step": 20321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0301254181397203e-06,
+      "loss": 0.987,
+      "step": 20322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.029520562524089e-06,
+      "loss": 0.9479,
+      "step": 20323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0289158748991192e-06,
+      "loss": 0.9184,
+      "step": 20324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.028311355276127e-06,
+      "loss": 0.9932,
+      "step": 20325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.027707003666443e-06,
+      "loss": 0.9277,
+      "step": 20326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.027102820081377e-06,
+      "loss": 0.9068,
+      "step": 20327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0264988045322455e-06,
+      "loss": 0.9509,
+      "step": 20328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.025894957030363e-06,
+      "loss": 0.9108,
+      "step": 20329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.025291277587033e-06,
+      "loss": 0.9758,
+      "step": 20330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0246877662135657e-06,
+      "loss": 0.9655,
+      "step": 20331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0240844229212576e-06,
+      "loss": 0.9831,
+      "step": 20332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0234812477214114e-06,
+      "loss": 0.9525,
+      "step": 20333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0228782406253222e-06,
+      "loss": 0.8992,
+      "step": 20334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0222754016442836e-06,
+      "loss": 0.9203,
+      "step": 20335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.02167273078958e-06,
+      "loss": 0.9874,
+      "step": 20336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0210702280725058e-06,
+      "loss": 0.9435,
+      "step": 20337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0204678935043366e-06,
+      "loss": 0.935,
+      "step": 20338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.019865727096355e-06,
+      "loss": 0.9582,
+      "step": 20339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0192637288598407e-06,
+      "loss": 0.9108,
+      "step": 20340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0186618988060603e-06,
+      "loss": 0.9917,
+      "step": 20341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0180602369462935e-06,
+      "loss": 0.9352,
+      "step": 20342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.017458743291798e-06,
+      "loss": 0.9187,
+      "step": 20343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.016857417853847e-06,
+      "loss": 0.943,
+      "step": 20344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0162562606436963e-06,
+      "loss": 0.9643,
+      "step": 20345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.015655271672604e-06,
+      "loss": 0.9457,
+      "step": 20346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.015054450951827e-06,
+      "loss": 0.9691,
+      "step": 20347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.014453798492614e-06,
+      "loss": 0.952,
+      "step": 20348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0138533143062176e-06,
+      "loss": 0.9323,
+      "step": 20349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0132529984038797e-06,
+      "loss": 1.027,
+      "step": 20350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0126528507968427e-06,
+      "loss": 0.9683,
+      "step": 20351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0120528714963462e-06,
+      "loss": 0.9514,
+      "step": 20352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0114530605136262e-06,
+      "loss": 0.8988,
+      "step": 20353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.010853417859916e-06,
+      "loss": 0.924,
+      "step": 20354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.010253943546445e-06,
+      "loss": 0.9594,
+      "step": 20355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.009654637584438e-06,
+      "loss": 0.9185,
+      "step": 20356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.00905549998512e-06,
+      "loss": 0.9378,
+      "step": 20357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0084565307597095e-06,
+      "loss": 0.9838,
+      "step": 20358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0078577299194245e-06,
+      "loss": 0.9421,
+      "step": 20359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0072590974754804e-06,
+      "loss": 0.9607,
+      "step": 20360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0066606334390826e-06,
+      "loss": 0.9744,
+      "step": 20361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.006062337821445e-06,
+      "loss": 0.986,
+      "step": 20362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0054642106337664e-06,
+      "loss": 0.9483,
+      "step": 20363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.004866251887253e-06,
+      "loss": 0.9596,
+      "step": 20364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0042684615930984e-06,
+      "loss": 1.0057,
+      "step": 20365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0036708397625003e-06,
+      "loss": 0.9551,
+      "step": 20366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.003073386406651e-06,
+      "loss": 0.9564,
+      "step": 20367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0024761015367335e-06,
+      "loss": 0.9473,
+      "step": 20368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.001878985163941e-06,
+      "loss": 1.0205,
+      "step": 20369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0012820372994494e-06,
+      "loss": 0.9645,
+      "step": 20370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0006852579544402e-06,
+      "loss": 0.9287,
+      "step": 20371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.0000886471400902e-06,
+      "loss": 0.9819,
+      "step": 20372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9994922048675713e-06,
+      "loss": 0.9649,
+      "step": 20373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.998895931148053e-06,
+      "loss": 0.9143,
+      "step": 20374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9982998259927034e-06,
+      "loss": 0.9538,
+      "step": 20375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9977038894126822e-06,
+      "loss": 0.9629,
+      "step": 20376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9971081214191523e-06,
+      "loss": 0.9547,
+      "step": 20377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9965125220232696e-06,
+      "loss": 0.9027,
+      "step": 20378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.995917091236188e-06,
+      "loss": 0.9825,
+      "step": 20379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9953218290690614e-06,
+      "loss": 0.9618,
+      "step": 20380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.994726735533029e-06,
+      "loss": 0.9603,
+      "step": 20381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9941318106392463e-06,
+      "loss": 0.9821,
+      "step": 20382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.993537054398845e-06,
+      "loss": 0.9502,
+      "step": 20383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.992942466822968e-06,
+      "loss": 0.9347,
+      "step": 20384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.992348047922749e-06,
+      "loss": 0.9396,
+      "step": 20385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9917537977093197e-06,
+      "loss": 0.9005,
+      "step": 20386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.99115971619381e-06,
+      "loss": 0.9854,
+      "step": 20387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.990565803387341e-06,
+      "loss": 0.952,
+      "step": 20388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9899720593010407e-06,
+      "loss": 0.9287,
+      "step": 20389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.989378483946024e-06,
+      "loss": 0.9576,
+      "step": 20390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9887850773334083e-06,
+      "loss": 0.9673,
+      "step": 20391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9881918394743073e-06,
+      "loss": 0.9965,
+      "step": 20392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9875987703798286e-06,
+      "loss": 0.9696,
+      "step": 20393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.987005870061082e-06,
+      "loss": 0.9089,
+      "step": 20394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.986413138529165e-06,
+      "loss": 0.9014,
+      "step": 20395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.985820575795183e-06,
+      "loss": 0.9187,
+      "step": 20396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.98522818187023e-06,
+      "loss": 0.9511,
+      "step": 20397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9846359567654016e-06,
+      "loss": 0.9541,
+      "step": 20398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9840439004917866e-06,
+      "loss": 0.9526,
+      "step": 20399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9834520130604762e-06,
+      "loss": 0.9045,
+      "step": 20400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.982860294482548e-06,
+      "loss": 0.9464,
+      "step": 20401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9822687447690927e-06,
+      "loss": 1.0143,
+      "step": 20402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9816773639311792e-06,
+      "loss": 0.9874,
+      "step": 20403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9810861519798862e-06,
+      "loss": 0.9064,
+      "step": 20404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.980495108926287e-06,
+      "loss": 0.9322,
+      "step": 20405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9799042347814446e-06,
+      "loss": 0.9099,
+      "step": 20406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9793135295564315e-06,
+      "loss": 1.0045,
+      "step": 20407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9787229932623043e-06,
+      "loss": 0.9899,
+      "step": 20408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.978132625910124e-06,
+      "loss": 0.9339,
+      "step": 20409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.977542427510947e-06,
+      "loss": 0.9146,
+      "step": 20410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9769523980758246e-06,
+      "loss": 0.8609,
+      "step": 20411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.976362537615807e-06,
+      "loss": 0.9629,
+      "step": 20412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9757728461419436e-06,
+      "loss": 0.9272,
+      "step": 20413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.975183323665272e-06,
+      "loss": 0.9761,
+      "step": 20414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9745939701968354e-06,
+      "loss": 0.923,
+      "step": 20415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9740047857476698e-06,
+      "loss": 0.9083,
+      "step": 20416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9734157703288093e-06,
+      "loss": 0.9371,
+      "step": 20417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9728269239512867e-06,
+      "loss": 0.8987,
+      "step": 20418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9722382466261227e-06,
+      "loss": 1.0108,
+      "step": 20419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.97164973836435e-06,
+      "loss": 0.9556,
+      "step": 20420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.971061399176982e-06,
+      "loss": 0.993,
+      "step": 20421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.970473229075042e-06,
+      "loss": 0.9561,
+      "step": 20422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9698852280695413e-06,
+      "loss": 0.9675,
+      "step": 20423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9692973961714936e-06,
+      "loss": 0.9551,
+      "step": 20424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9687097333919083e-06,
+      "loss": 1.0169,
+      "step": 20425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9681222397417843e-06,
+      "loss": 0.9843,
+      "step": 20426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9675349152321324e-06,
+      "loss": 0.9328,
+      "step": 20427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9669477598739454e-06,
+      "loss": 0.9697,
+      "step": 20428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.966360773678221e-06,
+      "loss": 0.9946,
+      "step": 20429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.965773956655951e-06,
+      "loss": 0.9911,
+      "step": 20430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.965187308818125e-06,
+      "loss": 0.9766,
+      "step": 20431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9646008301757316e-06,
+      "loss": 0.8815,
+      "step": 20432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.96401452073975e-06,
+      "loss": 0.9821,
+      "step": 20433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9634283805211606e-06,
+      "loss": 0.9168,
+      "step": 20434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.962842409530942e-06,
+      "loss": 0.892,
+      "step": 20435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9622566077800668e-06,
+      "loss": 0.8891,
+      "step": 20436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.961670975279505e-06,
+      "loss": 0.9512,
+      "step": 20437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.961085512040226e-06,
+      "loss": 0.9954,
+      "step": 20438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.960500218073189e-06,
+      "loss": 0.9275,
+      "step": 20439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9599150933893605e-06,
+      "loss": 0.9615,
+      "step": 20440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9593301379996943e-06,
+      "loss": 0.9716,
+      "step": 20441
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9587453519151467e-06,
+      "loss": 0.9948,
+      "step": 20442
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.958160735146669e-06,
+      "loss": 0.937,
+      "step": 20443
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9575762877052052e-06,
+      "loss": 0.93,
+      "step": 20444
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9569920096017084e-06,
+      "loss": 0.9839,
+      "step": 20445
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9564079008471113e-06,
+      "loss": 0.9564,
+      "step": 20446
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.955823961452361e-06,
+      "loss": 0.9371,
+      "step": 20447
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.955240191428387e-06,
+      "loss": 0.9272,
+      "step": 20448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.954656590786124e-06,
+      "loss": 0.9534,
+      "step": 20449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9540731595365005e-06,
+      "loss": 0.8516,
+      "step": 20450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9534898976904427e-06,
+      "loss": 0.9695,
+      "step": 20451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.952906805258875e-06,
+      "loss": 0.9433,
+      "step": 20452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9523238822527137e-06,
+      "loss": 0.9475,
+      "step": 20453
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.951741128682877e-06,
+      "loss": 0.923,
+      "step": 20454
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9511585445602778e-06,
+      "loss": 0.9392,
+      "step": 20455
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9505761298958257e-06,
+      "loss": 0.9193,
+      "step": 20456
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9499938847004297e-06,
+      "loss": 0.9431,
+      "step": 20457
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9494118089849936e-06,
+      "loss": 0.8982,
+      "step": 20458
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.948829902760414e-06,
+      "loss": 0.9502,
+      "step": 20459
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9482481660375918e-06,
+      "loss": 0.95,
+      "step": 20460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.947666598827419e-06,
+      "loss": 0.9928,
+      "step": 20461
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9470852011407893e-06,
+      "loss": 0.9603,
+      "step": 20462
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9465039729885905e-06,
+      "loss": 0.9446,
+      "step": 20463
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9459229143817026e-06,
+      "loss": 0.984,
+      "step": 20464
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9453420253310127e-06,
+      "loss": 0.9517,
+      "step": 20465
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9447613058473945e-06,
+      "loss": 0.927,
+      "step": 20466
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.944180755941729e-06,
+      "loss": 0.9575,
+      "step": 20467
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9436003756248836e-06,
+      "loss": 0.9566,
+      "step": 20468
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9430201649077274e-06,
+      "loss": 0.8736,
+      "step": 20469
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9424401238011292e-06,
+      "loss": 0.9373,
+      "step": 20470
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.941860252315945e-06,
+      "loss": 0.91,
+      "step": 20471
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9412805504630416e-06,
+      "loss": 0.9788,
+      "step": 20472
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9407010182532715e-06,
+      "loss": 0.9154,
+      "step": 20473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9401216556974866e-06,
+      "loss": 0.9776,
+      "step": 20474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9395424628065385e-06,
+      "loss": 0.91,
+      "step": 20475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.938963439591275e-06,
+      "loss": 0.9699,
+      "step": 20476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.938384586062533e-06,
+      "loss": 0.9449,
+      "step": 20477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9378059022311626e-06,
+      "loss": 0.905,
+      "step": 20478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9372273881079927e-06,
+      "loss": 0.9429,
+      "step": 20479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9366490437038608e-06,
+      "loss": 0.9182,
+      "step": 20480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9360708690295994e-06,
+      "loss": 0.8319,
+      "step": 20481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.935492864096028e-06,
+      "loss": 0.9468,
+      "step": 20482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.934915028913982e-06,
+      "loss": 0.9661,
+      "step": 20483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.934337363494273e-06,
+      "loss": 0.9084,
+      "step": 20484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9337598678477254e-06,
+      "loss": 1.0394,
+      "step": 20485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9331825419851512e-06,
+      "loss": 0.9049,
+      "step": 20486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.932605385917362e-06,
+      "loss": 0.9508,
+      "step": 20487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.932028399655167e-06,
+      "loss": 0.9346,
+      "step": 20488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9314515832093705e-06,
+      "loss": 0.9836,
+      "step": 20489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9308749365907766e-06,
+      "loss": 0.9515,
+      "step": 20490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9302984598101817e-06,
+      "loss": 1.0361,
+      "step": 20491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.929722152878383e-06,
+      "loss": 0.9376,
+      "step": 20492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9291460158061714e-06,
+      "loss": 0.9562,
+      "step": 20493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.928570048604338e-06,
+      "loss": 0.9359,
+      "step": 20494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.927994251283669e-06,
+      "loss": 0.9898,
+      "step": 20495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9274186238549476e-06,
+      "loss": 0.9785,
+      "step": 20496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.926843166328952e-06,
+      "loss": 0.9492,
+      "step": 20497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.926267878716459e-06,
+      "loss": 0.9445,
+      "step": 20498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9256927610282427e-06,
+      "loss": 1.0026,
+      "step": 20499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.925117813275074e-06,
+      "loss": 0.9451,
+      "step": 20500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92454303546772e-06,
+      "loss": 0.9825,
+      "step": 20501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9239684276169413e-06,
+      "loss": 0.9426,
+      "step": 20502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9233939897335042e-06,
+      "loss": 0.9034,
+      "step": 20503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92281972182816e-06,
+      "loss": 0.9148,
+      "step": 20504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9222456239116706e-06,
+      "loss": 1.0234,
+      "step": 20505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.92167169599478e-06,
+      "loss": 0.9168,
+      "step": 20506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.921097938088239e-06,
+      "loss": 0.916,
+      "step": 20507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9205243502027948e-06,
+      "loss": 0.998,
+      "step": 20508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.919950932349183e-06,
+      "loss": 0.9388,
+      "step": 20509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9193776845381496e-06,
+      "loss": 0.936,
+      "step": 20510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9188046067804234e-06,
+      "loss": 0.9197,
+      "step": 20511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.918231699086739e-06,
+      "loss": 0.917,
+      "step": 20512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9176589614678253e-06,
+      "loss": 0.9428,
+      "step": 20513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9170863939344087e-06,
+      "loss": 0.9399,
+      "step": 20514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9165139964972125e-06,
+      "loss": 0.8909,
+      "step": 20515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9159417691669524e-06,
+      "loss": 0.9448,
+      "step": 20516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.915369711954347e-06,
+      "loss": 0.9147,
+      "step": 20517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9147978248701084e-06,
+      "loss": 0.9583,
+      "step": 20518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.914226107924946e-06,
+      "loss": 0.8971,
+      "step": 20519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.913654561129569e-06,
+      "loss": 0.8912,
+      "step": 20520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9130831844946796e-06,
+      "loss": 1.0194,
+      "step": 20521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9125119780309744e-06,
+      "loss": 0.9459,
+      "step": 20522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9119409417491564e-06,
+      "loss": 0.9348,
+      "step": 20523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9113700756599142e-06,
+      "loss": 0.9662,
+      "step": 20524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.910799379773942e-06,
+      "loss": 0.8953,
+      "step": 20525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9102288541019264e-06,
+      "loss": 0.9884,
+      "step": 20526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9096584986545484e-06,
+      "loss": 0.9321,
+      "step": 20527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.909088313442495e-06,
+      "loss": 0.9527,
+      "step": 20528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9085182984764385e-06,
+      "loss": 1.0092,
+      "step": 20529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.907948453767058e-06,
+      "loss": 0.9198,
+      "step": 20530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9073787793250223e-06,
+      "loss": 0.9249,
+      "step": 20531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9068092751610004e-06,
+      "loss": 0.8585,
+      "step": 20532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.906239941285657e-06,
+      "loss": 0.9166,
+      "step": 20533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.905670777709655e-06,
+      "loss": 0.9504,
+      "step": 20534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9051017844436547e-06,
+      "loss": 0.979,
+      "step": 20535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9045329614983077e-06,
+      "loss": 0.8879,
+      "step": 20536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9039643088842674e-06,
+      "loss": 0.9146,
+      "step": 20537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9033958266121854e-06,
+      "loss": 0.9022,
+      "step": 20538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9028275146927057e-06,
+      "loss": 0.9599,
+      "step": 20539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9022593731364715e-06,
+      "loss": 0.9874,
+      "step": 20540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9016914019541243e-06,
+      "loss": 0.9658,
+      "step": 20541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.9011236011562975e-06,
+      "loss": 0.9365,
+      "step": 20542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.900555970753626e-06,
+      "loss": 0.9797,
+      "step": 20543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8999885107567386e-06,
+      "loss": 0.9777,
+      "step": 20544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.899421221176263e-06,
+      "loss": 0.9898,
+      "step": 20545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8988541020228246e-06,
+      "loss": 1.0103,
+      "step": 20546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.898287153307039e-06,
+      "loss": 0.9334,
+      "step": 20547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.897720375039531e-06,
+      "loss": 0.9839,
+      "step": 20548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8971537672309085e-06,
+      "loss": 0.8936,
+      "step": 20549
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8965873298917826e-06,
+      "loss": 1.0059,
+      "step": 20550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.896021063032764e-06,
+      "loss": 0.9617,
+      "step": 20551
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.895454966664455e-06,
+      "loss": 0.9532,
+      "step": 20552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.894889040797461e-06,
+      "loss": 0.9759,
+      "step": 20553
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.894323285442373e-06,
+      "loss": 0.8931,
+      "step": 20554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8937577006097917e-06,
+      "loss": 0.9106,
+      "step": 20555
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8931922863103057e-06,
+      "loss": 0.8978,
+      "step": 20556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8926270425545056e-06,
+      "loss": 1.0015,
+      "step": 20557
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8920619693529762e-06,
+      "loss": 0.9281,
+      "step": 20558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8914970667163e-06,
+      "loss": 0.9813,
+      "step": 20559
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8909323346550524e-06,
+      "loss": 0.9166,
+      "step": 20560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8903677731798164e-06,
+      "loss": 0.9698,
+      "step": 20561
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8898033823011575e-06,
+      "loss": 0.9232,
+      "step": 20562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8892391620296493e-06,
+      "loss": 1.0445,
+      "step": 20563
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8886751123758572e-06,
+      "loss": 0.8965,
+      "step": 20564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8881112333503404e-06,
+      "loss": 0.9776,
+      "step": 20565
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8875475249636655e-06,
+      "loss": 0.9715,
+      "step": 20566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8869839872263807e-06,
+      "loss": 0.9485,
+      "step": 20567
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.886420620149049e-06,
+      "loss": 1.0413,
+      "step": 20568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8858574237422143e-06,
+      "loss": 0.9575,
+      "step": 20569
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.885294398016424e-06,
+      "loss": 0.9717,
+      "step": 20570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.884731542982223e-06,
+      "loss": 0.9228,
+      "step": 20571
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8841688586501526e-06,
+      "loss": 0.9771,
+      "step": 20572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.883606345030752e-06,
+      "loss": 0.9455,
+      "step": 20573
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.88304400213455e-06,
+      "loss": 0.9446,
+      "step": 20574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8824818299720817e-06,
+      "loss": 0.969,
+      "step": 20575
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.881919828553872e-06,
+      "loss": 0.9982,
+      "step": 20576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8813579978904485e-06,
+      "loss": 0.9179,
+      "step": 20577
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.880796337992331e-06,
+      "loss": 0.9628,
+      "step": 20578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8802348488700407e-06,
+      "loss": 0.9976,
+      "step": 20579
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8796735305340886e-06,
+      "loss": 0.9869,
+      "step": 20580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8791123829949875e-06,
+      "loss": 0.9616,
+      "step": 20581
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.878551406263246e-06,
+      "loss": 0.8546,
+      "step": 20582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.877990600349371e-06,
+      "loss": 0.9405,
+      "step": 20583
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.877429965263866e-06,
+      "loss": 0.8927,
+      "step": 20584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8768695010172233e-06,
+      "loss": 0.9898,
+      "step": 20585
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8763092076199476e-06,
+      "loss": 0.9212,
+      "step": 20586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8757490850825234e-06,
+      "loss": 0.9923,
+      "step": 20587
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8751891334154484e-06,
+      "loss": 0.899,
+      "step": 20588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8746293526292025e-06,
+      "loss": 0.9314,
+      "step": 20589
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8740697427342703e-06,
+      "loss": 0.9369,
+      "step": 20590
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.873510303741135e-06,
+      "loss": 0.9568,
+      "step": 20591
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8729510356602654e-06,
+      "loss": 1.0184,
+      "step": 20592
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.872391938502145e-06,
+      "loss": 0.961,
+      "step": 20593
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8718330122772357e-06,
+      "loss": 1.0026,
+      "step": 20594
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8712742569960084e-06,
+      "loss": 1.0008,
+      "step": 20595
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.870715672668927e-06,
+      "loss": 0.9594,
+      "step": 20596
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.87015725930645e-06,
+      "loss": 0.9669,
+      "step": 20597
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.869599016919037e-06,
+      "loss": 0.9749,
+      "step": 20598
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8690409455171443e-06,
+      "loss": 0.9032,
+      "step": 20599
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.868483045111218e-06,
+      "loss": 0.9628,
+      "step": 20600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8679253157117073e-06,
+      "loss": 0.9357,
+      "step": 20601
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8673677573290584e-06,
+      "loss": 0.9325,
+      "step": 20602
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.866810369973712e-06,
+      "loss": 0.9237,
+      "step": 20603
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.866253153656107e-06,
+      "loss": 0.9644,
+      "step": 20604
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8656961083866754e-06,
+      "loss": 0.9695,
+      "step": 20605
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8651392341758535e-06,
+      "loss": 0.9844,
+      "step": 20606
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.864582531034067e-06,
+      "loss": 0.9633,
+      "step": 20607
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8640259989717426e-06,
+      "loss": 0.9871,
+      "step": 20608
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.863469637999301e-06,
+      "loss": 0.9662,
+      "step": 20609
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8629134481271627e-06,
+      "loss": 0.9391,
+      "step": 20610
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.862357429365744e-06,
+      "loss": 0.893,
+      "step": 20611
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8618015817254553e-06,
+      "loss": 0.9327,
+      "step": 20612
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8612459052167077e-06,
+      "loss": 0.9617,
+      "step": 20613
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.860690399849906e-06,
+      "loss": 0.9166,
+      "step": 20614
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8601350656354547e-06,
+      "loss": 0.8851,
+      "step": 20615
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8595799025837512e-06,
+      "loss": 0.901,
+      "step": 20616
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8590249107051973e-06,
+      "loss": 0.9701,
+      "step": 20617
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8584700900101804e-06,
+      "loss": 1.0312,
+      "step": 20618
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8579154405090925e-06,
+      "loss": 0.9332,
+      "step": 20619
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8573609622123224e-06,
+      "loss": 1.0026,
+      "step": 20620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.856806655130252e-06,
+      "loss": 0.9594,
+      "step": 20621
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.856252519273264e-06,
+      "loss": 0.9528,
+      "step": 20622
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8556985546517314e-06,
+      "loss": 0.9459,
+      "step": 20623
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8551447612760333e-06,
+      "loss": 0.9549,
+      "step": 20624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8545911391565354e-06,
+      "loss": 0.9496,
+      "step": 20625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8540376883036126e-06,
+      "loss": 0.9046,
+      "step": 20626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8534844087276225e-06,
+      "loss": 0.9581,
+      "step": 20627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8529313004389294e-06,
+      "loss": 0.8815,
+      "step": 20628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.852378363447893e-06,
+      "loss": 0.9623,
+      "step": 20629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8518255977648627e-06,
+      "loss": 1.0017,
+      "step": 20630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8512730034001982e-06,
+      "loss": 0.8551,
+      "step": 20631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8507205803642405e-06,
+      "loss": 1.0084,
+      "step": 20632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.850168328667338e-06,
+      "loss": 0.9288,
+      "step": 20633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8496162483198323e-06,
+      "loss": 0.9249,
+      "step": 20634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.849064339332063e-06,
+      "loss": 1.0208,
+      "step": 20635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.848512601714366e-06,
+      "loss": 0.9317,
+      "step": 20636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.847961035477074e-06,
+      "loss": 0.953,
+      "step": 20637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.847409640630513e-06,
+      "loss": 0.9523,
+      "step": 20638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8468584171850124e-06,
+      "loss": 0.9716,
+      "step": 20639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8463073651508935e-06,
+      "loss": 0.924,
+      "step": 20640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.845756484538476e-06,
+      "loss": 0.9613,
+      "step": 20641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.845205775358078e-06,
+      "loss": 0.9488,
+      "step": 20642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8446552376200067e-06,
+      "loss": 0.941,
+      "step": 20643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.844104871334581e-06,
+      "loss": 0.9293,
+      "step": 20644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8435546765121014e-06,
+      "loss": 0.9909,
+      "step": 20645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.843004653162873e-06,
+      "loss": 0.9191,
+      "step": 20646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8424548012971965e-06,
+      "loss": 0.925,
+      "step": 20647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8419051209253693e-06,
+      "loss": 0.9078,
+      "step": 20648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.841355612057686e-06,
+      "loss": 0.8827,
+      "step": 20649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.840806274704432e-06,
+      "loss": 0.9538,
+      "step": 20650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8402571088759028e-06,
+      "loss": 0.9597,
+      "step": 20651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.839708114582377e-06,
+      "loss": 0.8796,
+      "step": 20652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8391592918341374e-06,
+      "loss": 0.9649,
+      "step": 20653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8386106406414605e-06,
+      "loss": 0.9878,
+      "step": 20654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8380621610146234e-06,
+      "loss": 0.911,
+      "step": 20655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8375138529638983e-06,
+      "loss": 0.9122,
+      "step": 20656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.836965716499548e-06,
+      "loss": 0.9759,
+      "step": 20657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8364177516318427e-06,
+      "loss": 0.9402,
+      "step": 20658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.835869958371042e-06,
+      "loss": 0.9905,
+      "step": 20659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8353223367274034e-06,
+      "loss": 0.9447,
+      "step": 20660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8347748867111846e-06,
+      "loss": 0.963,
+      "step": 20661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8342276083326384e-06,
+      "loss": 0.9631,
+      "step": 20662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8336805016020087e-06,
+      "loss": 0.9182,
+      "step": 20663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.833133566529548e-06,
+      "loss": 0.9885,
+      "step": 20664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8325868031254933e-06,
+      "loss": 0.9705,
+      "step": 20665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8320402114000867e-06,
+      "loss": 0.9758,
+      "step": 20666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8314937913635643e-06,
+      "loss": 1.0236,
+      "step": 20667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8309475430261547e-06,
+      "loss": 0.9066,
+      "step": 20668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8304014663980954e-06,
+      "loss": 0.9316,
+      "step": 20669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8298555614896044e-06,
+      "loss": 0.9898,
+      "step": 20670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8293098283109113e-06,
+      "loss": 0.919,
+      "step": 20671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8287642668722327e-06,
+      "loss": 0.9264,
+      "step": 20672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8282188771837863e-06,
+      "loss": 0.9012,
+      "step": 20673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8276736592557856e-06,
+      "loss": 0.9725,
+      "step": 20674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82712861309844e-06,
+      "loss": 0.8796,
+      "step": 20675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82658373872196e-06,
+      "loss": 0.9136,
+      "step": 20676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8260390361365454e-06,
+      "loss": 0.9571,
+      "step": 20677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.825494505352399e-06,
+      "loss": 0.9905,
+      "step": 20678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.824950146379716e-06,
+      "loss": 0.9316,
+      "step": 20679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8244059592286956e-06,
+      "loss": 0.9735,
+      "step": 20680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.823861943909522e-06,
+      "loss": 0.9196,
+      "step": 20681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8233181004323897e-06,
+      "loss": 1.0477,
+      "step": 20682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.822774428807479e-06,
+      "loss": 0.9685,
+      "step": 20683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8222309290449712e-06,
+      "loss": 0.9332,
+      "step": 20684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.821687601155049e-06,
+      "loss": 0.9414,
+      "step": 20685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.82114444514788e-06,
+      "loss": 0.918,
+      "step": 20686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.820601461033643e-06,
+      "loss": 0.8991,
+      "step": 20687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8200586488225006e-06,
+      "loss": 0.9034,
+      "step": 20688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8195160085246246e-06,
+      "loss": 0.9294,
+      "step": 20689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8189735401501707e-06,
+      "loss": 0.9111,
+      "step": 20690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.818431243709301e-06,
+      "loss": 0.9563,
+      "step": 20691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8178891192121705e-06,
+      "loss": 0.8547,
+      "step": 20692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8173471666689315e-06,
+      "loss": 0.9616,
+      "step": 20693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8168053860897352e-06,
+      "loss": 0.9364,
+      "step": 20694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8162637774847237e-06,
+      "loss": 0.9678,
+      "step": 20695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8157223408640424e-06,
+      "loss": 0.9378,
+      "step": 20696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8151810762378288e-06,
+      "loss": 0.9056,
+      "step": 20697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.814639983616222e-06,
+      "loss": 0.8981,
+      "step": 20698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8140990630093525e-06,
+      "loss": 0.9419,
+      "step": 20699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8135583144273538e-06,
+      "loss": 0.9852,
+      "step": 20700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.813017737880347e-06,
+      "loss": 0.9394,
+      "step": 20701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.81247733337846e-06,
+      "loss": 0.9053,
+      "step": 20702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.81193710093181e-06,
+      "loss": 0.9151,
+      "step": 20703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8113970405505165e-06,
+      "loss": 0.8817,
+      "step": 20704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8108571522446926e-06,
+      "loss": 0.9551,
+      "step": 20705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.810317436024446e-06,
+      "loss": 0.9555,
+      "step": 20706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8097778918998896e-06,
+      "loss": 0.9908,
+      "step": 20707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8092385198811203e-06,
+      "loss": 0.9365,
+      "step": 20708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8086993199782477e-06,
+      "loss": 0.92,
+      "step": 20709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8081602922013618e-06,
+      "loss": 0.91,
+      "step": 20710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8076214365605595e-06,
+      "loss": 0.9895,
+      "step": 20711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.807082753065934e-06,
+      "loss": 0.9335,
+      "step": 20712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.806544241727569e-06,
+      "loss": 0.8857,
+      "step": 20713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8060059025555544e-06,
+      "loss": 0.948,
+      "step": 20714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8054677355599684e-06,
+      "loss": 0.9374,
+      "step": 20715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.804929740750889e-06,
+      "loss": 0.9084,
+      "step": 20716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.804391918138393e-06,
+      "loss": 0.9593,
+      "step": 20717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.80385426773255e-06,
+      "loss": 0.9435,
+      "step": 20718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.803316789543431e-06,
+      "loss": 0.8745,
+      "step": 20719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.802779483581103e-06,
+      "loss": 0.9046,
+      "step": 20720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.802242349855623e-06,
+      "loss": 0.9318,
+      "step": 20721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.801705388377052e-06,
+      "loss": 0.9324,
+      "step": 20722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.801168599155447e-06,
+      "loss": 0.9552,
+      "step": 20723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.80063198220086e-06,
+      "loss": 0.9167,
+      "step": 20724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.800095537523341e-06,
+      "loss": 0.9075,
+      "step": 20725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.799559265132932e-06,
+      "loss": 0.9174,
+      "step": 20726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7990231650396817e-06,
+      "loss": 0.9396,
+      "step": 20727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7984872372536255e-06,
+      "loss": 0.9597,
+      "step": 20728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7979514817848012e-06,
+      "loss": 0.945,
+      "step": 20729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.797415898643241e-06,
+      "loss": 0.9005,
+      "step": 20730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7968804878389767e-06,
+      "loss": 0.9396,
+      "step": 20731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7963452493820347e-06,
+      "loss": 0.893,
+      "step": 20732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7958101832824336e-06,
+      "loss": 0.9371,
+      "step": 20733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7952752895502023e-06,
+      "loss": 0.9529,
+      "step": 20734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.794740568195351e-06,
+      "loss": 0.9659,
+      "step": 20735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7942060192278953e-06,
+      "loss": 0.9523,
+      "step": 20736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.793671642657846e-06,
+      "loss": 0.9367,
+      "step": 20737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7931374384952107e-06,
+      "loss": 0.8793,
+      "step": 20738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7926034067499943e-06,
+      "loss": 0.9101,
+      "step": 20739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.792069547432195e-06,
+      "loss": 1.0087,
+      "step": 20740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.791535860551811e-06,
+      "loss": 0.9672,
+      "step": 20741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7910023461188386e-06,
+      "loss": 0.9532,
+      "step": 20742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.790469004143267e-06,
+      "loss": 0.9372,
+      "step": 20743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.789935834635086e-06,
+      "loss": 0.9692,
+      "step": 20744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7894028376042804e-06,
+      "loss": 0.9834,
+      "step": 20745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.788870013060827e-06,
+      "loss": 0.9999,
+      "step": 20746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7883373610147116e-06,
+      "loss": 0.9008,
+      "step": 20747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.787804881475903e-06,
+      "loss": 0.9265,
+      "step": 20748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7872725744543762e-06,
+      "loss": 0.9662,
+      "step": 20749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7867404399600998e-06,
+      "loss": 0.9815,
+      "step": 20750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7862084780030342e-06,
+      "loss": 0.9576,
+      "step": 20751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7856766885931496e-06,
+      "loss": 0.9643,
+      "step": 20752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7851450717403986e-06,
+      "loss": 0.9186,
+      "step": 20753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7846136274547377e-06,
+      "loss": 0.9035,
+      "step": 20754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.784082355746122e-06,
+      "loss": 1.0148,
+      "step": 20755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7835512566244985e-06,
+      "loss": 0.8972,
+      "step": 20756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.783020330099813e-06,
+      "loss": 0.9627,
+      "step": 20757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7824895761820116e-06,
+      "loss": 0.9899,
+      "step": 20758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7819589948810277e-06,
+      "loss": 0.9542,
+      "step": 20759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7814285862068026e-06,
+      "loss": 0.9198,
+      "step": 20760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.780898350169267e-06,
+      "loss": 0.9514,
+      "step": 20761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.78036828677835e-06,
+      "loss": 0.9598,
+      "step": 20762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7798383960439826e-06,
+      "loss": 0.9325,
+      "step": 20763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7793086779760804e-06,
+      "loss": 0.9692,
+      "step": 20764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.778779132584572e-06,
+      "loss": 0.9588,
+      "step": 20765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.778249759879368e-06,
+      "loss": 0.866,
+      "step": 20766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.777720559870384e-06,
+      "loss": 0.9692,
+      "step": 20767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7771915325675303e-06,
+      "loss": 0.9513,
+      "step": 20768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.776662677980715e-06,
+      "loss": 0.9791,
+      "step": 20769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.776133996119842e-06,
+      "loss": 0.9785,
+      "step": 20770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.775605486994808e-06,
+      "loss": 0.9476,
+      "step": 20771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7750771506155175e-06,
+      "loss": 0.9113,
+      "step": 20772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7745489869918577e-06,
+      "loss": 0.9438,
+      "step": 20773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.774020996133723e-06,
+      "loss": 0.9408,
+      "step": 20774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7734931780510014e-06,
+      "loss": 0.9873,
+      "step": 20775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7729655327535767e-06,
+      "loss": 0.9752,
+      "step": 20776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7724380602513303e-06,
+      "loss": 0.9356,
+      "step": 20777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7719107605541397e-06,
+      "loss": 0.9556,
+      "step": 20778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7713836336718793e-06,
+      "loss": 0.9181,
+      "step": 20779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.770856679614422e-06,
+      "loss": 0.9777,
+      "step": 20780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7703298983916352e-06,
+      "loss": 0.9449,
+      "step": 20781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.769803290013383e-06,
+      "loss": 0.9061,
+      "step": 20782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7692768544895313e-06,
+      "loss": 0.9923,
+      "step": 20783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.768750591829932e-06,
+      "loss": 0.9673,
+      "step": 20784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.768224502044448e-06,
+      "loss": 0.927,
+      "step": 20785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.767698585142925e-06,
+      "loss": 0.933,
+      "step": 20786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7671728411352153e-06,
+      "loss": 0.9751,
+      "step": 20787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7666472700311654e-06,
+      "loss": 0.9656,
+      "step": 20788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7661218718406128e-06,
+      "loss": 0.9288,
+      "step": 20789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.765596646573402e-06,
+      "loss": 0.9959,
+      "step": 20790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.765071594239365e-06,
+      "loss": 0.9537,
+      "step": 20791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7645467148483384e-06,
+      "loss": 0.9553,
+      "step": 20792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.764022008410149e-06,
+      "loss": 0.9618,
+      "step": 20793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7634974749346236e-06,
+      "loss": 0.854,
+      "step": 20794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.762973114431584e-06,
+      "loss": 0.9608,
+      "step": 20795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7624489269108516e-06,
+      "loss": 0.9875,
+      "step": 20796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7619249123822434e-06,
+      "loss": 0.9307,
+      "step": 20797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7614010708555717e-06,
+      "loss": 0.963,
+      "step": 20798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.760877402340645e-06,
+      "loss": 0.9302,
+      "step": 20799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.760353906847272e-06,
+      "loss": 0.9093,
+      "step": 20800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.759830584385255e-06,
+      "loss": 0.9189,
+      "step": 20801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7593074349643958e-06,
+      "loss": 0.9598,
+      "step": 20802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7587844585944912e-06,
+      "loss": 0.9478,
+      "step": 20803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.758261655285334e-06,
+      "loss": 0.9836,
+      "step": 20804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.757739025046714e-06,
+      "loss": 1.0038,
+      "step": 20805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.757216567888421e-06,
+      "loss": 0.8618,
+      "step": 20806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7566942838202374e-06,
+      "loss": 0.9595,
+      "step": 20807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7561721728519456e-06,
+      "loss": 0.9524,
+      "step": 20808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755650234993319e-06,
+      "loss": 0.9633,
+      "step": 20809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.755128470254138e-06,
+      "loss": 0.9455,
+      "step": 20810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.754606878644168e-06,
+      "loss": 0.9083,
+      "step": 20811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7540854601731837e-06,
+      "loss": 0.9612,
+      "step": 20812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7535642148509426e-06,
+      "loss": 1.0457,
+      "step": 20813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.75304314268721e-06,
+      "loss": 0.9668,
+      "step": 20814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7525222436917445e-06,
+      "loss": 0.9407,
+      "step": 20815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7520015178742976e-06,
+      "loss": 0.8851,
+      "step": 20816
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7514809652446232e-06,
+      "loss": 0.9119,
+      "step": 20817
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7509605858124703e-06,
+      "loss": 0.9375,
+      "step": 20818
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7504403795875825e-06,
+      "loss": 0.8988,
+      "step": 20819
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.749920346579702e-06,
+      "loss": 0.9719,
+      "step": 20820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7494004867985714e-06,
+      "loss": 0.9304,
+      "step": 20821
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7488808002539183e-06,
+      "loss": 1.0068,
+      "step": 20822
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7483612869554834e-06,
+      "loss": 0.9198,
+      "step": 20823
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7478419469129894e-06,
+      "loss": 0.9538,
+      "step": 20824
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7473227801361648e-06,
+      "loss": 0.9495,
+      "step": 20825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7468037866347323e-06,
+      "loss": 0.9414,
+      "step": 20826
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7462849664184078e-06,
+      "loss": 0.8945,
+      "step": 20827
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7457663194969122e-06,
+      "loss": 0.895,
+      "step": 20828
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.745247845879954e-06,
+      "loss": 0.9045,
+      "step": 20829
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7447295455772463e-06,
+      "loss": 1.0134,
+      "step": 20830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7442114185984928e-06,
+      "loss": 0.9498,
+      "step": 20831
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.743693464953398e-06,
+      "loss": 0.9149,
+      "step": 20832
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.743175684651659e-06,
+      "loss": 0.9888,
+      "step": 20833
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.742658077702976e-06,
+      "loss": 0.9759,
+      "step": 20834
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.742140644117041e-06,
+      "loss": 0.8965,
+      "step": 20835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7416233839035423e-06,
+      "loss": 0.9618,
+      "step": 20836
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.741106297072167e-06,
+      "loss": 0.9536,
+      "step": 20837
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7405893836326004e-06,
+      "loss": 0.9973,
+      "step": 20838
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7400726435945216e-06,
+      "loss": 0.8938,
+      "step": 20839
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7395560769676067e-06,
+      "loss": 0.9147,
+      "step": 20840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7390396837615343e-06,
+      "loss": 0.9575,
+      "step": 20841
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7385234639859686e-06,
+      "loss": 0.9107,
+      "step": 20842
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.738007417650579e-06,
+      "loss": 0.9322,
+      "step": 20843
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73749154476503e-06,
+      "loss": 0.9285,
+      "step": 20844
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.736975845338983e-06,
+      "loss": 0.9042,
+      "step": 20845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7364603193820963e-06,
+      "loss": 0.9256,
+      "step": 20846
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73594496690402e-06,
+      "loss": 0.9231,
+      "step": 20847
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73542978791441e-06,
+      "loss": 0.9586,
+      "step": 20848
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7349147824229095e-06,
+      "loss": 0.9805,
+      "step": 20849
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7343999504391697e-06,
+      "loss": 0.9044,
+      "step": 20850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7338852919728244e-06,
+      "loss": 0.9454,
+      "step": 20851
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7333708070335167e-06,
+      "loss": 0.9075,
+      "step": 20852
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7328564956308803e-06,
+      "loss": 0.9593,
+      "step": 20853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7323423577745433e-06,
+      "loss": 1.0012,
+      "step": 20854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73182839347414e-06,
+      "loss": 1.0156,
+      "step": 20855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.73131460273929e-06,
+      "loss": 0.9562,
+      "step": 20856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.730800985579618e-06,
+      "loss": 0.9521,
+      "step": 20857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7302875420047415e-06,
+      "loss": 0.9161,
+      "step": 20858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7297742720242763e-06,
+      "loss": 0.9428,
+      "step": 20859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.729261175647834e-06,
+      "loss": 0.9243,
+      "step": 20860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7287482528850265e-06,
+      "loss": 0.992,
+      "step": 20861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7282355037454535e-06,
+      "loss": 0.9162,
+      "step": 20862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.727722928238721e-06,
+      "loss": 0.9255,
+      "step": 20863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7272105263744264e-06,
+      "loss": 0.9682,
+      "step": 20864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.726698298162167e-06,
+      "loss": 0.9864,
+      "step": 20865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.726186243611536e-06,
+      "loss": 0.9672,
+      "step": 20866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7256743627321193e-06,
+      "loss": 0.9278,
+      "step": 20867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.725162655533508e-06,
+      "loss": 0.9618,
+      "step": 20868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7246511220252796e-06,
+      "loss": 0.9339,
+      "step": 20869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7241397622170153e-06,
+      "loss": 0.9274,
+      "step": 20870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.723628576118295e-06,
+      "loss": 0.9295,
+      "step": 20871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7231175637386844e-06,
+      "loss": 0.9417,
+      "step": 20872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.722606725087761e-06,
+      "loss": 1.0007,
+      "step": 20873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7220960601750845e-06,
+      "loss": 0.943,
+      "step": 20874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7215855690102245e-06,
+      "loss": 0.908,
+      "step": 20875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7210752516027374e-06,
+      "loss": 0.9696,
+      "step": 20876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.72056510796218e-06,
+      "loss": 0.8919,
+      "step": 20877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7200551380981056e-06,
+      "loss": 0.9496,
+      "step": 20878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7195453420200657e-06,
+      "loss": 0.931,
+      "step": 20879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7190357197376084e-06,
+      "loss": 0.9828,
+      "step": 20880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7185262712602746e-06,
+      "loss": 0.9542,
+      "step": 20881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7180169965976058e-06,
+      "loss": 0.9213,
+      "step": 20882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7175078957591386e-06,
+      "loss": 0.9341,
+      "step": 20883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7169989687544074e-06,
+      "loss": 0.8935,
+      "step": 20884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.716490215592944e-06,
+      "loss": 0.9046,
+      "step": 20885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7159816362842763e-06,
+      "loss": 0.9554,
+      "step": 20886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7154732308379263e-06,
+      "loss": 0.9207,
+      "step": 20887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7149649992634154e-06,
+      "loss": 0.9095,
+      "step": 20888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.714456941570262e-06,
+      "loss": 0.9512,
+      "step": 20889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7139490577679797e-06,
+      "loss": 0.9552,
+      "step": 20890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7134413478660818e-06,
+      "loss": 0.9017,
+      "step": 20891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7129338118740713e-06,
+      "loss": 0.978,
+      "step": 20892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.71242644980146e-06,
+      "loss": 0.9698,
+      "step": 20893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.711919261657744e-06,
+      "loss": 0.9242,
+      "step": 20894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.711412247452422e-06,
+      "loss": 0.92,
+      "step": 20895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7109054071949904e-06,
+      "loss": 0.9936,
+      "step": 20896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7103987408949407e-06,
+      "loss": 0.8808,
+      "step": 20897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.709892248561762e-06,
+      "loss": 0.9082,
+      "step": 20898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7093859302049365e-06,
+      "loss": 0.9728,
+      "step": 20899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.708879785833948e-06,
+      "loss": 0.9441,
+      "step": 20900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.708373815458275e-06,
+      "loss": 0.8832,
+      "step": 20901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.707868019087392e-06,
+      "loss": 0.9947,
+      "step": 20902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7073623967307727e-06,
+      "loss": 0.8681,
+      "step": 20903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7068569483978857e-06,
+      "loss": 1.0079,
+      "step": 20904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7063516740981934e-06,
+      "loss": 0.9962,
+      "step": 20905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7058465738411633e-06,
+      "loss": 0.9354,
+      "step": 20906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7053416476362506e-06,
+      "loss": 1.0075,
+      "step": 20907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.704836895492912e-06,
+      "loss": 0.8772,
+      "step": 20908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7043323174206017e-06,
+      "loss": 0.9786,
+      "step": 20909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.703827913428764e-06,
+      "loss": 0.9113,
+      "step": 20910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.703323683526852e-06,
+      "loss": 0.9602,
+      "step": 20911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7028196277243016e-06,
+      "loss": 1.031,
+      "step": 20912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7023157460305585e-06,
+      "loss": 0.989,
+      "step": 20913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.701812038455054e-06,
+      "loss": 0.9511,
+      "step": 20914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.701308505007223e-06,
+      "loss": 0.9176,
+      "step": 20915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.7008051456964957e-06,
+      "loss": 0.9116,
+      "step": 20916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.700301960532298e-06,
+      "loss": 0.9703,
+      "step": 20917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.699798949524054e-06,
+      "loss": 0.9388,
+      "step": 20918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6992961126811814e-06,
+      "loss": 0.9037,
+      "step": 20919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6987934500130977e-06,
+      "loss": 0.9218,
+      "step": 20920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6982909615292175e-06,
+      "loss": 0.9681,
+      "step": 20921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6977886472389495e-06,
+      "loss": 0.9387,
+      "step": 20922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6972865071517018e-06,
+      "loss": 1.0013,
+      "step": 20923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6967845412768793e-06,
+      "loss": 0.9803,
+      "step": 20924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.696282749623878e-06,
+      "loss": 0.9069,
+      "step": 20925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.695781132202098e-06,
+      "loss": 0.9525,
+      "step": 20926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6952796890209325e-06,
+      "loss": 1.0212,
+      "step": 20927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.694778420089772e-06,
+      "loss": 0.9202,
+      "step": 20928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6942773254180057e-06,
+      "loss": 0.9586,
+      "step": 20929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6937764050150118e-06,
+      "loss": 0.9242,
+      "step": 20930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.69327565889018e-06,
+      "loss": 0.8907,
+      "step": 20931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6927750870528784e-06,
+      "loss": 0.9237,
+      "step": 20932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.692274689512489e-06,
+      "loss": 0.9835,
+      "step": 20933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6917744662783788e-06,
+      "loss": 0.9277,
+      "step": 20934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.691274417359915e-06,
+      "loss": 0.9719,
+      "step": 20935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6907745427664656e-06,
+      "loss": 0.9421,
+      "step": 20936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6902748425073858e-06,
+      "loss": 0.9045,
+      "step": 20937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6897753165920403e-06,
+      "loss": 0.977,
+      "step": 20938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6892759650297798e-06,
+      "loss": 1.0094,
+      "step": 20939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6887767878299565e-06,
+      "loss": 0.9855,
+      "step": 20940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.688277785001917e-06,
+      "loss": 0.9743,
+      "step": 20941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6877789565550097e-06,
+      "loss": 0.943,
+      "step": 20942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.687280302498573e-06,
+      "loss": 0.9624,
+      "step": 20943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6867818228419475e-06,
+      "loss": 0.9451,
+      "step": 20944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.686283517594467e-06,
+      "loss": 0.9312,
+      "step": 20945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.6857853867654627e-06,
+      "loss": 0.9223,
+      "step": 20946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.685287430364263e-06,
+      "loss": 0.9865,
+      "step": 20947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.684789648400195e-06,
+      "loss": 0.8612,
+      "step": 20948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6842920408825814e-06,
+      "loss": 0.9166,
+      "step": 20949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6837946078207356e-06,
+      "loss": 1.0031,
+      "step": 20950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.683297349223979e-06,
+      "loss": 0.9206,
+      "step": 20951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.682800265101622e-06,
+      "loss": 0.8962,
+      "step": 20952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.682303355462971e-06,
+      "loss": 0.9842,
+      "step": 20953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6818066203173343e-06,
+      "loss": 0.9705,
+      "step": 20954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.681310059674014e-06,
+      "loss": 0.9776,
+      "step": 20955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6808136735423096e-06,
+      "loss": 0.9516,
+      "step": 20956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6803174619315153e-06,
+      "loss": 0.9271,
+      "step": 20957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.679821424850925e-06,
+      "loss": 0.9345,
+      "step": 20958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.679325562309826e-06,
+      "loss": 0.963,
+      "step": 20959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.678829874317507e-06,
+      "loss": 0.9673,
+      "step": 20960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6783343608832503e-06,
+      "loss": 0.9239,
+      "step": 20961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6778390220163353e-06,
+      "loss": 0.8842,
+      "step": 20962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.677343857726036e-06,
+      "loss": 1.0081,
+      "step": 20963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.676848868021628e-06,
+      "loss": 0.9585,
+      "step": 20964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.67635405291238e-06,
+      "loss": 0.9436,
+      "step": 20965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6758594124075578e-06,
+      "loss": 0.9579,
+      "step": 20966
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.675364946516428e-06,
+      "loss": 0.9578,
+      "step": 20967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.674870655248244e-06,
+      "loss": 0.9477,
+      "step": 20968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.674376538612269e-06,
+      "loss": 0.9509,
+      "step": 20969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6738825966177508e-06,
+      "loss": 0.934,
+      "step": 20970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6733888292739454e-06,
+      "loss": 0.8986,
+      "step": 20971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.672895236590095e-06,
+      "loss": 0.9709,
+      "step": 20972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.672401818575444e-06,
+      "loss": 0.9307,
+      "step": 20973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.671908575239236e-06,
+      "loss": 0.9889,
+      "step": 20974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6714155065907014e-06,
+      "loss": 0.9295,
+      "step": 20975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.670922612639081e-06,
+      "loss": 0.9413,
+      "step": 20976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6704298933936003e-06,
+      "loss": 0.958,
+      "step": 20977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6699373488634883e-06,
+      "loss": 0.9281,
+      "step": 20978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6694449790579684e-06,
+      "loss": 0.9685,
+      "step": 20979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.668952783986262e-06,
+      "loss": 0.9453,
+      "step": 20980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.668460763657587e-06,
+      "loss": 0.9348,
+      "step": 20981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6679689180811566e-06,
+      "loss": 0.9034,
+      "step": 20982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6674772472661806e-06,
+      "loss": 0.9701,
+      "step": 20983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.666985751221868e-06,
+      "loss": 0.9486,
+      "step": 20984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6664944299574227e-06,
+      "loss": 0.9522,
+      "step": 20985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.666003283482045e-06,
+      "loss": 0.9164,
+      "step": 20986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6655123118049354e-06,
+      "loss": 0.9962,
+      "step": 20987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.665021514935282e-06,
+      "loss": 0.9195,
+      "step": 20988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6645308928822847e-06,
+      "loss": 0.947,
+      "step": 20989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6640404456551252e-06,
+      "loss": 0.921,
+      "step": 20990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.663550173262989e-06,
+      "loss": 0.9032,
+      "step": 20991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6630600757150585e-06,
+      "loss": 0.9236,
+      "step": 20992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.662570153020512e-06,
+      "loss": 0.9078,
+      "step": 20993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6620804051885252e-06,
+      "loss": 0.9432,
+      "step": 20994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.661590832228266e-06,
+      "loss": 0.9237,
+      "step": 20995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6611014341489073e-06,
+      "loss": 0.9088,
+      "step": 20996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.66061221095961e-06,
+      "loss": 0.9482,
+      "step": 20997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.660123162669538e-06,
+      "loss": 0.951,
+      "step": 20998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.659634289287849e-06,
+      "loss": 1.0347,
+      "step": 20999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6591455908236985e-06,
+      "loss": 0.8733,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "eval_text_loss": 0.9264777898788452,
+      "eval_text_runtime": 1.5751,
+      "eval_text_samples_per_second": 46.981,
+      "eval_text_steps_per_second": 1.27,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "eval_svit_loss": 1.0736202001571655,
+      "eval_svit_runtime": 4.538,
+      "eval_svit_samples_per_second": 150.067,
+      "eval_svit_steps_per_second": 2.424,
+      "step": 21000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6586570672862402e-06,
+      "loss": 0.9281,
+      "step": 21001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6581687186846195e-06,
+      "loss": 0.9507,
+      "step": 21002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.657680545027982e-06,
+      "loss": 0.9342,
+      "step": 21003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6571925463254724e-06,
+      "loss": 0.9597,
+      "step": 21004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6567047225862267e-06,
+      "loss": 0.9511,
+      "step": 21005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6562170738193816e-06,
+      "loss": 0.9322,
+      "step": 21006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6557296000340715e-06,
+      "loss": 0.951,
+      "step": 21007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6552423012394202e-06,
+      "loss": 0.9308,
+      "step": 21008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6547551774445592e-06,
+      "loss": 0.9886,
+      "step": 21009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6542682286586068e-06,
+      "loss": 0.9713,
+      "step": 21010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.653781454890683e-06,
+      "loss": 0.9616,
+      "step": 21011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.653294856149905e-06,
+      "loss": 0.9957,
+      "step": 21012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.652808432445383e-06,
+      "loss": 0.9817,
+      "step": 21013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6523221837862292e-06,
+      "loss": 0.9789,
+      "step": 21014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.651836110181546e-06,
+      "loss": 0.9812,
+      "step": 21015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.65135021164044e-06,
+      "loss": 0.9293,
+      "step": 21016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.650864488172008e-06,
+      "loss": 0.9308,
+      "step": 21017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6503789397853478e-06,
+      "loss": 0.9765,
+      "step": 21018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.649893566489551e-06,
+      "loss": 0.9414,
+      "step": 21019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.649408368293707e-06,
+      "loss": 0.9324,
+      "step": 21020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.648923345206905e-06,
+      "loss": 0.9715,
+      "step": 21021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.648438497238225e-06,
+      "loss": 0.8812,
+      "step": 21022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6479538243967465e-06,
+      "loss": 0.9783,
+      "step": 21023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6474693266915477e-06,
+      "loss": 0.9597,
+      "step": 21024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6469850041317027e-06,
+      "loss": 0.9915,
+      "step": 21025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6465008567262775e-06,
+      "loss": 0.9354,
+      "step": 21026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.646016884484344e-06,
+      "loss": 0.9432,
+      "step": 21027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.645533087414961e-06,
+      "loss": 1.0268,
+      "step": 21028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6450494655271898e-06,
+      "loss": 0.9158,
+      "step": 21029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6445660188300904e-06,
+      "loss": 0.9388,
+      "step": 21030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6440827473327116e-06,
+      "loss": 0.9545,
+      "step": 21031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6435996510441076e-06,
+      "loss": 0.8812,
+      "step": 21032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6431167299733206e-06,
+      "loss": 0.9576,
+      "step": 21033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6426339841294003e-06,
+      "loss": 0.9822,
+      "step": 21034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.642151413521383e-06,
+      "loss": 0.9704,
+      "step": 21035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.641669018158307e-06,
+      "loss": 0.9451,
+      "step": 21036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.641186798049206e-06,
+      "loss": 0.8949,
+      "step": 21037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6407047532031106e-06,
+      "loss": 0.9368,
+      "step": 21038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.640222883629049e-06,
+      "loss": 0.8813,
+      "step": 21039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.639741189336044e-06,
+      "loss": 0.9171,
+      "step": 21040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6392596703331155e-06,
+      "loss": 0.9231,
+      "step": 21041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.638778326629283e-06,
+      "loss": 0.9835,
+      "step": 21042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6382971582335595e-06,
+      "loss": 0.9358,
+      "step": 21043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.637816165154955e-06,
+      "loss": 0.9854,
+      "step": 21044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.637335347402481e-06,
+      "loss": 0.9649,
+      "step": 21045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.636854704985136e-06,
+      "loss": 0.9383,
+      "step": 21046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.636374237911926e-06,
+      "loss": 0.9669,
+      "step": 21047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6358939461918465e-06,
+      "loss": 0.9664,
+      "step": 21048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6354138298338907e-06,
+      "loss": 0.9503,
+      "step": 21049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6349338888470533e-06,
+      "loss": 0.9194,
+      "step": 21050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6344541232403177e-06,
+      "loss": 0.9251,
+      "step": 21051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.633974533022674e-06,
+      "loss": 0.9927,
+      "step": 21052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6334951182030966e-06,
+      "loss": 0.8971,
+      "step": 21053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6330158787905704e-06,
+      "loss": 0.931,
+      "step": 21054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.632536814794067e-06,
+      "loss": 0.9966,
+      "step": 21055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.632057926222557e-06,
+      "loss": 0.9488,
+      "step": 21056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6315792130850106e-06,
+      "loss": 0.9892,
+      "step": 21057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.631100675390388e-06,
+      "loss": 0.9832,
+      "step": 21058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6306223131476584e-06,
+      "loss": 0.9773,
+      "step": 21059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.630144126365775e-06,
+      "loss": 0.9476,
+      "step": 21060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.629666115053693e-06,
+      "loss": 0.9589,
+      "step": 21061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6291882792203663e-06,
+      "loss": 0.9228,
+      "step": 21062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6287106188747416e-06,
+      "loss": 0.9411,
+      "step": 21063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.628233134025764e-06,
+      "loss": 0.9807,
+      "step": 21064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6277558246823777e-06,
+      "loss": 0.9884,
+      "step": 21065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.627278690853519e-06,
+      "loss": 0.9162,
+      "step": 21066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6268017325481234e-06,
+      "loss": 0.9227,
+      "step": 21067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.626324949775123e-06,
+      "loss": 0.963,
+      "step": 21068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6258483425434467e-06,
+      "loss": 0.9227,
+      "step": 21069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6253719108620223e-06,
+      "loss": 0.9291,
+      "step": 21070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6248956547397666e-06,
+      "loss": 0.9322,
+      "step": 21071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.624419574185605e-06,
+      "loss": 0.895,
+      "step": 21072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6239436692084475e-06,
+      "loss": 0.8659,
+      "step": 21073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.623467939817209e-06,
+      "loss": 1.0251,
+      "step": 21074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6229923860207987e-06,
+      "loss": 0.9315,
+      "step": 21075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6225170078281207e-06,
+      "loss": 0.9202,
+      "step": 21076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.622041805248081e-06,
+      "loss": 0.9449,
+      "step": 21077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6215667782895726e-06,
+      "loss": 0.9918,
+      "step": 21078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6210919269614988e-06,
+      "loss": 0.9075,
+      "step": 21079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.620617251272746e-06,
+      "loss": 0.9078,
+      "step": 21080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6201427512322068e-06,
+      "loss": 0.8761,
+      "step": 21081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6196684268487652e-06,
+      "loss": 0.947,
+      "step": 21082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6191942781313054e-06,
+      "loss": 0.8863,
+      "step": 21083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.618720305088708e-06,
+      "loss": 0.9671,
+      "step": 21084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.618246507729846e-06,
+      "loss": 0.9358,
+      "step": 21085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6177728860635942e-06,
+      "loss": 0.9508,
+      "step": 21086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.617299440098821e-06,
+      "loss": 0.8888,
+      "step": 21087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.616826169844393e-06,
+      "loss": 0.9605,
+      "step": 21088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.616353075309174e-06,
+      "loss": 0.9685,
+      "step": 21089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6158801565020244e-06,
+      "loss": 0.9151,
+      "step": 21090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6154074134317967e-06,
+      "loss": 0.9208,
+      "step": 21091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.61493484610735e-06,
+      "loss": 0.9423,
+      "step": 21092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6144624545375276e-06,
+      "loss": 0.9594,
+      "step": 21093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6139902387311803e-06,
+      "loss": 0.8825,
+      "step": 21094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6135181986971515e-06,
+      "loss": 0.9455,
+      "step": 21095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.613046334444276e-06,
+      "loss": 0.908,
+      "step": 21096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6125746459813986e-06,
+      "loss": 0.9719,
+      "step": 21097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6121031333173453e-06,
+      "loss": 0.9247,
+      "step": 21098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6116317964609497e-06,
+      "loss": 0.9082,
+      "step": 21099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6111606354210373e-06,
+      "loss": 0.9282,
+      "step": 21100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6106896502064327e-06,
+      "loss": 0.9384,
+      "step": 21101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6102188408259543e-06,
+      "loss": 0.9841,
+      "step": 21102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.609748207288423e-06,
+      "loss": 0.9532,
+      "step": 21103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.609277749602648e-06,
+      "loss": 0.9255,
+      "step": 21104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.608807467777441e-06,
+      "loss": 0.8862,
+      "step": 21105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6083373618216092e-06,
+      "loss": 0.9232,
+      "step": 21106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.607867431743956e-06,
+      "loss": 0.8631,
+      "step": 21107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.607397677553283e-06,
+      "loss": 0.9588,
+      "step": 21108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6069280992583835e-06,
+      "loss": 1.0209,
+      "step": 21109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6064586968680585e-06,
+      "loss": 0.9496,
+      "step": 21110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6059894703910915e-06,
+      "loss": 0.9425,
+      "step": 21111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.605520419836273e-06,
+      "loss": 1.0176,
+      "step": 21112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6050515452123848e-06,
+      "loss": 0.9373,
+      "step": 21113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.604582846528211e-06,
+      "loss": 0.9581,
+      "step": 21114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6041143237925274e-06,
+      "loss": 0.9638,
+      "step": 21115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.603645977014106e-06,
+      "loss": 0.8724,
+      "step": 21116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.603177806201721e-06,
+      "loss": 0.9434,
+      "step": 21117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.602709811364137e-06,
+      "loss": 0.9482,
+      "step": 21118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6022419925101198e-06,
+      "loss": 0.8758,
+      "step": 21119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6017743496484285e-06,
+      "loss": 0.9245,
+      "step": 21120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.601306882787823e-06,
+      "loss": 0.9584,
+      "step": 21121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.6008395919370568e-06,
+      "loss": 0.9573,
+      "step": 21122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.60037247710488e-06,
+      "loss": 0.9504,
+      "step": 21123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5999055383000418e-06,
+      "loss": 0.9546,
+      "step": 21124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5994387755312834e-06,
+      "loss": 0.9748,
+      "step": 21125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.59897218880735e-06,
+      "loss": 1.0103,
+      "step": 21126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5985057781369773e-06,
+      "loss": 0.9053,
+      "step": 21127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.598039543528901e-06,
+      "loss": 0.956,
+      "step": 21128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.597573484991848e-06,
+      "loss": 0.8929,
+      "step": 21129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5971076025345535e-06,
+      "loss": 0.9475,
+      "step": 21130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5966418961657357e-06,
+      "loss": 0.943,
+      "step": 21131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.596176365894119e-06,
+      "loss": 1.0207,
+      "step": 21132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5957110117284227e-06,
+      "loss": 0.9598,
+      "step": 21133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.595245833677356e-06,
+      "loss": 0.9625,
+      "step": 21134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.594780831749636e-06,
+      "loss": 0.9606,
+      "step": 21135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5943160059539664e-06,
+      "loss": 0.908,
+      "step": 21136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5938513562990577e-06,
+      "loss": 0.927,
+      "step": 21137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.593386882793606e-06,
+      "loss": 0.9956,
+      "step": 21138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5929225854463125e-06,
+      "loss": 0.9009,
+      "step": 21139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5924584642658704e-06,
+      "loss": 0.9082,
+      "step": 21140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.591994519260972e-06,
+      "loss": 0.9884,
+      "step": 21141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5915307504403077e-06,
+      "loss": 0.969,
+      "step": 21142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5910671578125594e-06,
+      "loss": 0.9913,
+      "step": 21143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.59060374138641e-06,
+      "loss": 0.9203,
+      "step": 21144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5901405011705364e-06,
+      "loss": 0.9093,
+      "step": 21145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.589677437173617e-06,
+      "loss": 1.0321,
+      "step": 21146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.589214549404321e-06,
+      "loss": 1.0091,
+      "step": 21147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5887518378713184e-06,
+      "loss": 0.947,
+      "step": 21148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5882893025832735e-06,
+      "loss": 0.9127,
+      "step": 21149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5878269435488494e-06,
+      "loss": 0.968,
+      "step": 21150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5873647607767017e-06,
+      "loss": 0.925,
+      "step": 21151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.586902754275489e-06,
+      "loss": 0.8823,
+      "step": 21152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.586440924053864e-06,
+      "loss": 0.9479,
+      "step": 21153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.58597927012047e-06,
+      "loss": 0.9058,
+      "step": 21154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5855177924839593e-06,
+      "loss": 0.9661,
+      "step": 21155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5850564911529677e-06,
+      "loss": 1.0006,
+      "step": 21156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5845953661361403e-06,
+      "loss": 0.9588,
+      "step": 21157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.584134417442108e-06,
+      "loss": 0.9406,
+      "step": 21158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5836736450795046e-06,
+      "loss": 0.9455,
+      "step": 21159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5832130490569586e-06,
+      "loss": 0.9549,
+      "step": 21160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5827526293830953e-06,
+      "loss": 0.9311,
+      "step": 21161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5822923860665386e-06,
+      "loss": 0.9602,
+      "step": 21162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5818323191159046e-06,
+      "loss": 0.9521,
+      "step": 21163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5813724285398123e-06,
+      "loss": 0.9678,
+      "step": 21164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5809127143468717e-06,
+      "loss": 0.9598,
+      "step": 21165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5804531765456956e-06,
+      "loss": 0.966,
+      "step": 21166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5799938151448824e-06,
+      "loss": 0.9681,
+      "step": 21167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5795346301530424e-06,
+      "loss": 0.9445,
+      "step": 21168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.579075621578771e-06,
+      "loss": 0.8814,
+      "step": 21169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5786167894306637e-06,
+      "loss": 0.8746,
+      "step": 21170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5781581337173156e-06,
+      "loss": 0.9806,
+      "step": 21171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5776996544473123e-06,
+      "loss": 0.9327,
+      "step": 21172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.577241351629244e-06,
+      "loss": 0.9643,
+      "step": 21173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5767832252716884e-06,
+      "loss": 0.9384,
+      "step": 21174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5763252753832313e-06,
+      "loss": 0.9674,
+      "step": 21175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5758675019724425e-06,
+      "loss": 0.9748,
+      "step": 21176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5754099050478978e-06,
+      "loss": 0.9615,
+      "step": 21177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5749524846181663e-06,
+      "loss": 0.9159,
+      "step": 21178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5744952406918148e-06,
+      "loss": 0.9795,
+      "step": 21179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5740381732774055e-06,
+      "loss": 0.9155,
+      "step": 21180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5735812823834974e-06,
+      "loss": 0.9593,
+      "step": 21181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.5731245680186467e-06,
+      "loss": 0.9684,
+      "step": 21182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.572668030191407e-06,
+      "loss": 1.0205,
+      "step": 21183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5722116689103276e-06,
+      "loss": 0.9634,
+      "step": 21184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5717554841839544e-06,
+      "loss": 0.9863,
+      "step": 21185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5712994760208324e-06,
+      "loss": 1.0255,
+      "step": 21186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5708436444294986e-06,
+      "loss": 0.9283,
+      "step": 21187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5703879894184905e-06,
+      "loss": 0.8896,
+      "step": 21188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5699325109963404e-06,
+      "loss": 0.8996,
+      "step": 21189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5694772091715796e-06,
+      "loss": 0.9638,
+      "step": 21190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.569022083952735e-06,
+      "loss": 0.9195,
+      "step": 21191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5685671353483255e-06,
+      "loss": 0.9359,
+      "step": 21192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.568112363366878e-06,
+      "loss": 0.9638,
+      "step": 21193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5676577680169014e-06,
+      "loss": 0.9644,
+      "step": 21194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5672033493069164e-06,
+      "loss": 0.9181,
+      "step": 21195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5667491072454255e-06,
+      "loss": 0.9597,
+      "step": 21196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5662950418409407e-06,
+      "loss": 0.9724,
+      "step": 21197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5658411531019647e-06,
+      "loss": 0.903,
+      "step": 21198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565387441036993e-06,
+      "loss": 1.0219,
+      "step": 21199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5649339056545297e-06,
+      "loss": 0.9024,
+      "step": 21200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.564480546963062e-06,
+      "loss": 0.9205,
+      "step": 21201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5640273649710814e-06,
+      "loss": 0.9405,
+      "step": 21202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5635743596870756e-06,
+      "loss": 0.9085,
+      "step": 21203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5631215311195294e-06,
+      "loss": 0.9465,
+      "step": 21204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5626688792769195e-06,
+      "loss": 1.0392,
+      "step": 21205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.562216404167727e-06,
+      "loss": 0.9499,
+      "step": 21206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5617641058004206e-06,
+      "loss": 0.997,
+      "step": 21207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5613119841834735e-06,
+      "loss": 0.9125,
+      "step": 21208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.560860039325353e-06,
+      "loss": 1.0087,
+      "step": 21209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5604082712345203e-06,
+      "loss": 0.9498,
+      "step": 21210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5599566799194387e-06,
+      "loss": 0.9679,
+      "step": 21211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5595052653885613e-06,
+      "loss": 0.9673,
+      "step": 21212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5590540276503476e-06,
+      "loss": 0.9514,
+      "step": 21213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.558602966713242e-06,
+      "loss": 0.973,
+      "step": 21214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5581520825856936e-06,
+      "loss": 0.9131,
+      "step": 21215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5577013752761485e-06,
+      "loss": 0.9162,
+      "step": 21216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5572508447930423e-06,
+      "loss": 0.9771,
+      "step": 21217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5568004911448177e-06,
+      "loss": 0.9628,
+      "step": 21218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.556350314339903e-06,
+      "loss": 0.9224,
+      "step": 21219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5559003143867345e-06,
+      "loss": 0.9513,
+      "step": 21220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555450491293734e-06,
+      "loss": 0.9718,
+      "step": 21221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.555000845069329e-06,
+      "loss": 0.9334,
+      "step": 21222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5545513757219367e-06,
+      "loss": 0.9511,
+      "step": 21223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.554102083259977e-06,
+      "loss": 0.9739,
+      "step": 21224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5536529676918654e-06,
+      "loss": 0.9344,
+      "step": 21225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5532040290260086e-06,
+      "loss": 0.946,
+      "step": 21226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.552755267270816e-06,
+      "loss": 0.9472,
+      "step": 21227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55230668243469e-06,
+      "loss": 0.9538,
+      "step": 21228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5518582745260347e-06,
+      "loss": 0.9079,
+      "step": 21229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.551410043553242e-06,
+      "loss": 0.9565,
+      "step": 21230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550961989524712e-06,
+      "loss": 0.9353,
+      "step": 21231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55051411244883e-06,
+      "loss": 1.0456,
+      "step": 21232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.550066412333988e-06,
+      "loss": 0.9543,
+      "step": 21233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5496188891885677e-06,
+      "loss": 0.9648,
+      "step": 21234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5491715430209495e-06,
+      "loss": 0.9625,
+      "step": 21235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5487243738395143e-06,
+      "loss": 0.9325,
+      "step": 21236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.54827738165263e-06,
+      "loss": 0.9265,
+      "step": 21237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.547830566468673e-06,
+      "loss": 0.8983,
+      "step": 21238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.547383928296008e-06,
+      "loss": 0.9385,
+      "step": 21239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5469374671430005e-06,
+      "loss": 0.9039,
+      "step": 21240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.546491183018011e-06,
+      "loss": 0.9211,
+      "step": 21241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5460450759293967e-06,
+      "loss": 0.9355,
+      "step": 21242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.545599145885515e-06,
+      "loss": 0.9266,
+      "step": 21243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5451533928947104e-06,
+      "loss": 0.9437,
+      "step": 21244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5447078169653363e-06,
+      "loss": 1.0175,
+      "step": 21245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5442624181057345e-06,
+      "loss": 0.9309,
+      "step": 21246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5438171963242453e-06,
+      "loss": 0.9492,
+      "step": 21247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5433721516292086e-06,
+      "loss": 0.9082,
+      "step": 21248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.542927284028959e-06,
+      "loss": 0.9091,
+      "step": 21249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5424825935318233e-06,
+      "loss": 0.9409,
+      "step": 21250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5420380801461357e-06,
+      "loss": 0.9863,
+      "step": 21251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5415937438802153e-06,
+      "loss": 0.8977,
+      "step": 21252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.541149584742385e-06,
+      "loss": 0.9639,
+      "step": 21253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5407056027409652e-06,
+      "loss": 0.9504,
+      "step": 21254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5402617978842637e-06,
+      "loss": 0.9569,
+      "step": 21255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539818170180599e-06,
+      "loss": 0.9129,
+      "step": 21256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5393747196382737e-06,
+      "loss": 0.9621,
+      "step": 21257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.538931446265596e-06,
+      "loss": 0.9446,
+      "step": 21258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5384883500708647e-06,
+      "loss": 0.9632,
+      "step": 21259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5380454310623783e-06,
+      "loss": 0.9073,
+      "step": 21260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.537602689248432e-06,
+      "loss": 0.9347,
+      "step": 21261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5371601246373167e-06,
+      "loss": 0.953,
+      "step": 21262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.536717737237321e-06,
+      "loss": 0.9279,
+      "step": 21263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5362755270567275e-06,
+      "loss": 0.9634,
+      "step": 21264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.535833494103819e-06,
+      "loss": 0.9496,
+      "step": 21265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.535391638386873e-06,
+      "loss": 0.9126,
+      "step": 21266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534949959914165e-06,
+      "loss": 0.959,
+      "step": 21267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5345084586939652e-06,
+      "loss": 0.94,
+      "step": 21268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534067134734544e-06,
+      "loss": 0.913,
+      "step": 21269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5336259880441618e-06,
+      "loss": 0.9881,
+      "step": 21270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5331850186310826e-06,
+      "loss": 0.9542,
+      "step": 21271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5327442265035654e-06,
+      "loss": 0.9784,
+      "step": 21272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5323036116698633e-06,
+      "loss": 0.925,
+      "step": 21273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5318631741382285e-06,
+      "loss": 0.9269,
+      "step": 21274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.531422913916907e-06,
+      "loss": 0.9259,
+      "step": 21275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5309828310141485e-06,
+      "loss": 0.9092,
+      "step": 21276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5305429254381874e-06,
+      "loss": 0.9608,
+      "step": 21277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530103197197269e-06,
+      "loss": 0.9024,
+      "step": 21278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.529663646299624e-06,
+      "loss": 0.9355,
+      "step": 21279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5292242727534843e-06,
+      "loss": 0.9474,
+      "step": 21280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.528785076567081e-06,
+      "loss": 0.9512,
+      "step": 21281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5283460577486332e-06,
+      "loss": 1.0106,
+      "step": 21282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.527907216306368e-06,
+      "loss": 0.9058,
+      "step": 21283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5274685522485e-06,
+      "loss": 0.914,
+      "step": 21284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.527030065583246e-06,
+      "loss": 0.9646,
+      "step": 21285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526591756318817e-06,
+      "loss": 0.9735,
+      "step": 21286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.526153624463421e-06,
+      "loss": 0.9378,
+      "step": 21287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.525715670025263e-06,
+      "loss": 0.9083,
+      "step": 21288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5252778930125466e-06,
+      "loss": 0.9221,
+      "step": 21289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.524840293433466e-06,
+      "loss": 0.9377,
+      "step": 21290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.524402871296219e-06,
+      "loss": 0.9921,
+      "step": 21291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5239656266089963e-06,
+      "loss": 0.9337,
+      "step": 21292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5235285593799875e-06,
+      "loss": 0.9936,
+      "step": 21293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523091669617377e-06,
+      "loss": 0.9136,
+      "step": 21294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5226549573293438e-06,
+      "loss": 0.953,
+      "step": 21295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.522218422524071e-06,
+      "loss": 0.9111,
+      "step": 21296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5217820652097294e-06,
+      "loss": 0.9219,
+      "step": 21297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.521345885394493e-06,
+      "loss": 0.9374,
+      "step": 21298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520909883086529e-06,
+      "loss": 0.9742,
+      "step": 21299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520474058294004e-06,
+      "loss": 0.9128,
+      "step": 21300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.520038411025079e-06,
+      "loss": 0.9334,
+      "step": 21301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519602941287911e-06,
+      "loss": 0.9195,
+      "step": 21302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.519167649090656e-06,
+      "loss": 0.9247,
+      "step": 21303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.518732534441466e-06,
+      "loss": 0.9182,
+      "step": 21304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5182975973484903e-06,
+      "loss": 0.922,
+      "step": 21305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5178628378198722e-06,
+      "loss": 0.8634,
+      "step": 21306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5174282558637554e-06,
+      "loss": 0.9365,
+      "step": 21307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5169938514882764e-06,
+      "loss": 0.9859,
+      "step": 21308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5165596247015714e-06,
+      "loss": 0.9326,
+      "step": 21309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.516125575511771e-06,
+      "loss": 0.9525,
+      "step": 21310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5156917039270047e-06,
+      "loss": 0.9906,
+      "step": 21311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5152580099553993e-06,
+      "loss": 0.9407,
+      "step": 21312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5148244936050715e-06,
+      "loss": 0.8912,
+      "step": 21313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5143911548841473e-06,
+      "loss": 0.9253,
+      "step": 21314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5139579938007332e-06,
+      "loss": 0.9496,
+      "step": 21315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5135250103629494e-06,
+      "loss": 0.9124,
+      "step": 21316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5130922045788985e-06,
+      "loss": 0.959,
+      "step": 21317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5126595764566876e-06,
+      "loss": 0.9226,
+      "step": 21318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.512227126004421e-06,
+      "loss": 0.9049,
+      "step": 21319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5117948532301906e-06,
+      "loss": 0.9963,
+      "step": 21320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5113627581421e-06,
+      "loss": 0.9142,
+      "step": 21321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510930840748235e-06,
+      "loss": 0.8542,
+      "step": 21322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.510499101056686e-06,
+      "loss": 0.9835,
+      "step": 21323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5100675390755384e-06,
+      "loss": 0.9503,
+      "step": 21324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.509636154812875e-06,
+      "loss": 0.9497,
+      "step": 21325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.509204948276771e-06,
+      "loss": 0.9562,
+      "step": 21326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.508773919475307e-06,
+      "loss": 0.9705,
+      "step": 21327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50834306841655e-06,
+      "loss": 0.9132,
+      "step": 21328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5079123951085705e-06,
+      "loss": 1.0009,
+      "step": 21329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507481899559434e-06,
+      "loss": 0.9666,
+      "step": 21330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5070515817772015e-06,
+      "loss": 0.9552,
+      "step": 21331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5066214417699335e-06,
+      "loss": 0.9105,
+      "step": 21332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5061914795456814e-06,
+      "loss": 0.9532,
+      "step": 21333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.505761695112503e-06,
+      "loss": 0.9486,
+      "step": 21334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.505332088478442e-06,
+      "loss": 0.9589,
+      "step": 21335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5049026596515436e-06,
+      "loss": 0.8996,
+      "step": 21336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5044734086398525e-06,
+      "loss": 0.8821,
+      "step": 21337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.504044335451407e-06,
+      "loss": 0.9789,
+      "step": 21338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5036154400942415e-06,
+      "loss": 0.8896,
+      "step": 21339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5031867225763864e-06,
+      "loss": 0.9511,
+      "step": 21340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5027581829058735e-06,
+      "loss": 0.9713,
+      "step": 21341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5023298210907255e-06,
+      "loss": 0.9797,
+      "step": 21342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5019016371389664e-06,
+      "loss": 0.9853,
+      "step": 21343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.501473631058612e-06,
+      "loss": 0.9257,
+      "step": 21344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5010458028576803e-06,
+      "loss": 0.9934,
+      "step": 21345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5006181525441844e-06,
+      "loss": 0.9905,
+      "step": 21346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5001906801261277e-06,
+      "loss": 0.9883,
+      "step": 21347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4997633856115203e-06,
+      "loss": 0.9481,
+      "step": 21348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.499336269008361e-06,
+      "loss": 0.9358,
+      "step": 21349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4989093303246517e-06,
+      "loss": 0.9394,
+      "step": 21350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4984825695683846e-06,
+      "loss": 0.9998,
+      "step": 21351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4980559867475553e-06,
+      "loss": 1.0035,
+      "step": 21352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.497629581870147e-06,
+      "loss": 0.918,
+      "step": 21353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.49720335494415e-06,
+      "loss": 0.9152,
+      "step": 21354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4967773059775435e-06,
+      "loss": 0.9285,
+      "step": 21355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.496351434978308e-06,
+      "loss": 0.9361,
+      "step": 21356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495925741954419e-06,
+      "loss": 0.9379,
+      "step": 21357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495500226913844e-06,
+      "loss": 0.9608,
+      "step": 21358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4950748898645584e-06,
+      "loss": 0.9165,
+      "step": 21359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4946497308145213e-06,
+      "loss": 0.9457,
+      "step": 21360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4942247497717e-06,
+      "loss": 1.0116,
+      "step": 21361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.493799946744049e-06,
+      "loss": 0.936,
+      "step": 21362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4933753217395265e-06,
+      "loss": 0.9993,
+      "step": 21363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4929508747660816e-06,
+      "loss": 0.9249,
+      "step": 21364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4925266058316658e-06,
+      "loss": 0.9074,
+      "step": 21365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.492102514944224e-06,
+      "loss": 0.9283,
+      "step": 21366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.491678602111696e-06,
+      "loss": 0.886,
+      "step": 21367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4912548673420214e-06,
+      "loss": 0.9902,
+      "step": 21368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4908313106431362e-06,
+      "loss": 0.9163,
+      "step": 21369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4904079320229735e-06,
+      "loss": 0.9056,
+      "step": 21370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4899847314894576e-06,
+      "loss": 0.9515,
+      "step": 21371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.48956170905052e-06,
+      "loss": 0.8902,
+      "step": 21372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4891388647140784e-06,
+      "loss": 0.9496,
+      "step": 21373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4887161984880516e-06,
+      "loss": 0.9356,
+      "step": 21374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4882937103803553e-06,
+      "loss": 0.8995,
+      "step": 21375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4878714003989025e-06,
+      "loss": 0.9985,
+      "step": 21376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487449268551602e-06,
+      "loss": 0.9843,
+      "step": 21377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4870273148463556e-06,
+      "loss": 1.0197,
+      "step": 21378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4866055392910705e-06,
+      "loss": 0.9398,
+      "step": 21379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.486183941893642e-06,
+      "loss": 0.8893,
+      "step": 21380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4857625226619666e-06,
+      "loss": 0.9135,
+      "step": 21381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.485341281603935e-06,
+      "loss": 0.9132,
+      "step": 21382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484920218727436e-06,
+      "loss": 0.9745,
+      "step": 21383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484499334040357e-06,
+      "loss": 0.9497,
+      "step": 21384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.484078627550577e-06,
+      "loss": 0.9902,
+      "step": 21385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4836580992659766e-06,
+      "loss": 0.986,
+      "step": 21386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4832377491944295e-06,
+      "loss": 0.939,
+      "step": 21387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4828175773438093e-06,
+      "loss": 0.9998,
+      "step": 21388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.482397583721983e-06,
+      "loss": 0.9293,
+      "step": 21389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4819777683368188e-06,
+      "loss": 0.981,
+      "step": 21390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4815581311961736e-06,
+      "loss": 0.9164,
+      "step": 21391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4811386723079112e-06,
+      "loss": 0.9399,
+      "step": 21392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4807193916798824e-06,
+      "loss": 1.0069,
+      "step": 21393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4803002893199424e-06,
+      "loss": 0.9711,
+      "step": 21394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.479881365235939e-06,
+      "loss": 1.0013,
+      "step": 21395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4794626194357144e-06,
+      "loss": 0.9153,
+      "step": 21396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4790440519271154e-06,
+      "loss": 0.9152,
+      "step": 21397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.478625662717976e-06,
+      "loss": 0.992,
+      "step": 21398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.478207451816136e-06,
+      "loss": 0.9554,
+      "step": 21399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.477789419229424e-06,
+      "loss": 0.9775,
+      "step": 21400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4773715649656683e-06,
+      "loss": 0.938,
+      "step": 21401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4769538890326964e-06,
+      "loss": 0.9411,
+      "step": 21402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4765363914383267e-06,
+      "loss": 0.9583,
+      "step": 21403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4761190721903817e-06,
+      "loss": 1.0158,
+      "step": 21404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4757019312966748e-06,
+      "loss": 0.9254,
+      "step": 21405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4752849687650164e-06,
+      "loss": 0.9324,
+      "step": 21406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4748681846032166e-06,
+      "loss": 0.8891,
+      "step": 21407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4744515788190793e-06,
+      "loss": 0.9847,
+      "step": 21408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.474035151420409e-06,
+      "loss": 0.9045,
+      "step": 21409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4736189024150026e-06,
+      "loss": 0.9617,
+      "step": 21410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.473202831810654e-06,
+      "loss": 0.9514,
+      "step": 21411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.472786939615155e-06,
+      "loss": 0.8641,
+      "step": 21412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4723712258362968e-06,
+      "loss": 0.9424,
+      "step": 21413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.471955690481861e-06,
+      "loss": 0.9068,
+      "step": 21414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4715403335596326e-06,
+      "loss": 0.9547,
+      "step": 21415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4711251550773864e-06,
+      "loss": 0.9093,
+      "step": 21416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4707101550429014e-06,
+      "loss": 0.9862,
+      "step": 21417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470295333463946e-06,
+      "loss": 0.927,
+      "step": 21418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4698806903482922e-06,
+      "loss": 0.8842,
+      "step": 21419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4694662257037006e-06,
+      "loss": 0.9796,
+      "step": 21420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4690519395379367e-06,
+      "loss": 0.9755,
+      "step": 21421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.468637831858758e-06,
+      "loss": 0.9514,
+      "step": 21422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4682239026739155e-06,
+      "loss": 0.9545,
+      "step": 21423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.467810151991168e-06,
+      "loss": 0.9162,
+      "step": 21424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.467396579818258e-06,
+      "loss": 0.9804,
+      "step": 21425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4669831861629323e-06,
+      "loss": 0.9161,
+      "step": 21426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.466569971032934e-06,
+      "loss": 0.9299,
+      "step": 21427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4661569344359992e-06,
+      "loss": 1.0111,
+      "step": 21428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.465744076379865e-06,
+      "loss": 0.9181,
+      "step": 21429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4653313968722617e-06,
+      "loss": 0.966,
+      "step": 21430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.464918895920916e-06,
+      "loss": 0.9588,
+      "step": 21431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.464506573533556e-06,
+      "loss": 0.9363,
+      "step": 21432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4640944297179018e-06,
+      "loss": 0.9715,
+      "step": 21433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4636824644816705e-06,
+      "loss": 0.9754,
+      "step": 21434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4632706778325813e-06,
+      "loss": 0.9247,
+      "step": 21435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4628590697783384e-06,
+      "loss": 0.9108,
+      "step": 21436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4624476403266573e-06,
+      "loss": 0.8837,
+      "step": 21437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.46203638948524e-06,
+      "loss": 0.9335,
+      "step": 21438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.461625317261786e-06,
+      "loss": 0.878,
+      "step": 21439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.461214423663997e-06,
+      "loss": 0.9166,
+      "step": 21440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.460803708699563e-06,
+      "loss": 1.0375,
+      "step": 21441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.460393172376182e-06,
+      "loss": 0.9628,
+      "step": 21442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.459982814701537e-06,
+      "loss": 1.0298,
+      "step": 21443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.459572635683315e-06,
+      "loss": 0.899,
+      "step": 21444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.459162635329198e-06,
+      "loss": 0.964,
+      "step": 21445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458752813646861e-06,
+      "loss": 0.9114,
+      "step": 21446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.458343170643982e-06,
+      "loss": 0.9093,
+      "step": 21447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457933706328232e-06,
+      "loss": 0.945,
+      "step": 21448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457524420707277e-06,
+      "loss": 0.9751,
+      "step": 21449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.457115313788784e-06,
+      "loss": 0.9139,
+      "step": 21450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456706385580412e-06,
+      "loss": 0.9733,
+      "step": 21451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.456297636089822e-06,
+      "loss": 0.9242,
+      "step": 21452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.455889065324667e-06,
+      "loss": 1.0062,
+      "step": 21453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4554806732925962e-06,
+      "loss": 0.9747,
+      "step": 21454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.455072460001262e-06,
+      "loss": 1.0095,
+      "step": 21455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4546644254583067e-06,
+      "loss": 0.9247,
+      "step": 21456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.454256569671371e-06,
+      "loss": 1.0011,
+      "step": 21457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4538488926480933e-06,
+      "loss": 0.9264,
+      "step": 21458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.453441394396108e-06,
+      "loss": 0.9146,
+      "step": 21459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.453034074923049e-06,
+      "loss": 0.9158,
+      "step": 21460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4526269342365387e-06,
+      "loss": 0.9351,
+      "step": 21461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4522199723442075e-06,
+      "loss": 0.8594,
+      "step": 21462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4518131892536735e-06,
+      "loss": 0.9601,
+      "step": 21463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4514065849725544e-06,
+      "loss": 0.9512,
+      "step": 21464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4510001595084656e-06,
+      "loss": 0.9009,
+      "step": 21465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4505939128690184e-06,
+      "loss": 0.9217,
+      "step": 21466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.45018784506182e-06,
+      "loss": 0.9313,
+      "step": 21467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.449781956094476e-06,
+      "loss": 0.9688,
+      "step": 21468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4493762459745854e-06,
+      "loss": 0.948,
+      "step": 21469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4489707147097467e-06,
+      "loss": 0.9086,
+      "step": 21470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4485653623075548e-06,
+      "loss": 0.9017,
+      "step": 21471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4481601887756015e-06,
+      "loss": 0.9938,
+      "step": 21472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4477551941214742e-06,
+      "loss": 0.9409,
+      "step": 21473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4473503783527543e-06,
+      "loss": 0.9287,
+      "step": 21474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4469457414770282e-06,
+      "loss": 0.964,
+      "step": 21475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.446541283501869e-06,
+      "loss": 0.9013,
+      "step": 21476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.446137004434852e-06,
+      "loss": 0.8647,
+      "step": 21477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.445732904283551e-06,
+      "loss": 0.9661,
+      "step": 21478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4453289830555297e-06,
+      "loss": 0.9054,
+      "step": 21479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4449252407583547e-06,
+      "loss": 0.8932,
+      "step": 21480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.444521677399585e-06,
+      "loss": 0.9616,
+      "step": 21481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4441182929867814e-06,
+      "loss": 0.9739,
+      "step": 21482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.443715087527495e-06,
+      "loss": 0.9075,
+      "step": 21483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4433120610292773e-06,
+      "loss": 0.923,
+      "step": 21484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.442909213499677e-06,
+      "loss": 0.9552,
+      "step": 21485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4425065449462374e-06,
+      "loss": 0.8932,
+      "step": 21486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4421040553764997e-06,
+      "loss": 0.9174,
+      "step": 21487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.441701744798001e-06,
+      "loss": 0.9698,
+      "step": 21488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4412996132182745e-06,
+      "loss": 0.8972,
+      "step": 21489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4408976606448523e-06,
+      "loss": 0.9733,
+      "step": 21490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4404958870852604e-06,
+      "loss": 0.9416,
+      "step": 21491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.440094292547025e-06,
+      "loss": 0.9386,
+      "step": 21492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4396928770376663e-06,
+      "loss": 0.9696,
+      "step": 21493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4392916405646998e-06,
+      "loss": 0.9747,
+      "step": 21494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4388905831356404e-06,
+      "loss": 0.9768,
+      "step": 21495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.438489704757999e-06,
+      "loss": 0.9363,
+      "step": 21496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.438089005439284e-06,
+      "loss": 0.9637,
+      "step": 21497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.437688485186998e-06,
+      "loss": 0.8786,
+      "step": 21498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4372881440086386e-06,
+      "loss": 0.9635,
+      "step": 21499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.43688798191171e-06,
+      "loss": 0.9864,
+      "step": 21500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4364879989036984e-06,
+      "loss": 0.9289,
+      "step": 21501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4360881949921013e-06,
+      "loss": 0.9542,
+      "step": 21502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4356885701844017e-06,
+      "loss": 0.9755,
+      "step": 21503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.435289124488083e-06,
+      "loss": 0.9424,
+      "step": 21504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4348898579106287e-06,
+      "loss": 0.9044,
+      "step": 21505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4344907704595136e-06,
+      "loss": 0.9156,
+      "step": 21506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4340918621422105e-06,
+      "loss": 0.9304,
+      "step": 21507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.433693132966193e-06,
+      "loss": 0.9677,
+      "step": 21508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4332945829389243e-06,
+      "loss": 0.9884,
+      "step": 21509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4328962120678713e-06,
+      "loss": 0.8669,
+      "step": 21510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4324980203604943e-06,
+      "loss": 0.9791,
+      "step": 21511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4321000078242474e-06,
+      "loss": 0.9198,
+      "step": 21512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4317021744665872e-06,
+      "loss": 0.9312,
+      "step": 21513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4313045202949617e-06,
+      "loss": 0.9082,
+      "step": 21514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.430907045316818e-06,
+      "loss": 0.9158,
+      "step": 21515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4305097495396036e-06,
+      "loss": 0.961,
+      "step": 21516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4301126329707518e-06,
+      "loss": 0.999,
+      "step": 21517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.429715695617705e-06,
+      "loss": 0.975,
+      "step": 21518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4293189374878924e-06,
+      "loss": 0.8784,
+      "step": 21519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.428922358588749e-06,
+      "loss": 0.9264,
+      "step": 21520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.428525958927698e-06,
+      "loss": 0.8701,
+      "step": 21521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.428129738512164e-06,
+      "loss": 0.9889,
+      "step": 21522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427733697349566e-06,
+      "loss": 0.9247,
+      "step": 21523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.427337835447322e-06,
+      "loss": 0.9783,
+      "step": 21524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4269421528128467e-06,
+      "loss": 0.9577,
+      "step": 21525
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4265466494535464e-06,
+      "loss": 0.8791,
+      "step": 21526
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.426151325376829e-06,
+      "loss": 0.9582,
+      "step": 21527
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4257561805900986e-06,
+      "loss": 0.8901,
+      "step": 21528
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4253612151007555e-06,
+      "loss": 0.9217,
+      "step": 21529
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4249664289161947e-06,
+      "loss": 0.9415,
+      "step": 21530
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.424571822043812e-06,
+      "loss": 0.9756,
+      "step": 21531
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4241773944909945e-06,
+      "loss": 0.9638,
+      "step": 21532
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4237831462651286e-06,
+      "loss": 0.9935,
+      "step": 21533
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4233890773735983e-06,
+      "loss": 1.0649,
+      "step": 21534
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.422995187823785e-06,
+      "loss": 0.953,
+      "step": 21535
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4226014776230635e-06,
+      "loss": 0.944,
+      "step": 21536
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.422207946778804e-06,
+      "loss": 0.9684,
+      "step": 21537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421814595298382e-06,
+      "loss": 0.9432,
+      "step": 21538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4214214231891587e-06,
+      "loss": 0.8865,
+      "step": 21539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.421028430458502e-06,
+      "loss": 0.9005,
+      "step": 21540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4206356171137668e-06,
+      "loss": 0.8924,
+      "step": 21541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.420242983162311e-06,
+      "loss": 0.9908,
+      "step": 21542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4198505286114896e-06,
+      "loss": 0.9529,
+      "step": 21543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4194582534686476e-06,
+      "loss": 0.9642,
+      "step": 21544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.419066157741137e-06,
+      "loss": 0.9376,
+      "step": 21545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.418674241436295e-06,
+      "loss": 0.9609,
+      "step": 21546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4182825045614637e-06,
+      "loss": 0.9109,
+      "step": 21547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.417890947123979e-06,
+      "loss": 0.9545,
+      "step": 21548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4174995691311734e-06,
+      "loss": 0.9351,
+      "step": 21549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4171083705903767e-06,
+      "loss": 0.9801,
+      "step": 21550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.416717351508916e-06,
+      "loss": 0.9086,
+      "step": 21551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4163265118941104e-06,
+      "loss": 0.8968,
+      "step": 21552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.415935851753283e-06,
+      "loss": 1.0093,
+      "step": 21553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.415545371093746e-06,
+      "loss": 0.9362,
+      "step": 21554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.415155069922816e-06,
+      "loss": 0.9702,
+      "step": 21555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4147649482478e-06,
+      "loss": 0.992,
+      "step": 21556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4143750060760008e-06,
+      "loss": 0.8799,
+      "step": 21557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4139852434147285e-06,
+      "loss": 0.9013,
+      "step": 21558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4135956602712754e-06,
+      "loss": 0.9672,
+      "step": 21559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4132062566529394e-06,
+      "loss": 0.9854,
+      "step": 21560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.412817032567013e-06,
+      "loss": 0.9399,
+      "step": 21561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.412427988020786e-06,
+      "loss": 0.8963,
+      "step": 21562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.412039123021545e-06,
+      "loss": 0.9616,
+      "step": 21563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.411650437576567e-06,
+      "loss": 0.9957,
+      "step": 21564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4112619316931374e-06,
+      "loss": 0.9374,
+      "step": 21565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.410873605378529e-06,
+      "loss": 0.9903,
+      "step": 21566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4104854586400118e-06,
+      "loss": 0.9504,
+      "step": 21567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4100974914848585e-06,
+      "loss": 1.0074,
+      "step": 21568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4097097039203324e-06,
+      "loss": 0.9075,
+      "step": 21569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.409322095953697e-06,
+      "loss": 0.956,
+      "step": 21570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408934667592209e-06,
+      "loss": 0.9381,
+      "step": 21571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4085474188431258e-06,
+      "loss": 0.9914,
+      "step": 21572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4081603497136974e-06,
+      "loss": 0.9296,
+      "step": 21573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407773460211176e-06,
+      "loss": 0.9002,
+      "step": 21574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407386750342802e-06,
+      "loss": 0.9437,
+      "step": 21575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.407000220115822e-06,
+      "loss": 0.9383,
+      "step": 21576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4066138695374697e-06,
+      "loss": 0.9048,
+      "step": 21577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.406227698614986e-06,
+      "loss": 0.9793,
+      "step": 21578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4058417073555977e-06,
+      "loss": 0.8823,
+      "step": 21579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4054558957665367e-06,
+      "loss": 0.9527,
+      "step": 21580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4050702638550275e-06,
+      "loss": 0.925,
+      "step": 21581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4046848116282885e-06,
+      "loss": 0.9775,
+      "step": 21582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.404299539093543e-06,
+      "loss": 0.8963,
+      "step": 21583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4039144462580033e-06,
+      "loss": 0.8979,
+      "step": 21584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4035295331288814e-06,
+      "loss": 0.9744,
+      "step": 21585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.403144799713385e-06,
+      "loss": 0.9553,
+      "step": 21586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.402760246018721e-06,
+      "loss": 0.9602,
+      "step": 21587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4023758720520898e-06,
+      "loss": 0.9306,
+      "step": 21588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4019916778206888e-06,
+      "loss": 0.918,
+      "step": 21589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4016076633317133e-06,
+      "loss": 0.9115,
+      "step": 21590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4012238285923543e-06,
+      "loss": 0.9414,
+      "step": 21591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400840173609801e-06,
+      "loss": 0.8946,
+      "step": 21592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400456698391238e-06,
+      "loss": 0.9814,
+      "step": 21593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.400073402943848e-06,
+      "loss": 0.9997,
+      "step": 21594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3996902872748043e-06,
+      "loss": 0.943,
+      "step": 21595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.399307351391287e-06,
+      "loss": 0.9591,
+      "step": 21596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.398924595300465e-06,
+      "loss": 0.9455,
+      "step": 21597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3985420190095046e-06,
+      "loss": 0.9861,
+      "step": 21598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3981596225255734e-06,
+      "loss": 1.0203,
+      "step": 21599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.397777405855829e-06,
+      "loss": 0.9347,
+      "step": 21600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.397395369007433e-06,
+      "loss": 0.9316,
+      "step": 21601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3970135119875363e-06,
+      "loss": 0.9272,
+      "step": 21602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3966318348032936e-06,
+      "loss": 0.9566,
+      "step": 21603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3962503374618498e-06,
+      "loss": 0.915,
+      "step": 21604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3958690199703497e-06,
+      "loss": 0.9727,
+      "step": 21605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3954878823359354e-06,
+      "loss": 0.9254,
+      "step": 21606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3951069245657433e-06,
+      "loss": 0.987,
+      "step": 21607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3947261466669096e-06,
+      "loss": 0.9199,
+      "step": 21608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.394345548646562e-06,
+      "loss": 0.9224,
+      "step": 21609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3939651305118296e-06,
+      "loss": 0.9864,
+      "step": 21610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.393584892269837e-06,
+      "loss": 0.997,
+      "step": 21611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.393204833927704e-06,
+      "loss": 1.0147,
+      "step": 21612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3928249554925477e-06,
+      "loss": 0.8922,
+      "step": 21613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3924452569714854e-06,
+      "loss": 0.9809,
+      "step": 21614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.392065738371623e-06,
+      "loss": 0.9423,
+      "step": 21615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.391686399700071e-06,
+      "loss": 0.8984,
+      "step": 21616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.391307240963932e-06,
+      "loss": 0.8321,
+      "step": 21617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.390928262170306e-06,
+      "loss": 0.9278,
+      "step": 21618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3905494633262917e-06,
+      "loss": 0.9587,
+      "step": 21619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3901708444389804e-06,
+      "loss": 0.9738,
+      "step": 21620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.389792405515467e-06,
+      "loss": 0.9462,
+      "step": 21621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.389414146562833e-06,
+      "loss": 0.972,
+      "step": 21622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3890360675881665e-06,
+      "loss": 0.9639,
+      "step": 21623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3886581685985456e-06,
+      "loss": 0.9469,
+      "step": 21624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3882804496010484e-06,
+      "loss": 0.9062,
+      "step": 21625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.387902910602749e-06,
+      "loss": 0.9718,
+      "step": 21626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3875255516107133e-06,
+      "loss": 1.0155,
+      "step": 21627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3871483726320133e-06,
+      "loss": 0.9965,
+      "step": 21628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3867713736737097e-06,
+      "loss": 0.9729,
+      "step": 21629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3863945547428636e-06,
+      "loss": 0.9022,
+      "step": 21630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3860179158465315e-06,
+      "loss": 0.9678,
+      "step": 21631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3856414569917656e-06,
+      "loss": 0.9219,
+      "step": 21632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3852651781856175e-06,
+      "loss": 0.9619,
+      "step": 21633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3848890794351345e-06,
+      "loss": 0.9314,
+      "step": 21634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384513160747357e-06,
+      "loss": 0.8865,
+      "step": 21635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.384137422129327e-06,
+      "loss": 0.9442,
+      "step": 21636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.38376186358808e-06,
+      "loss": 0.8758,
+      "step": 21637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.383386485130649e-06,
+      "loss": 0.9471,
+      "step": 21638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3830112867640656e-06,
+      "loss": 0.9251,
+      "step": 21639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.382636268495353e-06,
+      "loss": 0.9098,
+      "step": 21640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3822614303315373e-06,
+      "loss": 0.9716,
+      "step": 21641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.381886772279637e-06,
+      "loss": 0.9262,
+      "step": 21642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3815122943466663e-06,
+      "loss": 0.9552,
+      "step": 21643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.381137996539641e-06,
+      "loss": 0.9171,
+      "step": 21644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3807638788655696e-06,
+      "loss": 0.9091,
+      "step": 21645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3803899413314583e-06,
+      "loss": 0.9329,
+      "step": 21646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3800161839443096e-06,
+      "loss": 0.8688,
+      "step": 21647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3796426067111212e-06,
+      "loss": 0.9747,
+      "step": 21648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3792692096388926e-06,
+      "loss": 0.9585,
+      "step": 21649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3788959927346135e-06,
+      "loss": 0.9732,
+      "step": 21650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.378522956005275e-06,
+      "loss": 0.9216,
+      "step": 21651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3781500994578623e-06,
+      "loss": 1.008,
+      "step": 21652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.377777423099358e-06,
+      "loss": 0.9396,
+      "step": 21653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3774049269367407e-06,
+      "loss": 0.9383,
+      "step": 21654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3770326109769862e-06,
+      "loss": 0.9591,
+      "step": 21655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3766604752270674e-06,
+      "loss": 0.8943,
+      "step": 21656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3762885196939537e-06,
+      "loss": 0.8894,
+      "step": 21657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.375916744384609e-06,
+      "loss": 0.9467,
+      "step": 21658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.375545149305998e-06,
+      "loss": 0.9797,
+      "step": 21659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.375173734465076e-06,
+      "loss": 0.968,
+      "step": 21660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3748024998688038e-06,
+      "loss": 0.9619,
+      "step": 21661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3744314455241293e-06,
+      "loss": 0.909,
+      "step": 21662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.374060571438003e-06,
+      "loss": 0.9878,
+      "step": 21663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3736898776173704e-06,
+      "loss": 0.9637,
+      "step": 21664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.37331936406917e-06,
+      "loss": 0.9507,
+      "step": 21665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3729490308003457e-06,
+      "loss": 0.9672,
+      "step": 21666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3725788778178297e-06,
+      "loss": 0.9269,
+      "step": 21667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3722089051285537e-06,
+      "loss": 0.915,
+      "step": 21668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3718391127394482e-06,
+      "loss": 0.9478,
+      "step": 21669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3714695006574365e-06,
+      "loss": 0.9481,
+      "step": 21670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3711000688894416e-06,
+      "loss": 0.9965,
+      "step": 21671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.370730817442382e-06,
+      "loss": 0.9444,
+      "step": 21672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.37036174632317e-06,
+      "loss": 0.9665,
+      "step": 21673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.36999285553872e-06,
+      "loss": 0.9145,
+      "step": 21674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.369624145095939e-06,
+      "loss": 0.9174,
+      "step": 21675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3692556150017334e-06,
+      "loss": 0.9509,
+      "step": 21676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.368887265263004e-06,
+      "loss": 0.9086,
+      "step": 21677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.368519095886646e-06,
+      "loss": 0.9824,
+      "step": 21678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3681511068795585e-06,
+      "loss": 0.9323,
+      "step": 21679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3677832982486306e-06,
+      "loss": 0.8932,
+      "step": 21680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3674156700007503e-06,
+      "loss": 0.8982,
+      "step": 21681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3670482221428027e-06,
+      "loss": 0.8776,
+      "step": 21682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3666809546816683e-06,
+      "loss": 0.9698,
+      "step": 21683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.366313867624227e-06,
+      "loss": 0.9775,
+      "step": 21684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.365946960977349e-06,
+      "loss": 0.9412,
+      "step": 21685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.365580234747911e-06,
+      "loss": 0.9195,
+      "step": 21686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.365213688942776e-06,
+      "loss": 0.9248,
+      "step": 21687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.364847323568811e-06,
+      "loss": 0.9465,
+      "step": 21688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.364481138632875e-06,
+      "loss": 0.9136,
+      "step": 21689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3641151341418268e-06,
+      "loss": 0.9526,
+      "step": 21690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3637493101025226e-06,
+      "loss": 0.9764,
+      "step": 21691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.363383666521809e-06,
+      "loss": 0.9265,
+      "step": 21692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3630182034065353e-06,
+      "loss": 0.9032,
+      "step": 21693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3626529207635466e-06,
+      "loss": 0.9568,
+      "step": 21694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3622878185996814e-06,
+      "loss": 0.9803,
+      "step": 21695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.361922896921779e-06,
+      "loss": 0.8541,
+      "step": 21696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.361558155736674e-06,
+      "loss": 1.0186,
+      "step": 21697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3611935950511935e-06,
+      "loss": 0.9386,
+      "step": 21698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.360829214872168e-06,
+      "loss": 0.9376,
+      "step": 21699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3604650152064193e-06,
+      "loss": 0.9401,
+      "step": 21700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.360100996060768e-06,
+      "loss": 0.8867,
+      "step": 21701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3597371574420327e-06,
+      "loss": 0.9199,
+      "step": 21702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.359373499357024e-06,
+      "loss": 0.9837,
+      "step": 21703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3590100218125558e-06,
+      "loss": 0.9585,
+      "step": 21704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3586467248154312e-06,
+      "loss": 0.9151,
+      "step": 21705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.358283608372457e-06,
+      "loss": 0.9912,
+      "step": 21706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3579206724904316e-06,
+      "loss": 0.922,
+      "step": 21707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.357557917176152e-06,
+      "loss": 0.9592,
+      "step": 21708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3571953424364116e-06,
+      "loss": 0.9813,
+      "step": 21709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.356832948278002e-06,
+      "loss": 0.972,
+      "step": 21710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3564707347077055e-06,
+      "loss": 0.9371,
+      "step": 21711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3561087017323094e-06,
+      "loss": 0.9652,
+      "step": 21712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.355746849358591e-06,
+      "loss": 0.9891,
+      "step": 21713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.355385177593328e-06,
+      "loss": 0.9259,
+      "step": 21714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3550236864432943e-06,
+      "loss": 1.0134,
+      "step": 21715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3546623759152565e-06,
+      "loss": 0.9363,
+      "step": 21716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3543012460159845e-06,
+      "loss": 0.9202,
+      "step": 21717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3539402967522386e-06,
+      "loss": 0.9364,
+      "step": 21718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3535795281307794e-06,
+      "loss": 0.898,
+      "step": 21719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.353218940158364e-06,
+      "loss": 0.9254,
+      "step": 21720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3528585328417423e-06,
+      "loss": 0.9367,
+      "step": 21721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3524983061876673e-06,
+      "loss": 0.9734,
+      "step": 21722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.352138260202881e-06,
+      "loss": 0.9323,
+      "step": 21723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.35177839489413e-06,
+      "loss": 0.9543,
+      "step": 21724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3514187102681507e-06,
+      "loss": 0.9817,
+      "step": 21725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3510592063316797e-06,
+      "loss": 0.9141,
+      "step": 21726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.35069988309145e-06,
+      "loss": 0.9452,
+      "step": 21727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3503407405541893e-06,
+      "loss": 0.9119,
+      "step": 21728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3499817787266256e-06,
+      "loss": 0.9149,
+      "step": 21729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3496229976154798e-06,
+      "loss": 0.9586,
+      "step": 21730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.34926439722747e-06,
+      "loss": 0.9111,
+      "step": 21731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3489059775693125e-06,
+      "loss": 0.9608,
+      "step": 21732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.34854773864772e-06,
+      "loss": 0.992,
+      "step": 21733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3481896804694007e-06,
+      "loss": 0.8794,
+      "step": 21734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.347831803041061e-06,
+      "loss": 0.9942,
+      "step": 21735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3474741063694e-06,
+      "loss": 1.0035,
+      "step": 21736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3471165904611194e-06,
+      "loss": 0.9414,
+      "step": 21737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3467592553229125e-06,
+      "loss": 0.9448,
+      "step": 21738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3464021009614727e-06,
+      "loss": 0.9684,
+      "step": 21739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3460451273834886e-06,
+      "loss": 1.0031,
+      "step": 21740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3456883345956417e-06,
+      "loss": 1.0246,
+      "step": 21741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.345331722604619e-06,
+      "loss": 0.912,
+      "step": 21742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3449752914170933e-06,
+      "loss": 0.9229,
+      "step": 21743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3446190410397452e-06,
+      "loss": 0.9669,
+      "step": 21744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.344262971479242e-06,
+      "loss": 0.9068,
+      "step": 21745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.343907082742253e-06,
+      "loss": 0.9744,
+      "step": 21746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3435513748354425e-06,
+      "loss": 0.878,
+      "step": 21747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3431958477654727e-06,
+      "loss": 0.9196,
+      "step": 21748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3428405015390036e-06,
+      "loss": 0.9206,
+      "step": 21749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3424853361626856e-06,
+      "loss": 0.8973,
+      "step": 21750
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3421303516431717e-06,
+      "loss": 0.9523,
+      "step": 21751
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.34177554798711e-06,
+      "loss": 0.945,
+      "step": 21752
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3414209252011446e-06,
+      "loss": 0.9129,
+      "step": 21753
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3410664832919163e-06,
+      "loss": 0.9406,
+      "step": 21754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3407122222660653e-06,
+      "loss": 0.9302,
+      "step": 21755
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3403581421302225e-06,
+      "loss": 0.9076,
+      "step": 21756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3400042428910198e-06,
+      "loss": 1.0011,
+      "step": 21757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.339650524555084e-06,
+      "loss": 0.909,
+      "step": 21758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3392969871290408e-06,
+      "loss": 0.916,
+      "step": 21759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3389436306195113e-06,
+      "loss": 1.0126,
+      "step": 21760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.33859045503311e-06,
+      "loss": 1.0009,
+      "step": 21761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3382374603764547e-06,
+      "loss": 0.9225,
+      "step": 21762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.337884646656151e-06,
+      "loss": 0.8887,
+      "step": 21763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.337532013878812e-06,
+      "loss": 0.9412,
+      "step": 21764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3371795620510363e-06,
+      "loss": 0.9273,
+      "step": 21765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.336827291179427e-06,
+      "loss": 0.9206,
+      "step": 21766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3364752012705806e-06,
+      "loss": 0.9333,
+      "step": 21767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3361232923310884e-06,
+      "loss": 0.9649,
+      "step": 21768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.335771564367546e-06,
+      "loss": 0.9646,
+      "step": 21769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3354200173865352e-06,
+      "loss": 0.9142,
+      "step": 21770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.335068651394642e-06,
+      "loss": 0.9387,
+      "step": 21771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3347174663984455e-06,
+      "loss": 0.9506,
+      "step": 21772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3343664624045215e-06,
+      "loss": 0.8984,
+      "step": 21773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.334015639419446e-06,
+      "loss": 1.0155,
+      "step": 21774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.333664997449786e-06,
+      "loss": 0.9271,
+      "step": 21775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3333145365021105e-06,
+      "loss": 0.9352,
+      "step": 21776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.332964256582979e-06,
+      "loss": 0.9429,
+      "step": 21777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.332614157698957e-06,
+      "loss": 0.976,
+      "step": 21778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3322642398565935e-06,
+      "loss": 0.955,
+      "step": 21779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.331914503062448e-06,
+      "loss": 0.8767,
+      "step": 21780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3315649473230654e-06,
+      "loss": 0.909,
+      "step": 21781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.331215572644995e-06,
+      "loss": 0.9561,
+      "step": 21782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3308663790347776e-06,
+      "loss": 0.8753,
+      "step": 21783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3305173664989537e-06,
+      "loss": 0.9998,
+      "step": 21784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3301685350440597e-06,
+      "loss": 1.0004,
+      "step": 21785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3298198846766254e-06,
+      "loss": 0.9211,
+      "step": 21786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.329471415403183e-06,
+      "loss": 0.98,
+      "step": 21787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3291231272302574e-06,
+      "loss": 0.8625,
+      "step": 21788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.32877502016437e-06,
+      "loss": 1.0016,
+      "step": 21789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3284270942120414e-06,
+      "loss": 0.956,
+      "step": 21790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.328079349379786e-06,
+      "loss": 1.0148,
+      "step": 21791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327731785674117e-06,
+      "loss": 0.8835,
+      "step": 21792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.327384403101544e-06,
+      "loss": 0.9108,
+      "step": 21793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3270372016685695e-06,
+      "loss": 1.0237,
+      "step": 21794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3266901813816967e-06,
+      "loss": 0.9409,
+      "step": 21795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.326343342247426e-06,
+      "loss": 0.9252,
+      "step": 21796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3259966842722505e-06,
+      "loss": 0.9885,
+      "step": 21797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3256502074626654e-06,
+      "loss": 0.9135,
+      "step": 21798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3253039118251544e-06,
+      "loss": 0.9353,
+      "step": 21799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3249577973662065e-06,
+      "loss": 0.886,
+      "step": 21800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3246118640923024e-06,
+      "loss": 0.9524,
+      "step": 21801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3242661120099186e-06,
+      "loss": 0.8779,
+      "step": 21802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3239205411255317e-06,
+      "loss": 0.9339,
+      "step": 21803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3235751514456133e-06,
+      "loss": 0.9605,
+      "step": 21804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3232299429766316e-06,
+      "loss": 0.9996,
+      "step": 21805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3228849157250503e-06,
+      "loss": 0.9435,
+      "step": 21806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3225400696973323e-06,
+      "loss": 0.9638,
+      "step": 21807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.322195404899933e-06,
+      "loss": 0.9391,
+      "step": 21808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3218509213393094e-06,
+      "loss": 0.8943,
+      "step": 21809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3215066190219116e-06,
+      "loss": 0.9167,
+      "step": 21810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3211624979541882e-06,
+      "loss": 0.9562,
+      "step": 21811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3208185581425837e-06,
+      "loss": 1.0005,
+      "step": 21812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3204747995935374e-06,
+      "loss": 0.9604,
+      "step": 21813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3201312223134877e-06,
+      "loss": 0.9535,
+      "step": 21814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3197878263088683e-06,
+      "loss": 0.9502,
+      "step": 21815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3194446115861116e-06,
+      "loss": 0.9775,
+      "step": 21816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.319101578151644e-06,
+      "loss": 0.9599,
+      "step": 21817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3187587260118895e-06,
+      "loss": 0.8954,
+      "step": 21818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3184160551732674e-06,
+      "loss": 0.9443,
+      "step": 21819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3180735656421984e-06,
+      "loss": 0.9711,
+      "step": 21820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3177312574250933e-06,
+      "loss": 0.9504,
+      "step": 21821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3173891305283624e-06,
+      "loss": 0.922,
+      "step": 21822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3170471849584166e-06,
+      "loss": 0.94,
+      "step": 21823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3167054207216534e-06,
+      "loss": 0.9862,
+      "step": 21824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.316363837824478e-06,
+      "loss": 0.9977,
+      "step": 21825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.316022436273284e-06,
+      "loss": 0.9427,
+      "step": 21826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315681216074469e-06,
+      "loss": 0.946,
+      "step": 21827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.315340177234419e-06,
+      "loss": 0.9244,
+      "step": 21828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.314999319759522e-06,
+      "loss": 0.9031,
+      "step": 21829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3146586436561618e-06,
+      "loss": 0.9314,
+      "step": 21830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3143181489307166e-06,
+      "loss": 0.8824,
+      "step": 21831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3139778355895662e-06,
+      "loss": 0.9153,
+      "step": 21832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3136377036390803e-06,
+      "loss": 0.9469,
+      "step": 21833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3132977530856297e-06,
+      "loss": 0.9589,
+      "step": 21834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.312957983935581e-06,
+      "loss": 0.9269,
+      "step": 21835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3126183961952965e-06,
+      "loss": 0.8823,
+      "step": 21836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3122789898711363e-06,
+      "loss": 0.97,
+      "step": 21837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3119397649694577e-06,
+      "loss": 0.9481,
+      "step": 21838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.31160072149661e-06,
+      "loss": 0.8761,
+      "step": 21839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3112618594589445e-06,
+      "loss": 0.961,
+      "step": 21840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3109231788628074e-06,
+      "loss": 0.9606,
+      "step": 21841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3105846797145404e-06,
+      "loss": 0.9744,
+      "step": 21842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3102463620204844e-06,
+      "loss": 0.9214,
+      "step": 21843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3099082257869713e-06,
+      "loss": 0.9162,
+      "step": 21844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3095702710203374e-06,
+      "loss": 0.9729,
+      "step": 21845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3092324977269087e-06,
+      "loss": 0.8909,
+      "step": 21846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.308894905913014e-06,
+      "loss": 0.9884,
+      "step": 21847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3085574955849723e-06,
+      "loss": 0.9428,
+      "step": 21848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.308220266749103e-06,
+      "loss": 0.9098,
+      "step": 21849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.307883219411722e-06,
+      "loss": 0.9453,
+      "step": 21850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3075463535791412e-06,
+      "loss": 0.9463,
+      "step": 21851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3072096692576686e-06,
+      "loss": 0.8863,
+      "step": 21852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3068731664536096e-06,
+      "loss": 0.9716,
+      "step": 21853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3065368451732652e-06,
+      "loss": 0.9901,
+      "step": 21854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.306200705422934e-06,
+      "loss": 0.8744,
+      "step": 21855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.305864747208914e-06,
+      "loss": 0.9807,
+      "step": 21856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3055289705374907e-06,
+      "loss": 0.8953,
+      "step": 21857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3051933754149582e-06,
+      "loss": 0.985,
+      "step": 21858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3048579618475975e-06,
+      "loss": 0.9679,
+      "step": 21859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3045227298416906e-06,
+      "loss": 0.9673,
+      "step": 21860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.304187679403517e-06,
+      "loss": 0.9414,
+      "step": 21861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3038528105393467e-06,
+      "loss": 0.8935,
+      "step": 21862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3035181232554573e-06,
+      "loss": 0.992,
+      "step": 21863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3031836175581113e-06,
+      "loss": 0.9264,
+      "step": 21864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.302849293453576e-06,
+      "loss": 0.9382,
+      "step": 21865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3025151509481116e-06,
+      "loss": 0.9008,
+      "step": 21866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.302181190047974e-06,
+      "loss": 0.9106,
+      "step": 21867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.301847410759419e-06,
+      "loss": 0.9635,
+      "step": 21868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3015138130886974e-06,
+      "loss": 0.9467,
+      "step": 21869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.301180397042057e-06,
+      "loss": 0.9226,
+      "step": 21870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.300847162625739e-06,
+      "loss": 0.9338,
+      "step": 21871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.300514109845986e-06,
+      "loss": 0.9685,
+      "step": 21872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.300181238709034e-06,
+      "loss": 0.9493,
+      "step": 21873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.299848549221119e-06,
+      "loss": 0.9557,
+      "step": 21874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2995160413884678e-06,
+      "loss": 0.9154,
+      "step": 21875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2991837152173117e-06,
+      "loss": 0.9497,
+      "step": 21876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2988515707138696e-06,
+      "loss": 0.9035,
+      "step": 21877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.298519607884363e-06,
+      "loss": 0.9114,
+      "step": 21878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2981878267350103e-06,
+      "loss": 0.9525,
+      "step": 21879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.297856227272024e-06,
+      "loss": 0.9543,
+      "step": 21880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2975248095016135e-06,
+      "loss": 0.9416,
+      "step": 21881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2971935734299845e-06,
+      "loss": 0.9274,
+      "step": 21882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2968625190633433e-06,
+      "loss": 0.9464,
+      "step": 21883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.296531646407884e-06,
+      "loss": 0.8855,
+      "step": 21884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2962009554698084e-06,
+      "loss": 0.9138,
+      "step": 21885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2958704462553072e-06,
+      "loss": 0.9468,
+      "step": 21886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2955401187705695e-06,
+      "loss": 0.95,
+      "step": 21887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.295209973021783e-06,
+      "loss": 0.9187,
+      "step": 21888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.294880009015126e-06,
+      "loss": 0.9242,
+      "step": 21889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2945502267567834e-06,
+      "loss": 0.9203,
+      "step": 21890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2942206262529266e-06,
+      "loss": 0.96,
+      "step": 21891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.293891207509731e-06,
+      "loss": 0.9443,
+      "step": 21892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2935619705333634e-06,
+      "loss": 0.9249,
+      "step": 21893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2932329153299917e-06,
+      "loss": 0.9811,
+      "step": 21894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.292904041905777e-06,
+      "loss": 0.9054,
+      "step": 21895
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.292575350266878e-06,
+      "loss": 0.9511,
+      "step": 21896
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.29224684041945e-06,
+      "loss": 0.9387,
+      "step": 21897
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2919185123696454e-06,
+      "loss": 0.9823,
+      "step": 21898
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.291590366123613e-06,
+      "loss": 0.9695,
+      "step": 21899
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2912624016874963e-06,
+      "loss": 0.9232,
+      "step": 21900
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290934619067441e-06,
+      "loss": 0.9508,
+      "step": 21901
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2906070182695805e-06,
+      "loss": 0.9373,
+      "step": 21902
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.290279599300054e-06,
+      "loss": 0.9402,
+      "step": 21903
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.28995236216499e-06,
+      "loss": 0.9808,
+      "step": 21904
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289625306870519e-06,
+      "loss": 0.925,
+      "step": 21905
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2892984334227656e-06,
+      "loss": 0.9151,
+      "step": 21906
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2889717418278492e-06,
+      "loss": 0.9576,
+      "step": 21907
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2886452320918897e-06,
+      "loss": 0.9425,
+      "step": 21908
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.288318904221e-06,
+      "loss": 0.9815,
+      "step": 21909
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287992758221294e-06,
+      "loss": 0.9601,
+      "step": 21910
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2876667940988764e-06,
+      "loss": 0.9232,
+      "step": 21911
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2873410118598527e-06,
+      "loss": 1.0166,
+      "step": 21912
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.287015411510325e-06,
+      "loss": 0.9467,
+      "step": 21913
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.286689993056389e-06,
+      "loss": 0.952,
+      "step": 21914
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2863647565041403e-06,
+      "loss": 0.9155,
+      "step": 21915
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2860397018596695e-06,
+      "loss": 0.9735,
+      "step": 21916
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2857148291290623e-06,
+      "loss": 0.8802,
+      "step": 21917
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2853901383184033e-06,
+      "loss": 0.9872,
+      "step": 21918
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.285065629433775e-06,
+      "loss": 0.9441,
+      "step": 21919
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2847413024812507e-06,
+      "loss": 0.974,
+      "step": 21920
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2844171574669083e-06,
+      "loss": 0.9478,
+      "step": 21921
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.284093194396815e-06,
+      "loss": 0.9935,
+      "step": 21922
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2837694132770394e-06,
+      "loss": 1.0197,
+      "step": 21923
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2834458141136443e-06,
+      "loss": 0.9124,
+      "step": 21924
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2831223969126896e-06,
+      "loss": 0.9371,
+      "step": 21925
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.282799161680233e-06,
+      "loss": 0.9278,
+      "step": 21926
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2824761084223246e-06,
+      "loss": 0.9148,
+      "step": 21927
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2821532371450184e-06,
+      "loss": 0.9964,
+      "step": 21928
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.281830547854358e-06,
+      "loss": 0.9612,
+      "step": 21929
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2815080405563876e-06,
+      "loss": 0.9103,
+      "step": 21930
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.281185715257147e-06,
+      "loss": 0.9453,
+      "step": 21931
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2808635719626717e-06,
+      "loss": 0.9583,
+      "step": 21932
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280541610678994e-06,
+      "loss": 0.932,
+      "step": 21933
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.280219831412146e-06,
+      "loss": 0.9382,
+      "step": 21934
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2798982341681497e-06,
+      "loss": 0.963,
+      "step": 21935
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2795768189530312e-06,
+      "loss": 0.9542,
+      "step": 21936
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2792555857728075e-06,
+      "loss": 0.9477,
+      "step": 21937
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2789345346334947e-06,
+      "loss": 0.9218,
+      "step": 21938
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2786136655411062e-06,
+      "loss": 0.945,
+      "step": 21939
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2782929785016485e-06,
+      "loss": 0.904,
+      "step": 21940
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2779724735211307e-06,
+      "loss": 0.9625,
+      "step": 21941
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.277652150605553e-06,
+      "loss": 0.9165,
+      "step": 21942
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.277332009760913e-06,
+      "loss": 0.9528,
+      "step": 21943
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2770120509932082e-06,
+      "loss": 0.9038,
+      "step": 21944
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2766922743084274e-06,
+      "loss": 0.9318,
+      "step": 21945
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2763726797125644e-06,
+      "loss": 0.9178,
+      "step": 21946
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.276053267211598e-06,
+      "loss": 0.9468,
+      "step": 21947
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2757340368115146e-06,
+      "loss": 0.988,
+      "step": 21948
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2754149885182897e-06,
+      "loss": 0.9529,
+      "step": 21949
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.275096122337899e-06,
+      "loss": 0.9119,
+      "step": 21950
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2747774382763143e-06,
+      "loss": 0.9589,
+      "step": 21951
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.274458936339503e-06,
+      "loss": 0.9517,
+      "step": 21952
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2741406165334316e-06,
+      "loss": 0.9882,
+      "step": 21953
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2738224788640585e-06,
+      "loss": 1.035,
+      "step": 21954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2735045233373422e-06,
+      "loss": 0.9478,
+      "step": 21955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2731867499592375e-06,
+      "loss": 0.9466,
+      "step": 21956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.272869158735697e-06,
+      "loss": 0.9315,
+      "step": 21957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2725517496726647e-06,
+      "loss": 0.9127,
+      "step": 21958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.272234522776089e-06,
+      "loss": 0.9754,
+      "step": 21959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271917478051906e-06,
+      "loss": 0.9648,
+      "step": 21960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2716006155060567e-06,
+      "loss": 0.9341,
+      "step": 21961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2712839351444738e-06,
+      "loss": 0.9606,
+      "step": 21962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2709674369730862e-06,
+      "loss": 0.9802,
+      "step": 21963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2706511209978246e-06,
+      "loss": 0.9477,
+      "step": 21964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2703349872246086e-06,
+      "loss": 0.9707,
+      "step": 21965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.270019035659362e-06,
+      "loss": 0.9325,
+      "step": 21966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.269703266307998e-06,
+      "loss": 0.9575,
+      "step": 21967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2693876791764337e-06,
+      "loss": 0.9405,
+      "step": 21968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2690722742705762e-06,
+      "loss": 0.8975,
+      "step": 21969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2687570515963344e-06,
+      "loss": 0.9057,
+      "step": 21970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268442011159611e-06,
+      "loss": 0.9592,
+      "step": 21971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.268127152966303e-06,
+      "loss": 0.959,
+      "step": 21972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2678124770223112e-06,
+      "loss": 0.8972,
+      "step": 21973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2674979833335255e-06,
+      "loss": 1.012,
+      "step": 21974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.267183671905835e-06,
+      "loss": 1.0043,
+      "step": 21975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2668695427451282e-06,
+      "loss": 0.9157,
+      "step": 21976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2665555958572875e-06,
+      "loss": 0.9298,
+      "step": 21977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2662418312481894e-06,
+      "loss": 0.9207,
+      "step": 21978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2659282489237146e-06,
+      "loss": 0.9639,
+      "step": 21979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2656148488897315e-06,
+      "loss": 0.9225,
+      "step": 21980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2653016311521096e-06,
+      "loss": 0.9689,
+      "step": 21981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2649885957167162e-06,
+      "loss": 0.9504,
+      "step": 21982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264675742589412e-06,
+      "loss": 0.9697,
+      "step": 21983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264363071776058e-06,
+      "loss": 0.8963,
+      "step": 21984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.264050583282506e-06,
+      "loss": 0.9541,
+      "step": 21985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2637382771146123e-06,
+      "loss": 0.9867,
+      "step": 21986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.263426153278222e-06,
+      "loss": 1.0068,
+      "step": 21987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2631142117791818e-06,
+      "loss": 0.9216,
+      "step": 21988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.262802452623333e-06,
+      "loss": 0.9128,
+      "step": 21989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2624908758165143e-06,
+      "loss": 0.8893,
+      "step": 21990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2621794813645606e-06,
+      "loss": 1.0028,
+      "step": 21991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.261868269273303e-06,
+      "loss": 0.9229,
+      "step": 21992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.26155723954857e-06,
+      "loss": 0.9391,
+      "step": 21993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.261246392196186e-06,
+      "loss": 0.959,
+      "step": 21994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2609357272219725e-06,
+      "loss": 0.9498,
+      "step": 21995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.260625244631747e-06,
+      "loss": 0.8832,
+      "step": 21996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2603149444313255e-06,
+      "loss": 0.9795,
+      "step": 21997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2600048266265163e-06,
+      "loss": 0.9296,
+      "step": 21998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259694891223128e-06,
+      "loss": 0.9089,
+      "step": 21999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2593851382269663e-06,
+      "loss": 0.9105,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "eval_text_loss": 0.9234224557876587,
+      "eval_text_runtime": 1.5393,
+      "eval_text_samples_per_second": 48.074,
+      "eval_text_steps_per_second": 1.299,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "eval_svit_loss": 1.0712182521820068,
+      "eval_svit_runtime": 4.0434,
+      "eval_svit_samples_per_second": 168.423,
+      "eval_svit_steps_per_second": 2.72,
+      "step": 22000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2590755676438294e-06,
+      "loss": 0.9695,
+      "step": 22001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258766179479517e-06,
+      "loss": 0.9706,
+      "step": 22002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258456973739821e-06,
+      "loss": 0.9575,
+      "step": 22003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.258147950430535e-06,
+      "loss": 0.9255,
+      "step": 22004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.257839109557442e-06,
+      "loss": 0.9293,
+      "step": 22005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2575304511263286e-06,
+      "loss": 0.9364,
+      "step": 22006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2572219751429736e-06,
+      "loss": 0.888,
+      "step": 22007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2569136816131544e-06,
+      "loss": 0.9465,
+      "step": 22008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.256605570542646e-06,
+      "loss": 0.9082,
+      "step": 22009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2562976419372154e-06,
+      "loss": 0.9795,
+      "step": 22010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.255989895802632e-06,
+      "loss": 0.8924,
+      "step": 22011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2556823321446564e-06,
+      "loss": 0.9852,
+      "step": 22012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2553749509690493e-06,
+      "loss": 1.0007,
+      "step": 22013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.255067752281568e-06,
+      "loss": 0.9958,
+      "step": 22014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2547607360879655e-06,
+      "loss": 0.9003,
+      "step": 22015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2544539023939894e-06,
+      "loss": 0.9818,
+      "step": 22016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2541472512053887e-06,
+      "loss": 0.9494,
+      "step": 22017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.253840782527903e-06,
+      "loss": 0.95,
+      "step": 22018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2535344963672725e-06,
+      "loss": 0.9024,
+      "step": 22019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2532283927292343e-06,
+      "loss": 0.9027,
+      "step": 22020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.25292247161952e-06,
+      "loss": 0.9675,
+      "step": 22021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.252616733043859e-06,
+      "loss": 0.9237,
+      "step": 22022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2523111770079746e-06,
+      "loss": 0.9219,
+      "step": 22023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2520058035175932e-06,
+      "loss": 0.9156,
+      "step": 22024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251700612578429e-06,
+      "loss": 0.9414,
+      "step": 22025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2513956041962007e-06,
+      "loss": 0.9258,
+      "step": 22026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.251090778376619e-06,
+      "loss": 0.862,
+      "step": 22027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2507861351253913e-06,
+      "loss": 0.9604,
+      "step": 22028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2504816744482262e-06,
+      "loss": 0.9774,
+      "step": 22029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.25017739635082e-06,
+      "loss": 0.9234,
+      "step": 22030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2498733008388764e-06,
+      "loss": 1.0023,
+      "step": 22031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.249569387918086e-06,
+      "loss": 0.9256,
+      "step": 22032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.249265657594142e-06,
+      "loss": 0.9897,
+      "step": 22033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2489621098727333e-06,
+      "loss": 0.956,
+      "step": 22034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248658744759542e-06,
+      "loss": 0.9615,
+      "step": 22035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.248355562260253e-06,
+      "loss": 0.9466,
+      "step": 22036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2480525623805395e-06,
+      "loss": 0.8954,
+      "step": 22037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247749745126078e-06,
+      "loss": 0.9548,
+      "step": 22038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24744711050254e-06,
+      "loss": 0.8755,
+      "step": 22039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.247144658515591e-06,
+      "loss": 0.9149,
+      "step": 22040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.246842389170897e-06,
+      "loss": 0.9583,
+      "step": 22041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.24654030247412e-06,
+      "loss": 0.8829,
+      "step": 22042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2462383984309114e-06,
+      "loss": 1.016,
+      "step": 22043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245936677046932e-06,
+      "loss": 0.8946,
+      "step": 22044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2456351383278267e-06,
+      "loss": 0.9402,
+      "step": 22045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2453337822792454e-06,
+      "loss": 0.9996,
+      "step": 22046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245032608906832e-06,
+      "loss": 1.0387,
+      "step": 22047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244731618216223e-06,
+      "loss": 0.9516,
+      "step": 22048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2444308102130587e-06,
+      "loss": 1.0035,
+      "step": 22049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2441301849029698e-06,
+      "loss": 0.9261,
+      "step": 22050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.243829742291589e-06,
+      "loss": 0.8907,
+      "step": 22051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2435294823845413e-06,
+      "loss": 0.8722,
+      "step": 22052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.243229405187448e-06,
+      "loss": 0.8791,
+      "step": 22053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2429295107059315e-06,
+      "loss": 0.947,
+      "step": 22054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2426297989456062e-06,
+      "loss": 0.8925,
+      "step": 22055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2423302699120856e-06,
+      "loss": 0.9479,
+      "step": 22056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2420309236109775e-06,
+      "loss": 0.8933,
+      "step": 22057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2417317600478896e-06,
+      "loss": 0.8757,
+      "step": 22058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.241432779228423e-06,
+      "loss": 0.9344,
+      "step": 22059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.241133981158179e-06,
+      "loss": 0.9168,
+      "step": 22060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2408353658427496e-06,
+      "loss": 1.0262,
+      "step": 22061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2405369332877303e-06,
+      "loss": 0.9504,
+      "step": 22062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.240238683498708e-06,
+      "loss": 0.9162,
+      "step": 22063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2399406164812688e-06,
+      "loss": 1.026,
+      "step": 22064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2396427322409946e-06,
+      "loss": 0.9766,
+      "step": 22065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2393450307834622e-06,
+      "loss": 0.917,
+      "step": 22066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23904751211425e-06,
+      "loss": 0.8998,
+      "step": 22067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2387501762389266e-06,
+      "loss": 0.9045,
+      "step": 22068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238453023163062e-06,
+      "loss": 0.9095,
+      "step": 22069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.238156052892221e-06,
+      "loss": 0.9186,
+      "step": 22070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2378592654319632e-06,
+      "loss": 0.9842,
+      "step": 22071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2375626607878487e-06,
+      "loss": 0.9333,
+      "step": 22072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23726623896543e-06,
+      "loss": 0.8828,
+      "step": 22073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.236969999970261e-06,
+      "loss": 0.9797,
+      "step": 22074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2366739438078866e-06,
+      "loss": 0.943,
+      "step": 22075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2363780704838525e-06,
+      "loss": 0.9666,
+      "step": 22076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2360823800036983e-06,
+      "loss": 0.941,
+      "step": 22077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2357868723729627e-06,
+      "loss": 0.9129,
+      "step": 22078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23549154759718e-06,
+      "loss": 0.9105,
+      "step": 22079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.23519640568188e-06,
+      "loss": 0.9496,
+      "step": 22080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2349014466325883e-06,
+      "loss": 0.9887,
+      "step": 22081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.234606670454832e-06,
+      "loss": 0.9549,
+      "step": 22082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2343120771541283e-06,
+      "loss": 0.9872,
+      "step": 22083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.234017666735996e-06,
+      "loss": 0.9998,
+      "step": 22084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2337234392059496e-06,
+      "loss": 0.9489,
+      "step": 22085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2334293945694943e-06,
+      "loss": 0.9718,
+      "step": 22086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2331355328321417e-06,
+      "loss": 0.9481,
+      "step": 22087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2328418539993914e-06,
+      "loss": 0.9361,
+      "step": 22088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232548358076747e-06,
+      "loss": 0.947,
+      "step": 22089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232255045069701e-06,
+      "loss": 0.9799,
+      "step": 22090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231961914983748e-06,
+      "loss": 0.8975,
+      "step": 22091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.231668967824379e-06,
+      "loss": 0.9787,
+      "step": 22092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2313762035970766e-06,
+      "loss": 0.9419,
+      "step": 22093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2310836223073262e-06,
+      "loss": 0.9871,
+      "step": 22094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2307912239606056e-06,
+      "loss": 0.9689,
+      "step": 22095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2304990085623916e-06,
+      "loss": 0.9509,
+      "step": 22096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2302069761181545e-06,
+      "loss": 0.966,
+      "step": 22097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2299151266333656e-06,
+      "loss": 0.932,
+      "step": 22098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229623460113488e-06,
+      "loss": 1.0149,
+      "step": 22099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.229331976563988e-06,
+      "loss": 0.9045,
+      "step": 22100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2290406759903196e-06,
+      "loss": 0.9136,
+      "step": 22101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2287495583979394e-06,
+      "loss": 0.8816,
+      "step": 22102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2284586237923e-06,
+      "loss": 0.953,
+      "step": 22103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2281678721788493e-06,
+      "loss": 1.0106,
+      "step": 22104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227877303563033e-06,
+      "loss": 0.9199,
+      "step": 22105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2275869179502896e-06,
+      "loss": 0.9644,
+      "step": 22106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2272967153460615e-06,
+      "loss": 1.0092,
+      "step": 22107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227006695755779e-06,
+      "loss": 0.9108,
+      "step": 22108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.226716859184879e-06,
+      "loss": 0.9565,
+      "step": 22109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2264272056387826e-06,
+      "loss": 0.9901,
+      "step": 22110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.226137735122919e-06,
+      "loss": 0.9355,
+      "step": 22111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225848447642708e-06,
+      "loss": 0.9097,
+      "step": 22112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2255593432035652e-06,
+      "loss": 0.9216,
+      "step": 22113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.225270421810907e-06,
+      "loss": 0.8949,
+      "step": 22114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2249816834701423e-06,
+      "loss": 0.9516,
+      "step": 22115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2246931281866794e-06,
+      "loss": 0.9374,
+      "step": 22116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2244047559659223e-06,
+      "loss": 0.9302,
+      "step": 22117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2241165668132704e-06,
+      "loss": 0.9225,
+      "step": 22118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2238285607341205e-06,
+      "loss": 0.9039,
+      "step": 22119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.223540737733869e-06,
+      "loss": 0.9393,
+      "step": 22120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2232530978179026e-06,
+      "loss": 0.9646,
+      "step": 22121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.222965640991608e-06,
+      "loss": 0.9114,
+      "step": 22122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2226783672603715e-06,
+      "loss": 0.9734,
+      "step": 22123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2223912766295676e-06,
+      "loss": 0.977,
+      "step": 22124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2221043691045797e-06,
+      "loss": 0.923,
+      "step": 22125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2218176446907737e-06,
+      "loss": 0.9434,
+      "step": 22126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2215311033935253e-06,
+      "loss": 0.9089,
+      "step": 22127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2212447452181953e-06,
+      "loss": 0.8948,
+      "step": 22128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2209585701701497e-06,
+      "loss": 0.8938,
+      "step": 22129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.220672578254747e-06,
+      "loss": 0.9622,
+      "step": 22130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.220386769477342e-06,
+      "loss": 0.9353,
+      "step": 22131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.220101143843289e-06,
+      "loss": 0.9183,
+      "step": 22132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2198157013579337e-06,
+      "loss": 0.9292,
+      "step": 22133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2195304420266254e-06,
+      "loss": 0.9166,
+      "step": 22134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.219245365854702e-06,
+      "loss": 1.0095,
+      "step": 22135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.218960472847507e-06,
+      "loss": 0.9711,
+      "step": 22136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.218675763010372e-06,
+      "loss": 0.9721,
+      "step": 22137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2183912363486305e-06,
+      "loss": 0.9836,
+      "step": 22138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2181068928676087e-06,
+      "loss": 0.9475,
+      "step": 22139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2178227325726346e-06,
+      "loss": 0.9363,
+      "step": 22140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2175387554690276e-06,
+      "loss": 0.9622,
+      "step": 22141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2172549615621054e-06,
+      "loss": 0.9467,
+      "step": 22142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2169713508571856e-06,
+      "loss": 0.9968,
+      "step": 22143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.216687923359574e-06,
+      "loss": 0.9687,
+      "step": 22144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2164046790745844e-06,
+      "loss": 0.9719,
+      "step": 22145
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2161216180075175e-06,
+      "loss": 0.9199,
+      "step": 22146
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.215838740163675e-06,
+      "loss": 0.9646,
+      "step": 22147
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2155560455483553e-06,
+      "loss": 0.9618,
+      "step": 22148
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.21527353416685e-06,
+      "loss": 1.0072,
+      "step": 22149
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.214991206024453e-06,
+      "loss": 0.9301,
+      "step": 22150
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.214709061126448e-06,
+      "loss": 0.8733,
+      "step": 22151
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.214427099478122e-06,
+      "loss": 0.8323,
+      "step": 22152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2141453210847534e-06,
+      "loss": 0.8779,
+      "step": 22153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2138637259516183e-06,
+      "loss": 0.9579,
+      "step": 22154
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.213582314083993e-06,
+      "loss": 0.9203,
+      "step": 22155
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2133010854871452e-06,
+      "loss": 0.921,
+      "step": 22156
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.213020040166343e-06,
+      "loss": 0.9485,
+      "step": 22157
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.212739178126847e-06,
+      "loss": 0.9561,
+      "step": 22158
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2124584993739196e-06,
+      "loss": 0.9685,
+      "step": 22159
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.212178003912815e-06,
+      "loss": 0.9602,
+      "step": 22160
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211897691748788e-06,
+      "loss": 0.8973,
+      "step": 22161
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2116175628870875e-06,
+      "loss": 0.9366,
+      "step": 22162
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2113376173329596e-06,
+      "loss": 0.9168,
+      "step": 22163
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.211057855091645e-06,
+      "loss": 0.9743,
+      "step": 22164
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.210778276168385e-06,
+      "loss": 0.9819,
+      "step": 22165
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.210498880568415e-06,
+      "loss": 0.9546,
+      "step": 22166
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2102196682969657e-06,
+      "loss": 0.9424,
+      "step": 22167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2099406393592685e-06,
+      "loss": 0.9192,
+      "step": 22168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.209661793760546e-06,
+      "loss": 0.9815,
+      "step": 22169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2093831315060224e-06,
+      "loss": 0.9777,
+      "step": 22170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2091046526009132e-06,
+      "loss": 0.941,
+      "step": 22171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.208826357050438e-06,
+      "loss": 0.8898,
+      "step": 22172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.208548244859805e-06,
+      "loss": 0.9608,
+      "step": 22173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2082703160342235e-06,
+      "loss": 0.9429,
+      "step": 22174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2079925705788977e-06,
+      "loss": 0.9378,
+      "step": 22175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2077150084990297e-06,
+      "loss": 0.9014,
+      "step": 22176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.207437629799819e-06,
+      "loss": 0.8876,
+      "step": 22177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.207160434486456e-06,
+      "loss": 0.9283,
+      "step": 22178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2068834225641356e-06,
+      "loss": 0.9755,
+      "step": 22179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206606594038042e-06,
+      "loss": 0.9639,
+      "step": 22180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206329948913363e-06,
+      "loss": 0.9724,
+      "step": 22181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.206053487195277e-06,
+      "loss": 0.9504,
+      "step": 22182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2057772088889632e-06,
+      "loss": 0.9209,
+      "step": 22183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.205501113999592e-06,
+      "loss": 0.9191,
+      "step": 22184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2052252025323363e-06,
+      "loss": 0.8843,
+      "step": 22185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204949474492364e-06,
+      "loss": 0.949,
+      "step": 22186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2046739298848365e-06,
+      "loss": 0.9046,
+      "step": 22187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2043985687149165e-06,
+      "loss": 0.9777,
+      "step": 22188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.204123390987757e-06,
+      "loss": 0.9027,
+      "step": 22189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2038483967085157e-06,
+      "loss": 0.9635,
+      "step": 22190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2035735858823383e-06,
+      "loss": 0.9236,
+      "step": 22191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.203298958514374e-06,
+      "loss": 0.9681,
+      "step": 22192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2030245146097643e-06,
+      "loss": 0.921,
+      "step": 22193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.202750254173649e-06,
+      "loss": 0.8897,
+      "step": 22194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.202476177211166e-06,
+      "loss": 0.9523,
+      "step": 22195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2022022837274452e-06,
+      "loss": 0.9378,
+      "step": 22196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2019285737276164e-06,
+      "loss": 0.9763,
+      "step": 22197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2016550472168067e-06,
+      "loss": 0.9361,
+      "step": 22198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2013817042001373e-06,
+      "loss": 0.8871,
+      "step": 22199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2011085446827285e-06,
+      "loss": 0.9601,
+      "step": 22200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2008355686696948e-06,
+      "loss": 0.9565,
+      "step": 22201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2005627761661466e-06,
+      "loss": 0.8959,
+      "step": 22202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2002901671771963e-06,
+      "loss": 0.9298,
+      "step": 22203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2000177417079463e-06,
+      "loss": 0.9155,
+      "step": 22204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1997454997634984e-06,
+      "loss": 0.9672,
+      "step": 22205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1994734413489517e-06,
+      "loss": 0.9174,
+      "step": 22206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1992015664694003e-06,
+      "loss": 0.9529,
+      "step": 22207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1989298751299373e-06,
+      "loss": 1.0272,
+      "step": 22208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.198658367335648e-06,
+      "loss": 0.9237,
+      "step": 22209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.198387043091619e-06,
+      "loss": 0.9165,
+      "step": 22210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.198115902402931e-06,
+      "loss": 0.9143,
+      "step": 22211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197844945274661e-06,
+      "loss": 0.9296,
+      "step": 22212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1975741717118837e-06,
+      "loss": 0.9222,
+      "step": 22213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.197303581719671e-06,
+      "loss": 0.929,
+      "step": 22214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1970331753030896e-06,
+      "loss": 0.9214,
+      "step": 22215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196762952467201e-06,
+      "loss": 0.8756,
+      "step": 22216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.196492913217069e-06,
+      "loss": 0.9165,
+      "step": 22217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1962230575577483e-06,
+      "loss": 0.9003,
+      "step": 22218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195953385494294e-06,
+      "loss": 0.8968,
+      "step": 22219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195683897031755e-06,
+      "loss": 0.9321,
+      "step": 22220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1954145921751814e-06,
+      "loss": 0.9096,
+      "step": 22221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.195145470929612e-06,
+      "loss": 0.9226,
+      "step": 22222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194876533300088e-06,
+      "loss": 0.9038,
+      "step": 22223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1946077792916467e-06,
+      "loss": 0.9436,
+      "step": 22224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194339208909321e-06,
+      "loss": 0.9326,
+      "step": 22225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194070822158141e-06,
+      "loss": 0.9448,
+      "step": 22226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1938026190431296e-06,
+      "loss": 0.8949,
+      "step": 22227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.193534599569314e-06,
+      "loss": 1.0169,
+      "step": 22228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1932667637417095e-06,
+      "loss": 0.9535,
+      "step": 22229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192999111565335e-06,
+      "loss": 0.935,
+      "step": 22230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192731643045201e-06,
+      "loss": 0.9178,
+      "step": 22231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192464358186316e-06,
+      "loss": 0.9114,
+      "step": 22232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.192197256993688e-06,
+      "loss": 0.8941,
+      "step": 22233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.191930339472316e-06,
+      "loss": 0.9537,
+      "step": 22234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1916636056272e-06,
+      "loss": 0.9632,
+      "step": 22235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.191397055463335e-06,
+      "loss": 0.9665,
+      "step": 22236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1911306889857134e-06,
+      "loss": 0.9629,
+      "step": 22237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.190864506199322e-06,
+      "loss": 0.9141,
+      "step": 22238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1905985071091466e-06,
+      "loss": 0.9284,
+      "step": 22239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1903326917201674e-06,
+      "loss": 0.9202,
+      "step": 22240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1900670600373655e-06,
+      "loss": 0.9514,
+      "step": 22241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.189801612065712e-06,
+      "loss": 0.8915,
+      "step": 22242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1895363478101783e-06,
+      "loss": 0.966,
+      "step": 22243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1892712672757334e-06,
+      "loss": 1.0031,
+      "step": 22244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1890063704673417e-06,
+      "loss": 0.8253,
+      "step": 22245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1887416573899635e-06,
+      "loss": 0.8571,
+      "step": 22246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.188477128048554e-06,
+      "loss": 0.8946,
+      "step": 22247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1882127824480708e-06,
+      "loss": 0.9851,
+      "step": 22248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.187948620593461e-06,
+      "loss": 0.919,
+      "step": 22249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1876846424896725e-06,
+      "loss": 0.9859,
+      "step": 22250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1874208481416504e-06,
+      "loss": 0.9088,
+      "step": 22251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1871572375543337e-06,
+      "loss": 0.9442,
+      "step": 22252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.186893810732659e-06,
+      "loss": 0.9543,
+      "step": 22253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1866305676815583e-06,
+      "loss": 0.9147,
+      "step": 22254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1863675084059645e-06,
+      "loss": 0.9309,
+      "step": 22255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1861046329108005e-06,
+      "loss": 0.973,
+      "step": 22256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1858419412009915e-06,
+      "loss": 0.9469,
+      "step": 22257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185579433281455e-06,
+      "loss": 0.9106,
+      "step": 22258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1853171091571104e-06,
+      "loss": 0.9113,
+      "step": 22259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185054968832866e-06,
+      "loss": 0.9532,
+      "step": 22260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1847930123136333e-06,
+      "loss": 0.9764,
+      "step": 22261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1845312396043173e-06,
+      "loss": 0.9609,
+      "step": 22262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.184269650709822e-06,
+      "loss": 0.9158,
+      "step": 22263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1840082456350434e-06,
+      "loss": 0.9537,
+      "step": 22264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.183747024384878e-06,
+      "loss": 1.011,
+      "step": 22265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1834859869642192e-06,
+      "loss": 0.9395,
+      "step": 22266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.183225133377952e-06,
+      "loss": 0.8827,
+      "step": 22267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1829644636309655e-06,
+      "loss": 0.9493,
+      "step": 22268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.182703977728138e-06,
+      "loss": 0.8865,
+      "step": 22269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.18244367567435e-06,
+      "loss": 0.9474,
+      "step": 22270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1821835574744744e-06,
+      "loss": 0.9065,
+      "step": 22271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181923623133383e-06,
+      "loss": 0.9707,
+      "step": 22272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181663872655945e-06,
+      "loss": 0.8766,
+      "step": 22273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1814043060470228e-06,
+      "loss": 0.9894,
+      "step": 22274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.181144923311477e-06,
+      "loss": 0.9303,
+      "step": 22275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1808857244541673e-06,
+      "loss": 0.995,
+      "step": 22276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1806267094799464e-06,
+      "loss": 0.9072,
+      "step": 22277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1803678783936646e-06,
+      "loss": 0.959,
+      "step": 22278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1801092312001706e-06,
+      "loss": 0.9883,
+      "step": 22279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.179850767904307e-06,
+      "loss": 0.8847,
+      "step": 22280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1795924885109135e-06,
+      "loss": 0.9153,
+      "step": 22281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1793343930248267e-06,
+      "loss": 0.9036,
+      "step": 22282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1790764814508817e-06,
+      "loss": 0.943,
+      "step": 22283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178818753793908e-06,
+      "loss": 0.9721,
+      "step": 22284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1785612100587298e-06,
+      "loss": 0.9453,
+      "step": 22285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.178303850250173e-06,
+      "loss": 0.9188,
+      "step": 22286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1780466743730556e-06,
+      "loss": 0.874,
+      "step": 22287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1777896824321954e-06,
+      "loss": 0.9492,
+      "step": 22288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.177532874432402e-06,
+      "loss": 0.9534,
+      "step": 22289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1772762503784885e-06,
+      "loss": 0.9508,
+      "step": 22290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1770198102752583e-06,
+      "loss": 0.99,
+      "step": 22291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1767635541275134e-06,
+      "loss": 0.9219,
+      "step": 22292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1765074819400547e-06,
+      "loss": 0.9778,
+      "step": 22293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1762515937176756e-06,
+      "loss": 0.9861,
+      "step": 22294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175995889465169e-06,
+      "loss": 0.976,
+      "step": 22295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.175740369187324e-06,
+      "loss": 0.9343,
+      "step": 22296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1754850328889256e-06,
+      "loss": 0.9465,
+      "step": 22297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1752298805747554e-06,
+      "loss": 0.9731,
+      "step": 22298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1749749122495914e-06,
+      "loss": 0.8944,
+      "step": 22299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1747201279182074e-06,
+      "loss": 0.8933,
+      "step": 22300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.174465527585377e-06,
+      "loss": 0.8781,
+      "step": 22301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1742111112558663e-06,
+      "loss": 0.9079,
+      "step": 22302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1739568789344403e-06,
+      "loss": 0.9722,
+      "step": 22303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1737028306258616e-06,
+      "loss": 0.9115,
+      "step": 22304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1734489663348836e-06,
+      "loss": 0.9086,
+      "step": 22305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1731952860662654e-06,
+      "loss": 0.9742,
+      "step": 22306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1729417898247554e-06,
+      "loss": 0.883,
+      "step": 22307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1726884776150992e-06,
+      "loss": 0.9441,
+      "step": 22308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172435349442045e-06,
+      "loss": 0.9946,
+      "step": 22309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172182405310327e-06,
+      "loss": 0.9762,
+      "step": 22310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171929645224689e-06,
+      "loss": 0.9718,
+      "step": 22311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171677069189857e-06,
+      "loss": 0.9049,
+      "step": 22312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1714246772105687e-06,
+      "loss": 0.8736,
+      "step": 22313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171172469291545e-06,
+      "loss": 0.9421,
+      "step": 22314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1709204454375114e-06,
+      "loss": 0.8575,
+      "step": 22315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.170668605653188e-06,
+      "loss": 0.9671,
+      "step": 22316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1704169499432885e-06,
+      "loss": 0.9406,
+      "step": 22317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1701654783125284e-06,
+      "loss": 0.9126,
+      "step": 22318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1699141907656155e-06,
+      "loss": 0.895,
+      "step": 22319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1696630873072555e-06,
+      "loss": 0.9222,
+      "step": 22320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.169412167942151e-06,
+      "loss": 0.9822,
+      "step": 22321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1691614326750024e-06,
+      "loss": 0.9526,
+      "step": 22322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1689108815105043e-06,
+      "loss": 0.9499,
+      "step": 22323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168660514453349e-06,
+      "loss": 0.9207,
+      "step": 22324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.168410331508225e-06,
+      "loss": 0.8563,
+      "step": 22325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1681603326798163e-06,
+      "loss": 0.9124,
+      "step": 22326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.167910517972806e-06,
+      "loss": 0.9057,
+      "step": 22327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1676608873918714e-06,
+      "loss": 0.9455,
+      "step": 22328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1674114409416903e-06,
+      "loss": 0.9245,
+      "step": 22329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1671621786269293e-06,
+      "loss": 0.8915,
+      "step": 22330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1669131004522606e-06,
+      "loss": 0.9538,
+      "step": 22331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1666642064223466e-06,
+      "loss": 0.9765,
+      "step": 22332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1664154965418483e-06,
+      "loss": 0.9572,
+      "step": 22333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1661669708154233e-06,
+      "loss": 0.9123,
+      "step": 22334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1659186292477267e-06,
+      "loss": 0.968,
+      "step": 22335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1656704718434098e-06,
+      "loss": 0.9628,
+      "step": 22336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1654224986071174e-06,
+      "loss": 0.86,
+      "step": 22337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.165174709543494e-06,
+      "loss": 0.9038,
+      "step": 22338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164927104657182e-06,
+      "loss": 1.0088,
+      "step": 22339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1646796839528162e-06,
+      "loss": 0.9673,
+      "step": 22340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164432447435031e-06,
+      "loss": 0.9511,
+      "step": 22341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.164185395108457e-06,
+      "loss": 0.9305,
+      "step": 22342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.163938526977718e-06,
+      "loss": 0.9562,
+      "step": 22343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.163691843047441e-06,
+      "loss": 0.9667,
+      "step": 22344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.163445343322243e-06,
+      "loss": 0.8961,
+      "step": 22345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1631990278067407e-06,
+      "loss": 0.9174,
+      "step": 22346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162952896505547e-06,
+      "loss": 0.9229,
+      "step": 22347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1627069494232714e-06,
+      "loss": 0.9712,
+      "step": 22348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.162461186564521e-06,
+      "loss": 0.9352,
+      "step": 22349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1622156079338943e-06,
+      "loss": 0.9476,
+      "step": 22350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1619702135359945e-06,
+      "loss": 0.9256,
+      "step": 22351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1617250033754157e-06,
+      "loss": 0.9339,
+      "step": 22352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1614799774567486e-06,
+      "loss": 0.9259,
+      "step": 22353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1612351357845853e-06,
+      "loss": 0.9249,
+      "step": 22354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1609904783635064e-06,
+      "loss": 0.8823,
+      "step": 22355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1607460051980967e-06,
+      "loss": 0.9505,
+      "step": 22356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1605017162929334e-06,
+      "loss": 0.9271,
+      "step": 22357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160257611652591e-06,
+      "loss": 0.9677,
+      "step": 22358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1600136912816417e-06,
+      "loss": 0.922,
+      "step": 22359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.159769955184653e-06,
+      "loss": 0.9341,
+      "step": 22360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.15952640336619e-06,
+      "loss": 0.9584,
+      "step": 22361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1592830358308137e-06,
+      "loss": 0.9331,
+      "step": 22362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1590398525830803e-06,
+      "loss": 0.9571,
+      "step": 22363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158796853627544e-06,
+      "loss": 0.9065,
+      "step": 22364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1585540389687566e-06,
+      "loss": 0.9388,
+      "step": 22365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1583114086112656e-06,
+      "loss": 0.9297,
+      "step": 22366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1580689625596143e-06,
+      "loss": 0.9764,
+      "step": 22367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1578267008183408e-06,
+      "loss": 0.8847,
+      "step": 22368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1575846233919854e-06,
+      "loss": 0.9075,
+      "step": 22369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1573427302850794e-06,
+      "loss": 0.9277,
+      "step": 22370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1571010215021527e-06,
+      "loss": 0.9784,
+      "step": 22371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1568594970477333e-06,
+      "loss": 0.9062,
+      "step": 22372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1566181569263436e-06,
+      "loss": 0.9619,
+      "step": 22373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1563770011425037e-06,
+      "loss": 0.9228,
+      "step": 22374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1561360297007264e-06,
+      "loss": 0.943,
+      "step": 22375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1558952426055293e-06,
+      "loss": 1.0028,
+      "step": 22376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.155654639861418e-06,
+      "loss": 0.8893,
+      "step": 22377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1554142214728993e-06,
+      "loss": 0.9545,
+      "step": 22378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1551739874444766e-06,
+      "loss": 0.8847,
+      "step": 22379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.154933937780648e-06,
+      "loss": 0.954,
+      "step": 22380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.154694072485909e-06,
+      "loss": 0.9734,
+      "step": 22381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1544543915647513e-06,
+      "loss": 0.949,
+      "step": 22382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1542148950216637e-06,
+      "loss": 0.9156,
+      "step": 22383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.153975582861131e-06,
+      "loss": 1.0331,
+      "step": 22384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.153736455087634e-06,
+      "loss": 0.8874,
+      "step": 22385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1534975117056536e-06,
+      "loss": 0.996,
+      "step": 22386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1532587527196617e-06,
+      "loss": 0.9315,
+      "step": 22387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.15302017813413e-06,
+      "loss": 0.9444,
+      "step": 22388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1527817879535288e-06,
+      "loss": 0.9096,
+      "step": 22389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1525435821823184e-06,
+      "loss": 0.9768,
+      "step": 22390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1523055608249635e-06,
+      "loss": 1.0089,
+      "step": 22391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.152067723885919e-06,
+      "loss": 1.002,
+      "step": 22392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1518300713696395e-06,
+      "loss": 0.9777,
+      "step": 22393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.151592603280577e-06,
+      "loss": 0.9595,
+      "step": 22394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1513553196231754e-06,
+      "loss": 0.972,
+      "step": 22395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1511182204018825e-06,
+      "loss": 0.9267,
+      "step": 22396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.150881305621134e-06,
+      "loss": 0.9296,
+      "step": 22397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1506445752853703e-06,
+      "loss": 0.9119,
+      "step": 22398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.150408029399023e-06,
+      "loss": 0.936,
+      "step": 22399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.150171667966522e-06,
+      "loss": 0.9059,
+      "step": 22400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1499354909922946e-06,
+      "loss": 0.9696,
+      "step": 22401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.149699498480762e-06,
+      "loss": 0.9198,
+      "step": 22402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1494636904363445e-06,
+      "loss": 0.894,
+      "step": 22403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1492280668634582e-06,
+      "loss": 0.9724,
+      "step": 22404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1489926277665167e-06,
+      "loss": 0.9703,
+      "step": 22405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1487573731499256e-06,
+      "loss": 0.9869,
+      "step": 22406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1485223030180947e-06,
+      "loss": 0.9224,
+      "step": 22407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1482874173754235e-06,
+      "loss": 0.9176,
+      "step": 22408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.148052716226311e-06,
+      "loss": 0.9319,
+      "step": 22409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147818199575155e-06,
+      "loss": 0.9485,
+      "step": 22410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.147583867426343e-06,
+      "loss": 0.9625,
+      "step": 22411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1473497197842676e-06,
+      "loss": 0.9052,
+      "step": 22412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1471157566533107e-06,
+      "loss": 0.9388,
+      "step": 22413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.146881978037855e-06,
+      "loss": 0.9253,
+      "step": 22414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.146648383942278e-06,
+      "loss": 0.9975,
+      "step": 22415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1464149743709553e-06,
+      "loss": 1.001,
+      "step": 22416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1461817493282567e-06,
+      "loss": 0.9269,
+      "step": 22417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1459487088185508e-06,
+      "loss": 0.9426,
+      "step": 22418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.145715852846201e-06,
+      "loss": 0.9,
+      "step": 22419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1454831814155687e-06,
+      "loss": 0.9637,
+      "step": 22420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1452506945310093e-06,
+      "loss": 0.9248,
+      "step": 22421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1450183921968793e-06,
+      "loss": 0.959,
+      "step": 22422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1447862744175277e-06,
+      "loss": 1.043,
+      "step": 22423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1445543411973003e-06,
+      "loss": 0.9352,
+      "step": 22424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.144322592540543e-06,
+      "loss": 0.9189,
+      "step": 22425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1440910284515927e-06,
+      "loss": 0.9141,
+      "step": 22426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.14385964893479e-06,
+      "loss": 0.9227,
+      "step": 22427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1436284539944644e-06,
+      "loss": 0.9129,
+      "step": 22428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.143397443634947e-06,
+      "loss": 0.9414,
+      "step": 22429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.143166617860564e-06,
+      "loss": 0.9657,
+      "step": 22430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1429359766756366e-06,
+      "loss": 0.9329,
+      "step": 22431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142705520084486e-06,
+      "loss": 0.9462,
+      "step": 22432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.142475248091426e-06,
+      "loss": 1.0017,
+      "step": 22433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1422451607007723e-06,
+      "loss": 0.9329,
+      "step": 22434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.14201525791683e-06,
+      "loss": 0.9931,
+      "step": 22435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.141785539743906e-06,
+      "loss": 0.9553,
+      "step": 22436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1415560061863018e-06,
+      "loss": 0.9506,
+      "step": 22437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.141326657248317e-06,
+      "loss": 0.8981,
+      "step": 22438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1410974929342468e-06,
+      "loss": 0.8959,
+      "step": 22439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140868513248381e-06,
+      "loss": 0.8641,
+      "step": 22440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.140639718195009e-06,
+      "loss": 0.8954,
+      "step": 22441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1404111077784153e-06,
+      "loss": 0.8966,
+      "step": 22442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1401826820028807e-06,
+      "loss": 0.9873,
+      "step": 22443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1399544408726825e-06,
+      "loss": 0.8968,
+      "step": 22444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139726384392097e-06,
+      "loss": 0.8588,
+      "step": 22445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1394985125653925e-06,
+      "loss": 0.9398,
+      "step": 22446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139270825396838e-06,
+      "loss": 0.941,
+      "step": 22447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.139043322890697e-06,
+      "loss": 0.9376,
+      "step": 22448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.13881600505123e-06,
+      "loss": 0.9259,
+      "step": 22449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1385888718826944e-06,
+      "loss": 0.9296,
+      "step": 22450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.138361923389342e-06,
+      "loss": 0.9489,
+      "step": 22451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1381351595754256e-06,
+      "loss": 0.928,
+      "step": 22452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1379085804451883e-06,
+      "loss": 0.9512,
+      "step": 22453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1376821860028764e-06,
+      "loss": 0.8767,
+      "step": 22454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1374559762527295e-06,
+      "loss": 0.9335,
+      "step": 22455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1372299511989816e-06,
+      "loss": 0.9497,
+      "step": 22456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1370041108458673e-06,
+      "loss": 0.9518,
+      "step": 22457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1367784551976135e-06,
+      "loss": 0.9651,
+      "step": 22458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.136552984258451e-06,
+      "loss": 0.8802,
+      "step": 22459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1363276980325977e-06,
+      "loss": 0.9339,
+      "step": 22460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1361025965242734e-06,
+      "loss": 0.917,
+      "step": 22461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1358776797376947e-06,
+      "loss": 0.9068,
+      "step": 22462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1356529476770726e-06,
+      "loss": 0.9568,
+      "step": 22463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1354284003466165e-06,
+      "loss": 0.9448,
+      "step": 22464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1352040377505314e-06,
+      "loss": 0.9606,
+      "step": 22465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1349798598930186e-06,
+      "loss": 0.9619,
+      "step": 22466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1347558667782764e-06,
+      "loss": 0.9453,
+      "step": 22467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1345320584105002e-06,
+      "loss": 0.9484,
+      "step": 22468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1343084347938794e-06,
+      "loss": 0.9839,
+      "step": 22469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.134084995932604e-06,
+      "loss": 0.9302,
+      "step": 22470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1338617418308564e-06,
+      "loss": 0.9651,
+      "step": 22471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1336386724928193e-06,
+      "loss": 0.9272,
+      "step": 22472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133415787922669e-06,
+      "loss": 0.9184,
+      "step": 22473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.133193088124579e-06,
+      "loss": 0.874,
+      "step": 22474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.132970573102722e-06,
+      "loss": 0.9334,
+      "step": 22475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1327482428612625e-06,
+      "loss": 0.959,
+      "step": 22476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1325260974043664e-06,
+      "loss": 0.9568,
+      "step": 22477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.132304136736192e-06,
+      "loss": 0.8691,
+      "step": 22478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1320823608608958e-06,
+      "loss": 0.9528,
+      "step": 22479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1318607697826324e-06,
+      "loss": 0.8869,
+      "step": 22480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1316393635055514e-06,
+      "loss": 0.926,
+      "step": 22481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.131418142033798e-06,
+      "loss": 0.9422,
+      "step": 22482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1311971053715167e-06,
+      "loss": 0.9886,
+      "step": 22483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1309762535228452e-06,
+      "loss": 0.969,
+      "step": 22484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1307555864919208e-06,
+      "loss": 1.0076,
+      "step": 22485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.130535104282874e-06,
+      "loss": 0.9533,
+      "step": 22486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1303148068998347e-06,
+      "loss": 0.9241,
+      "step": 22487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1300946943469305e-06,
+      "loss": 0.9658,
+      "step": 22488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1298747666282794e-06,
+      "loss": 1.0171,
+      "step": 22489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1296550237480043e-06,
+      "loss": 0.977,
+      "step": 22490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129435465710217e-06,
+      "loss": 0.8894,
+      "step": 22491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.129216092519032e-06,
+      "loss": 0.8979,
+      "step": 22492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.128996904178554e-06,
+      "loss": 0.9253,
+      "step": 22493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1287779006928913e-06,
+      "loss": 0.9302,
+      "step": 22494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1285590820661442e-06,
+      "loss": 0.9531,
+      "step": 22495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1283404483024075e-06,
+      "loss": 0.9514,
+      "step": 22496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1281219994057803e-06,
+      "loss": 0.9433,
+      "step": 22497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12790373538035e-06,
+      "loss": 0.9441,
+      "step": 22498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1276856562302054e-06,
+      "loss": 0.9008,
+      "step": 22499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1274677619594298e-06,
+      "loss": 0.922,
+      "step": 22500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.127250052572105e-06,
+      "loss": 0.9568,
+      "step": 22501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.127032528072308e-06,
+      "loss": 0.9542,
+      "step": 22502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126815188464111e-06,
+      "loss": 0.9758,
+      "step": 22503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1265980337515847e-06,
+      "loss": 0.9122,
+      "step": 22504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1263810639387964e-06,
+      "loss": 0.9718,
+      "step": 22505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.126164279029809e-06,
+      "loss": 0.9393,
+      "step": 22506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1259476790286815e-06,
+      "loss": 0.9221,
+      "step": 22507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125731263939471e-06,
+      "loss": 0.9483,
+      "step": 22508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.125515033766229e-06,
+      "loss": 0.9426,
+      "step": 22509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1252989885130084e-06,
+      "loss": 0.9326,
+      "step": 22510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12508312818385e-06,
+      "loss": 0.8798,
+      "step": 22511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1248674527828006e-06,
+      "loss": 0.922,
+      "step": 22512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1246519623138975e-06,
+      "loss": 0.8783,
+      "step": 22513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.124436656781174e-06,
+      "loss": 0.996,
+      "step": 22514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1242215361886665e-06,
+      "loss": 0.8425,
+      "step": 22515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1240066005403995e-06,
+      "loss": 0.9514,
+      "step": 22516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.123791849840401e-06,
+      "loss": 0.9455,
+      "step": 22517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1235772840926917e-06,
+      "loss": 0.9651,
+      "step": 22518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1233629033012893e-06,
+      "loss": 0.9189,
+      "step": 22519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1231487074702086e-06,
+      "loss": 0.9122,
+      "step": 22520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1229346966034615e-06,
+      "loss": 0.9238,
+      "step": 22521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1227208707050564e-06,
+      "loss": 0.8831,
+      "step": 22522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1225072297789953e-06,
+      "loss": 0.906,
+      "step": 22523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1222937738292806e-06,
+      "loss": 0.9485,
+      "step": 22524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12208050285991e-06,
+      "loss": 0.945,
+      "step": 22525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1218674168748765e-06,
+      "loss": 0.9588,
+      "step": 22526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.121654515878171e-06,
+      "loss": 0.9761,
+      "step": 22527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1214417998737816e-06,
+      "loss": 0.9447,
+      "step": 22528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.12122926886569e-06,
+      "loss": 0.8582,
+      "step": 22529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1210169228578774e-06,
+      "loss": 0.9611,
+      "step": 22530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1208047618543194e-06,
+      "loss": 0.9719,
+      "step": 22531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1205927858589906e-06,
+      "loss": 0.9784,
+      "step": 22532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1203809948758604e-06,
+      "loss": 0.9107,
+      "step": 22533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1201693889088926e-06,
+      "loss": 0.9289,
+      "step": 22534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1199579679620534e-06,
+      "loss": 0.9315,
+      "step": 22535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1197467320392997e-06,
+      "loss": 0.9154,
+      "step": 22536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.119535681144589e-06,
+      "loss": 0.9764,
+      "step": 22537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1193248152818728e-06,
+      "loss": 0.9264,
+      "step": 22538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1191141344550994e-06,
+      "loss": 0.968,
+      "step": 22539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1189036386682163e-06,
+      "loss": 0.9407,
+      "step": 22540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1186933279251635e-06,
+      "loss": 0.8887,
+      "step": 22541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1184832022298796e-06,
+      "loss": 0.8782,
+      "step": 22542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1182732615863e-06,
+      "loss": 0.8955,
+      "step": 22543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.118063505998357e-06,
+      "loss": 0.9646,
+      "step": 22544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117853935469977e-06,
+      "loss": 0.9172,
+      "step": 22545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117644550005087e-06,
+      "loss": 0.924,
+      "step": 22546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1174353496076064e-06,
+      "loss": 0.8998,
+      "step": 22547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1172263342814536e-06,
+      "loss": 0.9485,
+      "step": 22548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.117017504030542e-06,
+      "loss": 0.957,
+      "step": 22549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1168088588587834e-06,
+      "loss": 0.9234,
+      "step": 22550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1166003987700857e-06,
+      "loss": 0.8371,
+      "step": 22551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1163921237683496e-06,
+      "loss": 0.9374,
+      "step": 22552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1161840338574804e-06,
+      "loss": 0.9558,
+      "step": 22553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.11597612904137e-06,
+      "loss": 0.9143,
+      "step": 22554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1157684093239153e-06,
+      "loss": 0.9123,
+      "step": 22555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.115560874709004e-06,
+      "loss": 0.9619,
+      "step": 22556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1153535252005244e-06,
+      "loss": 0.9486,
+      "step": 22557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1151463608023595e-06,
+      "loss": 0.9365,
+      "step": 22558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1149393815183874e-06,
+      "loss": 0.9525,
+      "step": 22559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114732587352485e-06,
+      "loss": 0.958,
+      "step": 22560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.114525978308526e-06,
+      "loss": 0.9621,
+      "step": 22561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1143195543903774e-06,
+      "loss": 0.9604,
+      "step": 22562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1141133156019058e-06,
+      "loss": 0.942,
+      "step": 22563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113907261946975e-06,
+      "loss": 0.9673,
+      "step": 22564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1137013934294415e-06,
+      "loss": 0.9206,
+      "step": 22565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113495710053163e-06,
+      "loss": 0.8763,
+      "step": 22566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113290211821988e-06,
+      "loss": 0.9288,
+      "step": 22567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.113084898739768e-06,
+      "loss": 0.9621,
+      "step": 22568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112879770810347e-06,
+      "loss": 0.9554,
+      "step": 22569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1126748280375654e-06,
+      "loss": 0.8495,
+      "step": 22570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1124700704252624e-06,
+      "loss": 0.9137,
+      "step": 22571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112265497977271e-06,
+      "loss": 0.9495,
+      "step": 22572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1120611106974243e-06,
+      "loss": 0.9368,
+      "step": 22573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111856908589548e-06,
+      "loss": 0.9372,
+      "step": 22574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1116528916574686e-06,
+      "loss": 0.9877,
+      "step": 22575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1114490599050043e-06,
+      "loss": 0.898,
+      "step": 22576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.111245413335974e-06,
+      "loss": 0.9543,
+      "step": 22577
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1110419519541903e-06,
+      "loss": 0.8789,
+      "step": 22578
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1108386757634625e-06,
+      "loss": 0.9031,
+      "step": 22579
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1106355847676e-06,
+      "loss": 0.8553,
+      "step": 22580
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1104326789704045e-06,
+      "loss": 0.938,
+      "step": 22581
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.110229958375675e-06,
+      "loss": 0.938,
+      "step": 22582
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.11002742298721e-06,
+      "loss": 0.9178,
+      "step": 22583
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1098250728088e-06,
+      "loss": 0.9603,
+      "step": 22584
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1096229078442368e-06,
+      "loss": 0.9473,
+      "step": 22585
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1094209280973044e-06,
+      "loss": 0.9209,
+      "step": 22586
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1092191335717874e-06,
+      "loss": 0.9465,
+      "step": 22587
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.109017524271462e-06,
+      "loss": 0.9663,
+      "step": 22588
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1088161002001064e-06,
+      "loss": 0.9164,
+      "step": 22589
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.10861486136149e-06,
+      "loss": 0.9561,
+      "step": 22590
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1084138077593856e-06,
+      "loss": 0.9947,
+      "step": 22591
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.108212939397553e-06,
+      "loss": 0.8729,
+      "step": 22592
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.108012256279758e-06,
+      "loss": 0.956,
+      "step": 22593
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107811758409757e-06,
+      "loss": 0.9329,
+      "step": 22594
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1076114457913057e-06,
+      "loss": 0.9153,
+      "step": 22595
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1074113184281545e-06,
+      "loss": 0.9357,
+      "step": 22596
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107211376324051e-06,
+      "loss": 0.9278,
+      "step": 22597
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.107011619482741e-06,
+      "loss": 0.9364,
+      "step": 22598
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1068120479079635e-06,
+      "loss": 0.9323,
+      "step": 22599
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1066126616034567e-06,
+      "loss": 0.9127,
+      "step": 22600
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.106413460572955e-06,
+      "loss": 0.9724,
+      "step": 22601
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1062144448201885e-06,
+      "loss": 0.9033,
+      "step": 22602
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1060156143488843e-06,
+      "loss": 0.9786,
+      "step": 22603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1058169691627653e-06,
+      "loss": 0.9029,
+      "step": 22604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1056185092655524e-06,
+      "loss": 0.9155,
+      "step": 22605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.105420234660961e-06,
+      "loss": 0.9363,
+      "step": 22606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1052221453527045e-06,
+      "loss": 0.952,
+      "step": 22607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1050242413444946e-06,
+      "loss": 0.9373,
+      "step": 22608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1048265226400346e-06,
+      "loss": 1.0349,
+      "step": 22609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1046289892430283e-06,
+      "loss": 1.0318,
+      "step": 22610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1044316411571767e-06,
+      "loss": 0.8948,
+      "step": 22611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1042344783861712e-06,
+      "loss": 0.9353,
+      "step": 22612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.104037500933709e-06,
+      "loss": 0.9551,
+      "step": 22613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1038407088034755e-06,
+      "loss": 0.8894,
+      "step": 22614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1036441019991574e-06,
+      "loss": 0.9984,
+      "step": 22615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1034476805244364e-06,
+      "loss": 1.0035,
+      "step": 22616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10325144438299e-06,
+      "loss": 0.944,
+      "step": 22617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.103055393578496e-06,
+      "loss": 0.8991,
+      "step": 22618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1028595281146226e-06,
+      "loss": 0.9397,
+      "step": 22619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1026638479950393e-06,
+      "loss": 0.8577,
+      "step": 22620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1024683532234092e-06,
+      "loss": 0.9109,
+      "step": 22621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.102273043803396e-06,
+      "loss": 0.9935,
+      "step": 22622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1020779197386545e-06,
+      "loss": 0.9535,
+      "step": 22623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1018829810328413e-06,
+      "loss": 0.9765,
+      "step": 22624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1016882276896046e-06,
+      "loss": 0.8841,
+      "step": 22625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1014936597125937e-06,
+      "loss": 0.9935,
+      "step": 22626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101299277105451e-06,
+      "loss": 0.9393,
+      "step": 22627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.101105079871816e-06,
+      "loss": 0.9247,
+      "step": 22628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100911068015328e-06,
+      "loss": 0.9671,
+      "step": 22629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1007172415396174e-06,
+      "loss": 0.9416,
+      "step": 22630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.100523600448317e-06,
+      "loss": 0.9657,
+      "step": 22631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.10033014474505e-06,
+      "loss": 0.9622,
+      "step": 22632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1001368744334405e-06,
+      "loss": 0.9251,
+      "step": 22633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0999437895171093e-06,
+      "loss": 0.8786,
+      "step": 22634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0997508899996702e-06,
+      "loss": 0.9529,
+      "step": 22635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0995581758847375e-06,
+      "loss": 0.9593,
+      "step": 22636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0993656471759183e-06,
+      "loss": 0.953,
+      "step": 22637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0991733038768193e-06,
+      "loss": 0.9779,
+      "step": 22638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0989811459910428e-06,
+      "loss": 0.9079,
+      "step": 22639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0987891735221858e-06,
+      "loss": 0.9813,
+      "step": 22640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0985973864738444e-06,
+      "loss": 0.8776,
+      "step": 22641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0984057848496112e-06,
+      "loss": 0.9833,
+      "step": 22642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0982143686530735e-06,
+      "loss": 0.949,
+      "step": 22643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.098023137887814e-06,
+      "loss": 0.9324,
+      "step": 22644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.097832092557417e-06,
+      "loss": 0.9742,
+      "step": 22645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0976412326654584e-06,
+      "loss": 0.9661,
+      "step": 22646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0974505582155125e-06,
+      "loss": 0.9804,
+      "step": 22647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0972600692111515e-06,
+      "loss": 0.9341,
+      "step": 22648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0970697656559416e-06,
+      "loss": 0.9772,
+      "step": 22649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096879647553445e-06,
+      "loss": 0.9202,
+      "step": 22650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096689714907225e-06,
+      "loss": 0.9142,
+      "step": 22651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096499967720838e-06,
+      "loss": 0.919,
+      "step": 22652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.096310405997836e-06,
+      "loss": 0.9526,
+      "step": 22653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09612102974177e-06,
+      "loss": 1.0199,
+      "step": 22654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0959318389561848e-06,
+      "loss": 0.9283,
+      "step": 22655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.095742833644625e-06,
+      "loss": 0.977,
+      "step": 22656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0955540138106303e-06,
+      "loss": 0.9797,
+      "step": 22657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0953653794577368e-06,
+      "loss": 1.0019,
+      "step": 22658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0951769305894755e-06,
+      "loss": 0.9888,
+      "step": 22659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094988667209376e-06,
+      "loss": 0.9262,
+      "step": 22660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0948005893209647e-06,
+      "loss": 0.9301,
+      "step": 22661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094612696927764e-06,
+      "loss": 0.9701,
+      "step": 22662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0944249900332914e-06,
+      "loss": 0.9413,
+      "step": 22663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.094237468641063e-06,
+      "loss": 0.9632,
+      "step": 22664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0940501327545907e-06,
+      "loss": 0.8679,
+      "step": 22665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0938629823773804e-06,
+      "loss": 0.9497,
+      "step": 22666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0936760175129404e-06,
+      "loss": 0.9492,
+      "step": 22667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09348923816477e-06,
+      "loss": 0.9432,
+      "step": 22668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.093302644336367e-06,
+      "loss": 0.9311,
+      "step": 22669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0931162360312267e-06,
+      "loss": 0.9988,
+      "step": 22670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0929300132528387e-06,
+      "loss": 0.9477,
+      "step": 22671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0927439760046923e-06,
+      "loss": 0.9226,
+      "step": 22672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0925581242902683e-06,
+      "loss": 0.9252,
+      "step": 22673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.09237245811305e-06,
+      "loss": 0.967,
+      "step": 22674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0921869774765125e-06,
+      "loss": 0.9123,
+      "step": 22675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.092001682384132e-06,
+      "loss": 0.899,
+      "step": 22676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0918165728393758e-06,
+      "loss": 0.9577,
+      "step": 22677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.091631648845711e-06,
+      "loss": 0.9657,
+      "step": 22678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.091446910406603e-06,
+      "loss": 0.954,
+      "step": 22679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0912623575255076e-06,
+      "loss": 0.8942,
+      "step": 22680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0910779902058842e-06,
+      "loss": 0.9781,
+      "step": 22681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0908938084511833e-06,
+      "loss": 0.9293,
+      "step": 22682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090709812264855e-06,
+      "loss": 0.9383,
+      "step": 22683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090526001650345e-06,
+      "loss": 0.8871,
+      "step": 22684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0903423766110954e-06,
+      "loss": 0.905,
+      "step": 22685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.090158937150544e-06,
+      "loss": 0.9795,
+      "step": 22686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0899756832721287e-06,
+      "loss": 0.8731,
+      "step": 22687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0897926149792793e-06,
+      "loss": 0.8916,
+      "step": 22688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0896097322754242e-06,
+      "loss": 0.9954,
+      "step": 22689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089427035163988e-06,
+      "loss": 0.9518,
+      "step": 22690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089244523648393e-06,
+      "loss": 0.9563,
+      "step": 22691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.089062197732058e-06,
+      "loss": 0.8534,
+      "step": 22692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0888800574183945e-06,
+      "loss": 0.905,
+      "step": 22693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0886981027108158e-06,
+      "loss": 0.8849,
+      "step": 22694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0885163336127282e-06,
+      "loss": 0.9528,
+      "step": 22695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088334750127538e-06,
+      "loss": 1.0089,
+      "step": 22696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.088153352258642e-06,
+      "loss": 0.9354,
+      "step": 22697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087972140009439e-06,
+      "loss": 0.9182,
+      "step": 22698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0877911133833243e-06,
+      "loss": 0.9782,
+      "step": 22699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0876102723836845e-06,
+      "loss": 0.9259,
+      "step": 22700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08742961701391e-06,
+      "loss": 0.9199,
+      "step": 22701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.087249147277382e-06,
+      "loss": 0.9466,
+      "step": 22702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0870688631774794e-06,
+      "loss": 0.9138,
+      "step": 22703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08688876471758e-06,
+      "loss": 0.9724,
+      "step": 22704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.086708851901054e-06,
+      "loss": 0.9141,
+      "step": 22705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0865291247312736e-06,
+      "loss": 0.8958,
+      "step": 22706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0863495832116043e-06,
+      "loss": 0.886,
+      "step": 22707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0861702273454066e-06,
+      "loss": 0.948,
+      "step": 22708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08599105713604e-06,
+      "loss": 0.8953,
+      "step": 22709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0858120725868606e-06,
+      "loss": 1.0302,
+      "step": 22710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0856332737012186e-06,
+      "loss": 0.8881,
+      "step": 22711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0854546604824645e-06,
+      "loss": 0.9538,
+      "step": 22712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0852762329339406e-06,
+      "loss": 0.9706,
+      "step": 22713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0850979910589922e-06,
+      "loss": 0.918,
+      "step": 22714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.084919934860953e-06,
+      "loss": 0.9386,
+      "step": 22715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0847420643431596e-06,
+      "loss": 0.947,
+      "step": 22716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0845643795089435e-06,
+      "loss": 0.8347,
+      "step": 22717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0843868803616307e-06,
+      "loss": 0.9658,
+      "step": 22718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0842095669045474e-06,
+      "loss": 0.9226,
+      "step": 22719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0840324391410106e-06,
+      "loss": 0.9696,
+      "step": 22720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083855497074341e-06,
+      "loss": 0.952,
+      "step": 22721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0836787407078498e-06,
+      "loss": 0.9771,
+      "step": 22722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0835021700448488e-06,
+      "loss": 0.9101,
+      "step": 22723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0833257850886436e-06,
+      "loss": 0.9702,
+      "step": 22724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.083149585842538e-06,
+      "loss": 0.9061,
+      "step": 22725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082973572309831e-06,
+      "loss": 0.9576,
+      "step": 22726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08279774449382e-06,
+      "loss": 0.9611,
+      "step": 22727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0826221023977956e-06,
+      "loss": 0.9509,
+      "step": 22728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0824466460250495e-06,
+      "loss": 0.9519,
+      "step": 22729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.082271375378866e-06,
+      "loss": 0.973,
+      "step": 22730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0820962904625277e-06,
+      "loss": 0.9,
+      "step": 22731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0819213912793135e-06,
+      "loss": 0.9649,
+      "step": 22732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081746677832498e-06,
+      "loss": 0.8878,
+      "step": 22733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081572150125355e-06,
+      "loss": 0.9048,
+      "step": 22734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.081397808161152e-06,
+      "loss": 0.9006,
+      "step": 22735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0812236519431517e-06,
+      "loss": 0.9707,
+      "step": 22736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08104968147462e-06,
+      "loss": 1.0008,
+      "step": 22737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0808758967588095e-06,
+      "loss": 0.9204,
+      "step": 22738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08070229779898e-06,
+      "loss": 0.9192,
+      "step": 22739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0805288845983788e-06,
+      "loss": 1.0359,
+      "step": 22740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.080355657160255e-06,
+      "loss": 0.9481,
+      "step": 22741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0801826154878513e-06,
+      "loss": 1.0613,
+      "step": 22742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0800097595844095e-06,
+      "loss": 0.8749,
+      "step": 22743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0798370894531674e-06,
+      "loss": 0.9499,
+      "step": 22744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079664605097357e-06,
+      "loss": 0.9377,
+      "step": 22745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.079492306520209e-06,
+      "loss": 0.964,
+      "step": 22746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0793201937249497e-06,
+      "loss": 0.9192,
+      "step": 22747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0791482667148023e-06,
+      "loss": 0.9438,
+      "step": 22748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0789765254929874e-06,
+      "loss": 0.939,
+      "step": 22749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0788049700627214e-06,
+      "loss": 0.951,
+      "step": 22750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078633600427215e-06,
+      "loss": 0.9664,
+      "step": 22751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0784624165896795e-06,
+      "loss": 0.8893,
+      "step": 22752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078291418553319e-06,
+      "loss": 0.9978,
+      "step": 22753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.078120606321337e-06,
+      "loss": 0.8986,
+      "step": 22754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077949979896932e-06,
+      "loss": 0.8221,
+      "step": 22755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077779539283299e-06,
+      "loss": 0.9464,
+      "step": 22756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0776092844836304e-06,
+      "loss": 0.9427,
+      "step": 22757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077439215501114e-06,
+      "loss": 0.9157,
+      "step": 22758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077269332338936e-06,
+      "loss": 0.9414,
+      "step": 22759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.077099635000276e-06,
+      "loss": 0.9359,
+      "step": 22760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0769301234883125e-06,
+      "loss": 0.9054,
+      "step": 22761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.07676079780622e-06,
+      "loss": 0.9595,
+      "step": 22762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0765916579571712e-06,
+      "loss": 0.9649,
+      "step": 22763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076422703944331e-06,
+      "loss": 0.9261,
+      "step": 22764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0762539357708646e-06,
+      "loss": 0.9514,
+      "step": 22765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076085353439932e-06,
+      "loss": 0.9327,
+      "step": 22766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0759169569546913e-06,
+      "loss": 0.9217,
+      "step": 22767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075748746318295e-06,
+      "loss": 0.9079,
+      "step": 22768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0755807215338933e-06,
+      "loss": 0.9139,
+      "step": 22769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075412882604634e-06,
+      "loss": 0.9735,
+      "step": 22770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.075245229533658e-06,
+      "loss": 0.8875,
+      "step": 22771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0750777623241073e-06,
+      "loss": 0.9617,
+      "step": 22772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074910480979117e-06,
+      "loss": 0.8419,
+      "step": 22773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0747433855018195e-06,
+      "loss": 0.9495,
+      "step": 22774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0745764758953446e-06,
+      "loss": 0.9718,
+      "step": 22775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0744097521628173e-06,
+      "loss": 0.9013,
+      "step": 22776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074243214307361e-06,
+      "loss": 0.9119,
+      "step": 22777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.074076862332093e-06,
+      "loss": 0.8908,
+      "step": 22778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0739106962401305e-06,
+      "loss": 0.9503,
+      "step": 22779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0737447160345837e-06,
+      "loss": 0.9548,
+      "step": 22780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073578921718561e-06,
+      "loss": 0.9304,
+      "step": 22781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0734133132951684e-06,
+      "loss": 0.9945,
+      "step": 22782
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073247890767506e-06,
+      "loss": 0.9209,
+      "step": 22783
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.073082654138672e-06,
+      "loss": 0.8556,
+      "step": 22784
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0729176034117606e-06,
+      "loss": 0.8982,
+      "step": 22785
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072752738589864e-06,
+      "loss": 0.942,
+      "step": 22786
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0725880596760685e-06,
+      "loss": 0.9146,
+      "step": 22787
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0724235666734573e-06,
+      "loss": 0.8914,
+      "step": 22788
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0722592595851123e-06,
+      "loss": 0.9515,
+      "step": 22789
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.072095138414111e-06,
+      "loss": 0.9043,
+      "step": 22790
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0719312031635254e-06,
+      "loss": 0.9361,
+      "step": 22791
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0717674538364253e-06,
+      "loss": 0.9253,
+      "step": 22792
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.071603890435878e-06,
+      "loss": 0.9662,
+      "step": 22793
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0714405129649473e-06,
+      "loss": 0.99,
+      "step": 22794
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0712773214266926e-06,
+      "loss": 0.9514,
+      "step": 22795
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0711143158241676e-06,
+      "loss": 0.9817,
+      "step": 22796
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0709514961604284e-06,
+      "loss": 0.9576,
+      "step": 22797
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070788862438521e-06,
+      "loss": 1.0073,
+      "step": 22798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0706264146614942e-06,
+      "loss": 0.9142,
+      "step": 22799
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070464152832387e-06,
+      "loss": 0.8998,
+      "step": 22800
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0703020769542404e-06,
+      "loss": 0.9469,
+      "step": 22801
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.070140187030089e-06,
+      "loss": 0.9211,
+      "step": 22802
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0699784830629634e-06,
+      "loss": 1.0127,
+      "step": 22803
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0698169650558933e-06,
+      "loss": 0.9305,
+      "step": 22804
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0696556330119022e-06,
+      "loss": 0.9081,
+      "step": 22805
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069494486934013e-06,
+      "loss": 0.8904,
+      "step": 22806
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0693335268252427e-06,
+      "loss": 0.9488,
+      "step": 22807
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.069172752688604e-06,
+      "loss": 0.9099,
+      "step": 22808
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06901216452711e-06,
+      "loss": 0.8987,
+      "step": 22809
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.068851762343768e-06,
+      "loss": 0.9474,
+      "step": 22810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0686915461415805e-06,
+      "loss": 0.9008,
+      "step": 22811
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0685315159235487e-06,
+      "loss": 0.9478,
+      "step": 22812
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0683716716926695e-06,
+      "loss": 0.8904,
+      "step": 22813
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0682120134519347e-06,
+      "loss": 0.8893,
+      "step": 22814
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0680525412043376e-06,
+      "loss": 0.9505,
+      "step": 22815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.067893254952861e-06,
+      "loss": 0.8395,
+      "step": 22816
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.067734154700489e-06,
+      "loss": 0.9192,
+      "step": 22817
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.067575240450204e-06,
+      "loss": 0.9523,
+      "step": 22818
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0674165122049774e-06,
+      "loss": 0.8829,
+      "step": 22819
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0672579699677855e-06,
+      "loss": 0.9539,
+      "step": 22820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0670996137415946e-06,
+      "loss": 0.9062,
+      "step": 22821
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0669414435293725e-06,
+      "loss": 0.8913,
+      "step": 22822
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0667834593340788e-06,
+      "loss": 0.9198,
+      "step": 22823
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0666256611586747e-06,
+      "loss": 0.9002,
+      "step": 22824
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066468049006113e-06,
+      "loss": 0.8974,
+      "step": 22825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066310622879346e-06,
+      "loss": 0.9224,
+      "step": 22826
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.066153382781323e-06,
+      "loss": 0.8971,
+      "step": 22827
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065996328714989e-06,
+      "loss": 0.9479,
+      "step": 22828
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065839460683282e-06,
+      "loss": 0.9641,
+      "step": 22829
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065682778689142e-06,
+      "loss": 0.9572,
+      "step": 22830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065526282735503e-06,
+      "loss": 0.9698,
+      "step": 22831
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065369972825295e-06,
+      "loss": 0.9526,
+      "step": 22832
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065213848961447e-06,
+      "loss": 0.8864,
+      "step": 22833
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.065057911146881e-06,
+      "loss": 0.975,
+      "step": 22834
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0649021593845176e-06,
+      "loss": 0.9379,
+      "step": 22835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.064746593677274e-06,
+      "loss": 1.0017,
+      "step": 22836
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0645912140280633e-06,
+      "loss": 0.9869,
+      "step": 22837
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0644360204397942e-06,
+      "loss": 0.9295,
+      "step": 22838
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0642810129153746e-06,
+      "loss": 0.9077,
+      "step": 22839
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0641261914577075e-06,
+      "loss": 0.936,
+      "step": 22840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0639715560696907e-06,
+      "loss": 0.9091,
+      "step": 22841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0638171067542215e-06,
+      "loss": 0.9177,
+      "step": 22842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0636628435141916e-06,
+      "loss": 0.9531,
+      "step": 22843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063508766352489e-06,
+      "loss": 0.9624,
+      "step": 22844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0633548752720012e-06,
+      "loss": 0.9342,
+      "step": 22845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0632011702756086e-06,
+      "loss": 0.9806,
+      "step": 22846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.063047651366189e-06,
+      "loss": 0.937,
+      "step": 22847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.06289431854662e-06,
+      "loss": 0.9343,
+      "step": 22848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0627411718197713e-06,
+      "loss": 0.9348,
+      "step": 22849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0625882111885102e-06,
+      "loss": 0.9012,
+      "step": 22850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0624354366557023e-06,
+      "loss": 0.9575,
+      "step": 22851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0622828482242085e-06,
+      "loss": 0.9193,
+      "step": 22852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0621304458968875e-06,
+      "loss": 0.8908,
+      "step": 22853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.06197822967659e-06,
+      "loss": 0.9681,
+      "step": 22854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0618261995661693e-06,
+      "loss": 0.9562,
+      "step": 22855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0616743555684717e-06,
+      "loss": 0.8888,
+      "step": 22856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061522697686341e-06,
+      "loss": 0.9219,
+      "step": 22857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0613712259226183e-06,
+      "loss": 0.9856,
+      "step": 22858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0612199402801373e-06,
+      "loss": 0.9584,
+      "step": 22859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0610688407617343e-06,
+      "loss": 0.9088,
+      "step": 22860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0609179273702364e-06,
+      "loss": 0.9911,
+      "step": 22861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0607672001084726e-06,
+      "loss": 0.9146,
+      "step": 22862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060616658979263e-06,
+      "loss": 0.8959,
+      "step": 22863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060466303985427e-06,
+      "loss": 0.9799,
+      "step": 22864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060316135129782e-06,
+      "loss": 0.9231,
+      "step": 22865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0601661524151387e-06,
+      "loss": 0.9453,
+      "step": 22866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.060016355844308e-06,
+      "loss": 0.9827,
+      "step": 22867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059866745420092e-06,
+      "loss": 0.9612,
+      "step": 22868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0597173211452937e-06,
+      "loss": 0.8622,
+      "step": 22869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0595680830227116e-06,
+      "loss": 0.9784,
+      "step": 22870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0594190310551412e-06,
+      "loss": 0.9902,
+      "step": 22871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.059270165245373e-06,
+      "loss": 0.9476,
+      "step": 22872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0591214855961962e-06,
+      "loss": 0.9783,
+      "step": 22873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058972992110393e-06,
+      "loss": 0.9681,
+      "step": 22874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0588246847907445e-06,
+      "loss": 0.9835,
+      "step": 22875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0586765636400288e-06,
+      "loss": 0.9512,
+      "step": 22876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0585286286610206e-06,
+      "loss": 0.9058,
+      "step": 22877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0583808798564886e-06,
+      "loss": 0.9665,
+      "step": 22878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0582333172292013e-06,
+      "loss": 0.9106,
+      "step": 22879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.058085940781921e-06,
+      "loss": 0.9213,
+      "step": 22880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057938750517408e-06,
+      "loss": 0.9207,
+      "step": 22881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057791746438418e-06,
+      "loss": 0.9093,
+      "step": 22882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0576449285477054e-06,
+      "loss": 0.9595,
+      "step": 22883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0574982968480176e-06,
+      "loss": 0.9065,
+      "step": 22884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0573518513421034e-06,
+      "loss": 0.9134,
+      "step": 22885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057205592032703e-06,
+      "loss": 0.9502,
+      "step": 22886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.057059518922556e-06,
+      "loss": 0.8988,
+      "step": 22887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056913632014399e-06,
+      "loss": 0.937,
+      "step": 22888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0567679313109625e-06,
+      "loss": 0.9449,
+      "step": 22889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0566224168149752e-06,
+      "loss": 0.963,
+      "step": 22890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056477088529163e-06,
+      "loss": 0.9418,
+      "step": 22891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0563319464562473e-06,
+      "loss": 0.959,
+      "step": 22892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0561869905989462e-06,
+      "loss": 1.0158,
+      "step": 22893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.056042220959973e-06,
+      "loss": 0.9427,
+      "step": 22894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0558976375420417e-06,
+      "loss": 0.935,
+      "step": 22895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0557532403478566e-06,
+      "loss": 0.9168,
+      "step": 22896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0556090293801237e-06,
+      "loss": 0.9495,
+      "step": 22897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0554650046415438e-06,
+      "loss": 0.9398,
+      "step": 22898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0553211661348116e-06,
+      "loss": 0.9513,
+      "step": 22899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0551775138626254e-06,
+      "loss": 0.8317,
+      "step": 22900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0550340478276712e-06,
+      "loss": 0.8933,
+      "step": 22901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054890768032637e-06,
+      "loss": 0.8334,
+      "step": 22902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0547476744802066e-06,
+      "loss": 0.8891,
+      "step": 22903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0546047671730597e-06,
+      "loss": 0.939,
+      "step": 22904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0544620461138723e-06,
+      "loss": 0.9558,
+      "step": 22905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054319511305316e-06,
+      "loss": 0.889,
+      "step": 22906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0541771627500618e-06,
+      "loss": 0.9221,
+      "step": 22907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054035000450775e-06,
+      "loss": 0.94,
+      "step": 22908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0538930244101165e-06,
+      "loss": 0.9551,
+      "step": 22909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0537512346307465e-06,
+      "loss": 0.8751,
+      "step": 22910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053609631115321e-06,
+      "loss": 0.995,
+      "step": 22911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05346821386649e-06,
+      "loss": 0.9641,
+      "step": 22912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.053326982886903e-06,
+      "loss": 0.9855,
+      "step": 22913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0531859381792043e-06,
+      "loss": 0.922,
+      "step": 22914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0530450797460347e-06,
+      "loss": 0.9421,
+      "step": 22915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052904407590034e-06,
+      "loss": 0.9215,
+      "step": 22916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052763921713834e-06,
+      "loss": 0.8937,
+      "step": 22917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0526236221200694e-06,
+      "loss": 0.9377,
+      "step": 22918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.052483508811363e-06,
+      "loss": 0.938,
+      "step": 22919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0523435817903417e-06,
+      "loss": 0.9001,
+      "step": 22920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0522038410596246e-06,
+      "loss": 0.9144,
+      "step": 22921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0520642866218302e-06,
+      "loss": 0.9919,
+      "step": 22922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.05192491847957e-06,
+      "loss": 0.9423,
+      "step": 22923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0517857366354546e-06,
+      "loss": 0.9111,
+      "step": 22924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0516467410920907e-06,
+      "loss": 0.9302,
+      "step": 22925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0515079318520815e-06,
+      "loss": 1.0006,
+      "step": 22926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0513693089180266e-06,
+      "loss": 0.9299,
+      "step": 22927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0512308722925213e-06,
+      "loss": 0.9261,
+      "step": 22928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.051092621978159e-06,
+      "loss": 1.0488,
+      "step": 22929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050954557977527e-06,
+      "loss": 0.9104,
+      "step": 22930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0508166802932143e-06,
+      "loss": 0.903,
+      "step": 22931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050678988927798e-06,
+      "loss": 0.9305,
+      "step": 22932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0505414838838614e-06,
+      "loss": 0.9124,
+      "step": 22933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0504041651639763e-06,
+      "loss": 0.9125,
+      "step": 22934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0502670327707156e-06,
+      "loss": 0.9764,
+      "step": 22935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.050130086706648e-06,
+      "loss": 0.8766,
+      "step": 22936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0499933269743356e-06,
+      "loss": 0.8918,
+      "step": 22937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0498567535763426e-06,
+      "loss": 0.954,
+      "step": 22938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0497203665152256e-06,
+      "loss": 0.9326,
+      "step": 22939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0495841657935368e-06,
+      "loss": 0.8827,
+      "step": 22940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.04944815141383e-06,
+      "loss": 0.9546,
+      "step": 22941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0493123233786495e-06,
+      "loss": 0.9074,
+      "step": 22942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0491766816905426e-06,
+      "loss": 0.9871,
+      "step": 22943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0490412263520443e-06,
+      "loss": 0.9463,
+      "step": 22944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0489059573656958e-06,
+      "loss": 0.9488,
+      "step": 22945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0487708747340275e-06,
+      "loss": 0.9024,
+      "step": 22946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0486359784595713e-06,
+      "loss": 0.886,
+      "step": 22947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0485012685448513e-06,
+      "loss": 0.961,
+      "step": 22948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0483667449923916e-06,
+      "loss": 0.9191,
+      "step": 22949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0482324078047114e-06,
+      "loss": 0.9364,
+      "step": 22950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0480982569843255e-06,
+      "loss": 0.9256,
+      "step": 22951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0479642925337474e-06,
+      "loss": 0.9517,
+      "step": 22952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047830514455485e-06,
+      "loss": 0.9351,
+      "step": 22953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047696922752043e-06,
+      "loss": 0.893,
+      "step": 22954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0475635174259235e-06,
+      "loss": 0.9255,
+      "step": 22955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.047430298479627e-06,
+      "loss": 0.9228,
+      "step": 22956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0472972659156456e-06,
+      "loss": 0.9134,
+      "step": 22957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0471644197364717e-06,
+      "loss": 0.8799,
+      "step": 22958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0470317599445923e-06,
+      "loss": 0.95,
+      "step": 22959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0468992865424933e-06,
+      "loss": 0.9615,
+      "step": 22960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046766999532655e-06,
+      "loss": 0.9721,
+      "step": 22961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0466348989175524e-06,
+      "loss": 0.8587,
+      "step": 22962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0465029846996632e-06,
+      "loss": 0.8821,
+      "step": 22963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0463712568814546e-06,
+      "loss": 0.9483,
+      "step": 22964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046239715465396e-06,
+      "loss": 0.886,
+      "step": 22965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.046108360453948e-06,
+      "loss": 0.9797,
+      "step": 22966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045977191849573e-06,
+      "loss": 0.8989,
+      "step": 22967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045846209654726e-06,
+      "loss": 0.9665,
+      "step": 22968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0457154138718617e-06,
+      "loss": 0.9202,
+      "step": 22969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0455848045034267e-06,
+      "loss": 0.9254,
+      "step": 22970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0454543815518692e-06,
+      "loss": 0.9268,
+      "step": 22971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0453241450196295e-06,
+      "loss": 0.9712,
+      "step": 22972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0451940949091487e-06,
+      "loss": 0.9795,
+      "step": 22973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.045064231222862e-06,
+      "loss": 0.9109,
+      "step": 22974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0449345539632005e-06,
+      "loss": 0.9534,
+      "step": 22975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0448050631325926e-06,
+      "loss": 0.9889,
+      "step": 22976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0446757587334632e-06,
+      "loss": 0.9499,
+      "step": 22977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044546640768236e-06,
+      "loss": 0.9458,
+      "step": 22978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.044417709239326e-06,
+      "loss": 0.9797,
+      "step": 22979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0442889641491484e-06,
+      "loss": 0.8638,
+      "step": 22980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0441604055001155e-06,
+      "loss": 0.966,
+      "step": 22981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0440320332946344e-06,
+      "loss": 0.9169,
+      "step": 22982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043903847535108e-06,
+      "loss": 0.8897,
+      "step": 22983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043775848223938e-06,
+      "loss": 0.9102,
+      "step": 22984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0436480353635216e-06,
+      "loss": 0.9389,
+      "step": 22985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0435204089562514e-06,
+      "loss": 0.8412,
+      "step": 22986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0433929690045177e-06,
+      "loss": 0.8815,
+      "step": 22987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0432657155107073e-06,
+      "loss": 0.9626,
+      "step": 22988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0431386484772044e-06,
+      "loss": 0.9152,
+      "step": 22989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.043011767906387e-06,
+      "loss": 0.9303,
+      "step": 22990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0428850738006303e-06,
+      "loss": 0.9394,
+      "step": 22991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0427585661623097e-06,
+      "loss": 0.9296,
+      "step": 22992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0426322449937922e-06,
+      "loss": 0.9167,
+      "step": 22993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0425061102974456e-06,
+      "loss": 0.9013,
+      "step": 22994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042380162075628e-06,
+      "loss": 0.8789,
+      "step": 22995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0422544003307025e-06,
+      "loss": 0.937,
+      "step": 22996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.042128825065022e-06,
+      "loss": 0.8734,
+      "step": 22997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0420034362809382e-06,
+      "loss": 0.9267,
+      "step": 22998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0418782339808e-06,
+      "loss": 0.9401,
+      "step": 22999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041753218166951e-06,
+      "loss": 0.902,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "eval_text_loss": 0.9234684109687805,
+      "eval_text_runtime": 1.5468,
+      "eval_text_samples_per_second": 47.839,
+      "eval_text_steps_per_second": 1.293,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "eval_svit_loss": 1.0713400840759277,
+      "eval_svit_runtime": 4.1428,
+      "eval_svit_samples_per_second": 164.382,
+      "eval_svit_steps_per_second": 2.655,
+      "step": 23000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0416283888417345e-06,
+      "loss": 0.9699,
+      "step": 23001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041503746007485e-06,
+      "loss": 0.9071,
+      "step": 23002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041379289666539e-06,
+      "loss": 0.9277,
+      "step": 23003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0412550198212266e-06,
+      "loss": 0.9329,
+      "step": 23004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0411309364738754e-06,
+      "loss": 0.8862,
+      "step": 23005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.041007039626808e-06,
+      "loss": 0.9018,
+      "step": 23006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040883329282345e-06,
+      "loss": 0.9237,
+      "step": 23007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040759805442805e-06,
+      "loss": 0.8778,
+      "step": 23008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0406364681104986e-06,
+      "loss": 0.9477,
+      "step": 23009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.040513317287737e-06,
+      "loss": 0.9607,
+      "step": 23010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0403903529768256e-06,
+      "loss": 0.9514,
+      "step": 23011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0402675751800686e-06,
+      "loss": 0.9384,
+      "step": 23012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0401449838997637e-06,
+      "loss": 0.9376,
+      "step": 23013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0400225791382068e-06,
+      "loss": 0.9078,
+      "step": 23014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.039900360897692e-06,
+      "loss": 0.9555,
+      "step": 23015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0397783291805053e-06,
+      "loss": 0.9293,
+      "step": 23016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0396564839889346e-06,
+      "loss": 0.9214,
+      "step": 23017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0395348253252604e-06,
+      "loss": 0.9052,
+      "step": 23018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0394133531917606e-06,
+      "loss": 0.9221,
+      "step": 23019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0392920675907104e-06,
+      "loss": 0.9214,
+      "step": 23020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0391709685243825e-06,
+      "loss": 0.9394,
+      "step": 23021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0390500559950434e-06,
+      "loss": 0.9021,
+      "step": 23022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0389293300049564e-06,
+      "loss": 0.941,
+      "step": 23023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038808790556385e-06,
+      "loss": 0.8914,
+      "step": 23024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038688437651584e-06,
+      "loss": 0.931,
+      "step": 23025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038568271292808e-06,
+      "loss": 0.9269,
+      "step": 23026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0384482914823092e-06,
+      "loss": 0.8811,
+      "step": 23027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038328498222332e-06,
+      "loss": 0.959,
+      "step": 23028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03820889151512e-06,
+      "loss": 0.926,
+      "step": 23029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.038089471362915e-06,
+      "loss": 0.996,
+      "step": 23030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037970237767951e-06,
+      "loss": 0.957,
+      "step": 23031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037851190732464e-06,
+      "loss": 0.8559,
+      "step": 23032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0377323302586794e-06,
+      "loss": 0.9721,
+      "step": 23033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037613656348827e-06,
+      "loss": 0.9462,
+      "step": 23034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037495169005126e-06,
+      "loss": 0.9914,
+      "step": 23035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037376868229798e-06,
+      "loss": 0.9573,
+      "step": 23036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0372587540250566e-06,
+      "loss": 0.8767,
+      "step": 23037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.037140826393114e-06,
+      "loss": 0.9321,
+      "step": 23038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0370230853361797e-06,
+      "loss": 0.8933,
+      "step": 23039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036905530856457e-06,
+      "loss": 0.8988,
+      "step": 23040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0367881629561483e-06,
+      "loss": 0.9796,
+      "step": 23041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036670981637452e-06,
+      "loss": 0.9471,
+      "step": 23042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036553986902561e-06,
+      "loss": 0.8823,
+      "step": 23043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036437178753669e-06,
+      "loss": 0.8793,
+      "step": 23044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0363205571929603e-06,
+      "loss": 0.9715,
+      "step": 23045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0362041222226206e-06,
+      "loss": 0.9734,
+      "step": 23046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.036087873844831e-06,
+      "loss": 0.8683,
+      "step": 23047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035971812061767e-06,
+      "loss": 0.9149,
+      "step": 23048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0358559368756018e-06,
+      "loss": 0.9955,
+      "step": 23049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035740248288507e-06,
+      "loss": 1.0019,
+      "step": 23050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0356247463026493e-06,
+      "loss": 0.9804,
+      "step": 23051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0355094309201907e-06,
+      "loss": 0.9238,
+      "step": 23052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.03539430214329e-06,
+      "loss": 0.8794,
+      "step": 23053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035279359974104e-06,
+      "loss": 0.9191,
+      "step": 23054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0351646044147856e-06,
+      "loss": 0.9927,
+      "step": 23055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.035050035467484e-06,
+      "loss": 0.8844,
+      "step": 23056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0349356531343448e-06,
+      "loss": 0.965,
+      "step": 23057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0348214574175074e-06,
+      "loss": 0.9863,
+      "step": 23058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0347074483191143e-06,
+      "loss": 0.9302,
+      "step": 23059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034593625841298e-06,
+      "loss": 0.934,
+      "step": 23060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0344799899861915e-06,
+      "loss": 0.9915,
+      "step": 23061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034366540755921e-06,
+      "loss": 0.9507,
+      "step": 23062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034253278152614e-06,
+      "loss": 0.9533,
+      "step": 23063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0341402021783886e-06,
+      "loss": 0.9148,
+      "step": 23064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.034027312835363e-06,
+      "loss": 0.9413,
+      "step": 23065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0339146101256527e-06,
+      "loss": 0.9693,
+      "step": 23066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.033802094051367e-06,
+      "loss": 1.0005,
+      "step": 23067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0336897646146137e-06,
+      "loss": 0.9658,
+      "step": 23068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0335776218174964e-06,
+      "loss": 0.8965,
+      "step": 23069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0334656656621146e-06,
+      "loss": 0.9076,
+      "step": 23070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0333538961505664e-06,
+      "loss": 0.9112,
+      "step": 23071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0332423132849422e-06,
+      "loss": 0.9351,
+      "step": 23072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0331309170673343e-06,
+      "loss": 0.9267,
+      "step": 23073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0330197074998272e-06,
+      "loss": 0.919,
+      "step": 23074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0329086845845038e-06,
+      "loss": 0.9354,
+      "step": 23075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0327978483234434e-06,
+      "loss": 0.9199,
+      "step": 23076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0326871987187227e-06,
+      "loss": 0.9441,
+      "step": 23077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0325767357724117e-06,
+      "loss": 0.843,
+      "step": 23078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0324664594865806e-06,
+      "loss": 0.9282,
+      "step": 23079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0323563698632945e-06,
+      "loss": 0.9413,
+      "step": 23080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.032246466904614e-06,
+      "loss": 0.916,
+      "step": 23081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.032136750612599e-06,
+      "loss": 0.9176,
+      "step": 23082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0320272209893026e-06,
+      "loss": 0.9332,
+      "step": 23083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0319178780367767e-06,
+      "loss": 0.9576,
+      "step": 23084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0318087217570683e-06,
+      "loss": 0.9471,
+      "step": 23085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0316997521522223e-06,
+      "loss": 0.9155,
+      "step": 23086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0315909692242786e-06,
+      "loss": 0.913,
+      "step": 23087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0314823729752754e-06,
+      "loss": 0.9044,
+      "step": 23088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031373963407246e-06,
+      "loss": 0.9417,
+      "step": 23089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0312657405222205e-06,
+      "loss": 0.945,
+      "step": 23090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0311577043222253e-06,
+      "loss": 1.03,
+      "step": 23091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.031049854809284e-06,
+      "loss": 0.9702,
+      "step": 23092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030942191985416e-06,
+      "loss": 0.9801,
+      "step": 23093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0308347158526377e-06,
+      "loss": 0.9455,
+      "step": 23094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030727426412962e-06,
+      "loss": 0.9582,
+      "step": 23095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0306203236683977e-06,
+      "loss": 1.0121,
+      "step": 23096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0305134076209513e-06,
+      "loss": 0.9243,
+      "step": 23097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0304066782726243e-06,
+      "loss": 0.9824,
+      "step": 23098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.030300135625415e-06,
+      "loss": 0.9598,
+      "step": 23099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.03019377968132e-06,
+      "loss": 0.9289,
+      "step": 23100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0300876104423295e-06,
+      "loss": 0.9156,
+      "step": 23101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0299816279104338e-06,
+      "loss": 0.9665,
+      "step": 23102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029875832087615e-06,
+      "loss": 0.932,
+      "step": 23103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0297702229758554e-06,
+      "loss": 0.9735,
+      "step": 23104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0296648005771335e-06,
+      "loss": 0.9292,
+      "step": 23105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029559564893423e-06,
+      "loss": 0.9477,
+      "step": 23106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0294545159266956e-06,
+      "loss": 0.9905,
+      "step": 23107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029349653678917e-06,
+      "loss": 0.9316,
+      "step": 23108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.029244978152052e-06,
+      "loss": 0.9306,
+      "step": 23109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0291404893480594e-06,
+      "loss": 0.9698,
+      "step": 23110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0290361872688983e-06,
+      "loss": 0.9579,
+      "step": 23111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02893207191652e-06,
+      "loss": 0.9493,
+      "step": 23112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028828143292876e-06,
+      "loss": 0.9426,
+      "step": 23113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0287244013999097e-06,
+      "loss": 1.0043,
+      "step": 23114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0286208462395676e-06,
+      "loss": 0.9656,
+      "step": 23115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0285174778137864e-06,
+      "loss": 0.9651,
+      "step": 23116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028414296124504e-06,
+      "loss": 0.9041,
+      "step": 23117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028311301173649e-06,
+      "loss": 0.914,
+      "step": 23118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028208492963154e-06,
+      "loss": 0.9359,
+      "step": 23119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028105871494942e-06,
+      "loss": 0.9795,
+      "step": 23120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.028003436770937e-06,
+      "loss": 0.8642,
+      "step": 23121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0279011887930547e-06,
+      "loss": 0.9612,
+      "step": 23122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0277991275632106e-06,
+      "loss": 1.0007,
+      "step": 23123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0276972530833186e-06,
+      "loss": 0.9327,
+      "step": 23124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0275955653552827e-06,
+      "loss": 0.9218,
+      "step": 23125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0274940643810095e-06,
+      "loss": 0.8852,
+      "step": 23126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.027392750162399e-06,
+      "loss": 0.9132,
+      "step": 23127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0272916227013487e-06,
+      "loss": 0.9463,
+      "step": 23128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0271906819997536e-06,
+      "loss": 0.986,
+      "step": 23129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0270899280595014e-06,
+      "loss": 0.9145,
+      "step": 23130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026989360882481e-06,
+      "loss": 0.9013,
+      "step": 23131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0268889804705762e-06,
+      "loss": 0.9608,
+      "step": 23132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0267887868256644e-06,
+      "loss": 0.9524,
+      "step": 23133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026688779949624e-06,
+      "loss": 0.9164,
+      "step": 23134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0265889598443268e-06,
+      "loss": 0.9364,
+      "step": 23135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026489326511644e-06,
+      "loss": 0.9256,
+      "step": 23136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0263898799534377e-06,
+      "loss": 0.9223,
+      "step": 23137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026290620171574e-06,
+      "loss": 0.9844,
+      "step": 23138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0261915471679096e-06,
+      "loss": 0.9896,
+      "step": 23139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.026092660944301e-06,
+      "loss": 0.972,
+      "step": 23140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025993961502598e-06,
+      "loss": 0.9787,
+      "step": 23141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0258954488446527e-06,
+      "loss": 0.9847,
+      "step": 23142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0257971229723062e-06,
+      "loss": 0.8793,
+      "step": 23143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0256989838874023e-06,
+      "loss": 0.9332,
+      "step": 23144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0256010315917776e-06,
+      "loss": 0.9301,
+      "step": 23145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0255032660872657e-06,
+      "loss": 0.8945,
+      "step": 23146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0254056873757003e-06,
+      "loss": 0.9103,
+      "step": 23147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0253082954589055e-06,
+      "loss": 0.8805,
+      "step": 23148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.025211090338708e-06,
+      "loss": 0.9151,
+      "step": 23149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0251140720169264e-06,
+      "loss": 0.9538,
+      "step": 23150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0250172404953774e-06,
+      "loss": 0.9311,
+      "step": 23151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0249205957758756e-06,
+      "loss": 0.9272,
+      "step": 23152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0248241378602296e-06,
+      "loss": 0.963,
+      "step": 23153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024727866750247e-06,
+      "loss": 0.967,
+      "step": 23154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0246317824477305e-06,
+      "loss": 0.9217,
+      "step": 23155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0245358849544786e-06,
+      "loss": 0.9714,
+      "step": 23156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0244401742722875e-06,
+      "loss": 0.9104,
+      "step": 23157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02434465040295e-06,
+      "loss": 0.9824,
+      "step": 23158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0242493133482547e-06,
+      "loss": 0.9142,
+      "step": 23159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.024154163109987e-06,
+      "loss": 0.9681,
+      "step": 23160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0240591996899284e-06,
+      "loss": 0.9455,
+      "step": 23161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023964423089858e-06,
+      "loss": 0.9863,
+      "step": 23162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0238698333115505e-06,
+      "loss": 0.9866,
+      "step": 23163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023775430356776e-06,
+      "loss": 0.9859,
+      "step": 23164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0236812142273047e-06,
+      "loss": 0.9048,
+      "step": 23165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0235871849249003e-06,
+      "loss": 0.8991,
+      "step": 23166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0234933424513216e-06,
+      "loss": 0.9539,
+      "step": 23167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0233996868083285e-06,
+      "loss": 0.904,
+      "step": 23168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0233062179976737e-06,
+      "loss": 0.9852,
+      "step": 23169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0232129360211084e-06,
+      "loss": 0.9164,
+      "step": 23170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023119840880379e-06,
+      "loss": 0.8516,
+      "step": 23171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.023026932577229e-06,
+      "loss": 1.0279,
+      "step": 23172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0229342111133976e-06,
+      "loss": 0.9623,
+      "step": 23173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0228416764906226e-06,
+      "loss": 0.8655,
+      "step": 23174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.022749328710635e-06,
+      "loss": 0.9285,
+      "step": 23175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0226571677751663e-06,
+      "loss": 0.9204,
+      "step": 23176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0225651936859406e-06,
+      "loss": 0.9238,
+      "step": 23177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0224734064446817e-06,
+      "loss": 0.8927,
+      "step": 23178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0223818060531066e-06,
+      "loss": 0.9393,
+      "step": 23179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0222903925129335e-06,
+      "loss": 0.9056,
+      "step": 23180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0221991658258725e-06,
+      "loss": 0.9244,
+      "step": 23181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0221081259936317e-06,
+      "loss": 0.9652,
+      "step": 23182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0220172730179166e-06,
+      "loss": 0.9351,
+      "step": 23183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0219266069004297e-06,
+      "loss": 0.8969,
+      "step": 23184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021836127642867e-06,
+      "loss": 0.9059,
+      "step": 23185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0217458352469236e-06,
+      "loss": 0.9445,
+      "step": 23186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0216557297142917e-06,
+      "loss": 0.9071,
+      "step": 23187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0215658110466563e-06,
+      "loss": 0.9583,
+      "step": 23188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0214760792457037e-06,
+      "loss": 0.9192,
+      "step": 23189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0213865343131124e-06,
+      "loss": 0.9681,
+      "step": 23190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.02129717625056e-06,
+      "loss": 0.9648,
+      "step": 23191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021208005059721e-06,
+      "loss": 0.9514,
+      "step": 23192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021119020742264e-06,
+      "loss": 0.8967,
+      "step": 23193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.021030223299855e-06,
+      "loss": 0.8903,
+      "step": 23194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0209416127341575e-06,
+      "loss": 1.0171,
+      "step": 23195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020853189046832e-06,
+      "loss": 0.9587,
+      "step": 23196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0207649522395334e-06,
+      "loss": 0.9545,
+      "step": 23197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020676902313914e-06,
+      "loss": 0.945,
+      "step": 23198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020589039271622e-06,
+      "loss": 0.9318,
+      "step": 23199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0205013631143058e-06,
+      "loss": 0.9469,
+      "step": 23200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020413873843604e-06,
+      "loss": 0.8885,
+      "step": 23201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020326571461157e-06,
+      "loss": 0.925,
+      "step": 23202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.020239455968598e-06,
+      "loss": 0.9691,
+      "step": 23203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0201525273675594e-06,
+      "loss": 0.92,
+      "step": 23204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0200657856596695e-06,
+      "loss": 0.9236,
+      "step": 23205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019979230846553e-06,
+      "loss": 0.8993,
+      "step": 23206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019892862929829e-06,
+      "loss": 0.9418,
+      "step": 23207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0198066819111175e-06,
+      "loss": 0.9296,
+      "step": 23208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0197206877920297e-06,
+      "loss": 0.9926,
+      "step": 23209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0196348805741774e-06,
+      "loss": 0.9551,
+      "step": 23210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0195492602591682e-06,
+      "loss": 0.9376,
+      "step": 23211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019463826848605e-06,
+      "loss": 0.9316,
+      "step": 23212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019378580344086e-06,
+      "loss": 0.9636,
+      "step": 23213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0192935207472098e-06,
+      "loss": 0.9124,
+      "step": 23214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0192086480595686e-06,
+      "loss": 0.9751,
+      "step": 23215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.019123962282752e-06,
+      "loss": 0.9531,
+      "step": 23216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0190394634183455e-06,
+      "loss": 0.8716,
+      "step": 23217
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018955151467932e-06,
+      "loss": 0.88,
+      "step": 23218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0188710264330905e-06,
+      "loss": 0.8712,
+      "step": 23219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018787088315395e-06,
+      "loss": 0.9767,
+      "step": 23220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018703337116419e-06,
+      "loss": 1.0182,
+      "step": 23221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018619772837731e-06,
+      "loss": 0.884,
+      "step": 23222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0185363954808947e-06,
+      "loss": 0.8907,
+      "step": 23223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0184532050474718e-06,
+      "loss": 0.9607,
+      "step": 23224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018370201539022e-06,
+      "loss": 0.9185,
+      "step": 23225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018287384957096e-06,
+      "loss": 0.8891,
+      "step": 23226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0182047553032484e-06,
+      "loss": 0.9214,
+      "step": 23227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.018122312579025e-06,
+      "loss": 0.9169,
+      "step": 23228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0180400567859695e-06,
+      "loss": 0.9392,
+      "step": 23229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017957987925623e-06,
+      "loss": 0.8678,
+      "step": 23230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0178761059995224e-06,
+      "loss": 0.9886,
+      "step": 23231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0177944110091998e-06,
+      "loss": 0.9356,
+      "step": 23232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0177129029561873e-06,
+      "loss": 0.9347,
+      "step": 23233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0176315818420093e-06,
+      "loss": 0.9115,
+      "step": 23234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01755044766819e-06,
+      "loss": 0.8832,
+      "step": 23235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017469500436248e-06,
+      "loss": 0.9168,
+      "step": 23236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017388740147699e-06,
+      "loss": 0.9121,
+      "step": 23237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0173081668040574e-06,
+      "loss": 1.0055,
+      "step": 23238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01722778040683e-06,
+      "loss": 0.9498,
+      "step": 23239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0171475809575216e-06,
+      "loss": 0.9337,
+      "step": 23240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.017067568457637e-06,
+      "loss": 0.8467,
+      "step": 23241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0169877429086717e-06,
+      "loss": 0.887,
+      "step": 23242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016908104312122e-06,
+      "loss": 0.9641,
+      "step": 23243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0168286526694792e-06,
+      "loss": 0.8605,
+      "step": 23244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0167493879822323e-06,
+      "loss": 1.0073,
+      "step": 23245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016670310251863e-06,
+      "loss": 0.9178,
+      "step": 23246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0165914194798545e-06,
+      "loss": 0.9374,
+      "step": 23247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0165127156676836e-06,
+      "loss": 0.9537,
+      "step": 23248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0164341988168227e-06,
+      "loss": 0.9378,
+      "step": 23249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0163558689287444e-06,
+      "loss": 0.9666,
+      "step": 23250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0162777260049153e-06,
+      "loss": 0.9113,
+      "step": 23251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0161997700467974e-06,
+      "loss": 0.9659,
+      "step": 23252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0161220010558505e-06,
+      "loss": 0.8988,
+      "step": 23253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.016044419033533e-06,
+      "loss": 0.9402,
+      "step": 23254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015967023981296e-06,
+      "loss": 0.9764,
+      "step": 23255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0158898159005905e-06,
+      "loss": 0.9375,
+      "step": 23256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01581279479286e-06,
+      "loss": 0.9205,
+      "step": 23257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0157359606595486e-06,
+      "loss": 0.9483,
+      "step": 23258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0156593135020954e-06,
+      "loss": 0.8926,
+      "step": 23259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015582853321935e-06,
+      "loss": 0.9826,
+      "step": 23260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015506580120499e-06,
+      "loss": 0.9643,
+      "step": 23261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0154304938992167e-06,
+      "loss": 0.9429,
+      "step": 23262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015354594659512e-06,
+      "loss": 0.905,
+      "step": 23263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015278882402807e-06,
+      "loss": 0.9203,
+      "step": 23264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0152033571305197e-06,
+      "loss": 0.9275,
+      "step": 23265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015128018844064e-06,
+      "loss": 0.9167,
+      "step": 23266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.015052867544851e-06,
+      "loss": 0.908,
+      "step": 23267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014977903234288e-06,
+      "loss": 0.8903,
+      "step": 23268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0149031259137776e-06,
+      "loss": 0.9297,
+      "step": 23269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0148285355847234e-06,
+      "loss": 0.9399,
+      "step": 23270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014754132248519e-06,
+      "loss": 0.8895,
+      "step": 23271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014679915906558e-06,
+      "loss": 0.9402,
+      "step": 23272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014605886560233e-06,
+      "loss": 0.8866,
+      "step": 23273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0145320442109274e-06,
+      "loss": 0.9203,
+      "step": 23274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0144583888600258e-06,
+      "loss": 0.9501,
+      "step": 23275
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014384920508907e-06,
+      "loss": 0.9475,
+      "step": 23276
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014311639158947e-06,
+      "loss": 0.8982,
+      "step": 23277
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0142385448115166e-06,
+      "loss": 0.8507,
+      "step": 23278
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0141656374679877e-06,
+      "loss": 0.8939,
+      "step": 23279
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.014092917129723e-06,
+      "loss": 0.9354,
+      "step": 23280
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0140203837980855e-06,
+      "loss": 0.9105,
+      "step": 23281
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0139480374744337e-06,
+      "loss": 0.9668,
+      "step": 23282
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0138758781601213e-06,
+      "loss": 1.0105,
+      "step": 23283
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013803905856501e-06,
+      "loss": 0.9538,
+      "step": 23284
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0137321205649197e-06,
+      "loss": 0.9049,
+      "step": 23285
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0136605222867224e-06,
+      "loss": 0.8814,
+      "step": 23286
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0135891110232486e-06,
+      "loss": 0.9304,
+      "step": 23287
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.013517886775838e-06,
+      "loss": 0.9113,
+      "step": 23288
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0134468495458218e-06,
+      "loss": 0.9138,
+      "step": 23289
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0133759993345326e-06,
+      "loss": 0.8637,
+      "step": 23290
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0133053361432953e-06,
+      "loss": 0.9147,
+      "step": 23291
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0132348599734344e-06,
+      "loss": 0.9392,
+      "step": 23292
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0131645708262693e-06,
+      "loss": 0.957,
+      "step": 23293
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0130944687031162e-06,
+      "loss": 0.983,
+      "step": 23294
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0130245536052885e-06,
+      "loss": 0.9222,
+      "step": 23295
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012954825534095e-06,
+      "loss": 0.9872,
+      "step": 23296
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012885284490841e-06,
+      "loss": 0.9227,
+      "step": 23297
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012815930476829e-06,
+      "loss": 0.9567,
+      "step": 23298
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0127467634933596e-06,
+      "loss": 0.9084,
+      "step": 23299
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0126777835417256e-06,
+      "loss": 0.9264,
+      "step": 23300
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012608990623221e-06,
+      "loss": 0.9362,
+      "step": 23301
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0125403847391313e-06,
+      "loss": 0.977,
+      "step": 23302
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0124719658907446e-06,
+      "loss": 0.9766,
+      "step": 23303
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01240373407934e-06,
+      "loss": 0.9396,
+      "step": 23304
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.012335689306195e-06,
+      "loss": 0.9157,
+      "step": 23305
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122678315725855e-06,
+      "loss": 0.9108,
+      "step": 23306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0122001608797815e-06,
+      "loss": 0.9859,
+      "step": 23307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0121326772290496e-06,
+      "loss": 1.0203,
+      "step": 23308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0120653806216552e-06,
+      "loss": 0.8935,
+      "step": 23309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011998271058857e-06,
+      "loss": 0.887,
+      "step": 23310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0119313485419126e-06,
+      "loss": 0.8923,
+      "step": 23311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0118646130720745e-06,
+      "loss": 0.9143,
+      "step": 23312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.011798064650593e-06,
+      "loss": 1.0032,
+      "step": 23313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0117317032787146e-06,
+      "loss": 0.9402,
+      "step": 23314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0116655289576818e-06,
+      "loss": 0.9372,
+      "step": 23315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0115995416887325e-06,
+      "loss": 0.9478,
+      "step": 23316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011533741473105e-06,
+      "loss": 0.9806,
+      "step": 23317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011468128312031e-06,
+      "loss": 1.0001,
+      "step": 23318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0114027022067376e-06,
+      "loss": 0.8845,
+      "step": 23319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0113374631584504e-06,
+      "loss": 0.921,
+      "step": 23320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0112724111683924e-06,
+      "loss": 0.9124,
+      "step": 23321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011207546237781e-06,
+      "loss": 0.9147,
+      "step": 23322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0111428683678314e-06,
+      "loss": 0.9677,
+      "step": 23323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0110783775597544e-06,
+      "loss": 0.9399,
+      "step": 23324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.011014073814758e-06,
+      "loss": 0.8486,
+      "step": 23325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0109499571340463e-06,
+      "loss": 0.8899,
+      "step": 23326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0108860275188194e-06,
+      "loss": 0.97,
+      "step": 23327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010822284970275e-06,
+      "loss": 0.8827,
+      "step": 23328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0107587294896083e-06,
+      "loss": 0.935,
+      "step": 23329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0106953610780067e-06,
+      "loss": 0.9424,
+      "step": 23330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0106321797366593e-06,
+      "loss": 0.9219,
+      "step": 23331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0105691854667467e-06,
+      "loss": 0.9262,
+      "step": 23332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0105063782694515e-06,
+      "loss": 0.91,
+      "step": 23333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010443758145948e-06,
+      "loss": 0.9712,
+      "step": 23334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0103813250974102e-06,
+      "loss": 0.9859,
+      "step": 23335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0103190791250063e-06,
+      "loss": 0.9324,
+      "step": 23336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0102570202299013e-06,
+      "loss": 0.9182,
+      "step": 23337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.01019514841326e-06,
+      "loss": 0.8878,
+      "step": 23338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0101334636762385e-06,
+      "loss": 0.9512,
+      "step": 23339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.010071966019992e-06,
+      "loss": 0.9808,
+      "step": 23340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0100106554456745e-06,
+      "loss": 0.9396,
+      "step": 23341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009949531954432e-06,
+      "loss": 0.9356,
+      "step": 23342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00988859554741e-06,
+      "loss": 0.89,
+      "step": 23343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0098278462257493e-06,
+      "loss": 0.925,
+      "step": 23344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0097672839905883e-06,
+      "loss": 0.9156,
+      "step": 23345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00970690884306e-06,
+      "loss": 0.9351,
+      "step": 23346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0096467207842965e-06,
+      "loss": 0.9492,
+      "step": 23347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0095867198154242e-06,
+      "loss": 0.972,
+      "step": 23348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0095269059375656e-06,
+      "loss": 0.9222,
+      "step": 23349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0094672791518425e-06,
+      "loss": 0.842,
+      "step": 23350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009407839459371e-06,
+      "loss": 0.9571,
+      "step": 23351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009348586861264e-06,
+      "loss": 0.9246,
+      "step": 23352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009289521358632e-06,
+      "loss": 0.9122,
+      "step": 23353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00923064295258e-06,
+      "loss": 0.9289,
+      "step": 23354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009171951644211e-06,
+      "loss": 0.9629,
+      "step": 23355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009113447434624e-06,
+      "loss": 0.939,
+      "step": 23356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.009055130324915e-06,
+      "loss": 0.9836,
+      "step": 23357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0089970003161765e-06,
+      "loss": 0.9009,
+      "step": 23358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008939057409496e-06,
+      "loss": 0.9836,
+      "step": 23359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0088813016059587e-06,
+      "loss": 0.9421,
+      "step": 23360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0088237329066465e-06,
+      "loss": 0.9453,
+      "step": 23361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008766351312638e-06,
+      "loss": 1.0074,
+      "step": 23362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0087091568250072e-06,
+      "loss": 0.9643,
+      "step": 23363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0086521494448256e-06,
+      "loss": 0.8992,
+      "step": 23364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00859532917316e-06,
+      "loss": 0.9377,
+      "step": 23365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0085386960110755e-06,
+      "loss": 0.9541,
+      "step": 23366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0084822499596326e-06,
+      "loss": 0.9348,
+      "step": 23367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0084259910198867e-06,
+      "loss": 0.9923,
+      "step": 23368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.008369919192893e-06,
+      "loss": 0.9442,
+      "step": 23369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0083140344797016e-06,
+      "loss": 0.921,
+      "step": 23370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0082583368813593e-06,
+      "loss": 0.898,
+      "step": 23371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0082028263989076e-06,
+      "loss": 0.9124,
+      "step": 23372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0081475030333866e-06,
+      "loss": 0.9132,
+      "step": 23373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0080923667858335e-06,
+      "loss": 0.885,
+      "step": 23374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0080374176572795e-06,
+      "loss": 0.8856,
+      "step": 23375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007982655648754e-06,
+      "loss": 0.9251,
+      "step": 23376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0079280807612837e-06,
+      "loss": 0.9609,
+      "step": 23377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007873692995889e-06,
+      "loss": 0.9788,
+      "step": 23378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007819492353589e-06,
+      "loss": 0.9406,
+      "step": 23379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0077654788353985e-06,
+      "loss": 0.8795,
+      "step": 23380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0077116524423293e-06,
+      "loss": 0.9543,
+      "step": 23381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076580131753887e-06,
+      "loss": 0.9132,
+      "step": 23382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0076045610355834e-06,
+      "loss": 0.9212,
+      "step": 23383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007551296023912e-06,
+      "loss": 0.9571,
+      "step": 23384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0074982181413727e-06,
+      "loss": 0.9165,
+      "step": 23385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0074453273889607e-06,
+      "loss": 0.9477,
+      "step": 23386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0073926237676648e-06,
+      "loss": 0.9484,
+      "step": 23387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0073401072784715e-06,
+      "loss": 0.9512,
+      "step": 23388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007287777922368e-06,
+      "loss": 0.9356,
+      "step": 23389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0072356357003288e-06,
+      "loss": 0.9373,
+      "step": 23390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007183680613335e-06,
+      "loss": 1.0068,
+      "step": 23391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.007131912662358e-06,
+      "loss": 0.973,
+      "step": 23392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070803318483667e-06,
+      "loss": 0.9682,
+      "step": 23393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0070289381723277e-06,
+      "loss": 0.9036,
+      "step": 23394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0069777316352025e-06,
+      "loss": 0.9134,
+      "step": 23395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006926712237951e-06,
+      "loss": 0.9398,
+      "step": 23396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0068758799815283e-06,
+      "loss": 0.9213,
+      "step": 23397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006825234866886e-06,
+      "loss": 0.9369,
+      "step": 23398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006774776894974e-06,
+      "loss": 0.9543,
+      "step": 23399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0067245060667364e-06,
+      "loss": 0.9255,
+      "step": 23400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006674422383114e-06,
+      "loss": 0.9735,
+      "step": 23401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006624525845045e-06,
+      "loss": 1.0098,
+      "step": 23402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006574816453463e-06,
+      "loss": 0.88,
+      "step": 23403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006525294209301e-06,
+      "loss": 0.9524,
+      "step": 23404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0064759591134854e-06,
+      "loss": 0.9415,
+      "step": 23405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0064268111669395e-06,
+      "loss": 0.9542,
+      "step": 23406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0063778503705836e-06,
+      "loss": 1.0105,
+      "step": 23407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0063290767253365e-06,
+      "loss": 0.9167,
+      "step": 23408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006280490232109e-06,
+      "loss": 0.9769,
+      "step": 23409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006232090891813e-06,
+      "loss": 0.9419,
+      "step": 23410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061838787053537e-06,
+      "loss": 0.9463,
+      "step": 23411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0061358536736344e-06,
+      "loss": 0.9871,
+      "step": 23412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.006088015797555e-06,
+      "loss": 0.9873,
+      "step": 23413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0060403650780106e-06,
+      "loss": 0.9475,
+      "step": 23414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0059929015158934e-06,
+      "loss": 0.9416,
+      "step": 23415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0059456251120925e-06,
+      "loss": 0.8734,
+      "step": 23416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0058985358674943e-06,
+      "loss": 0.9132,
+      "step": 23417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005851633782978e-06,
+      "loss": 0.9155,
+      "step": 23418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0058049188594256e-06,
+      "loss": 0.9472,
+      "step": 23419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005758391097709e-06,
+      "loss": 0.9557,
+      "step": 23420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005712050498701e-06,
+      "loss": 0.9764,
+      "step": 23421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0056658970632683e-06,
+      "loss": 0.9838,
+      "step": 23422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0056199307922752e-06,
+      "loss": 0.8993,
+      "step": 23423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005574151686585e-06,
+      "loss": 0.9322,
+      "step": 23424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005528559747052e-06,
+      "loss": 0.9336,
+      "step": 23425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0054831549745315e-06,
+      "loss": 0.9404,
+      "step": 23426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0054379373698732e-06,
+      "loss": 0.9128,
+      "step": 23427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0053929069339243e-06,
+      "loss": 0.923,
+      "step": 23428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0053480636675272e-06,
+      "loss": 0.949,
+      "step": 23429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005303407571524e-06,
+      "loss": 0.9541,
+      "step": 23430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005258938646748e-06,
+      "loss": 0.9315,
+      "step": 23431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0052146568940336e-06,
+      "loss": 0.9001,
+      "step": 23432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0051705623142107e-06,
+      "loss": 0.9315,
+      "step": 23433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.005126654908104e-06,
+      "loss": 0.9585,
+      "step": 23434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050829346765356e-06,
+      "loss": 0.897,
+      "step": 23435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0050394016203245e-06,
+      "loss": 0.9696,
+      "step": 23436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004996055740286e-06,
+      "loss": 0.904,
+      "step": 23437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0049528970372326e-06,
+      "loss": 0.9504,
+      "step": 23438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004909925511971e-06,
+      "loss": 0.9542,
+      "step": 23439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0048671411653076e-06,
+      "loss": 0.9724,
+      "step": 23440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004824543998042e-06,
+      "loss": 1.0323,
+      "step": 23441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0047821340109734e-06,
+      "loss": 0.9518,
+      "step": 23442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0047399112048952e-06,
+      "loss": 0.9287,
+      "step": 23443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004697875580597e-06,
+      "loss": 0.8934,
+      "step": 23444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0046560271388684e-06,
+      "loss": 0.9467,
+      "step": 23445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0046143658804918e-06,
+      "loss": 0.9914,
+      "step": 23446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004572891806248e-06,
+      "loss": 0.9565,
+      "step": 23447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004531604916912e-06,
+      "loss": 0.9083,
+      "step": 23448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044905052132594e-06,
+      "loss": 0.8944,
+      "step": 23449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044495926960584e-06,
+      "loss": 0.9033,
+      "step": 23450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0044088673660744e-06,
+      "loss": 0.9078,
+      "step": 23451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043683292240724e-06,
+      "loss": 1.0094,
+      "step": 23452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0043279782708092e-06,
+      "loss": 0.9103,
+      "step": 23453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0042878145070425e-06,
+      "loss": 0.9833,
+      "step": 23454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004247837933523e-06,
+      "loss": 0.9288,
+      "step": 23455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0042080485509998e-06,
+      "loss": 0.9197,
+      "step": 23456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0041684463602184e-06,
+      "loss": 0.909,
+      "step": 23457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00412903136192e-06,
+      "loss": 0.9633,
+      "step": 23458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0040898035568424e-06,
+      "loss": 0.9106,
+      "step": 23459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0040507629457207e-06,
+      "loss": 0.9548,
+      "step": 23460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.004011909529287e-06,
+      "loss": 0.9857,
+      "step": 23461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039732433082673e-06,
+      "loss": 0.8992,
+      "step": 23462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0039347642833853e-06,
+      "loss": 0.9458,
+      "step": 23463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0038964724553635e-06,
+      "loss": 0.9415,
+      "step": 23464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003858367824918e-06,
+      "loss": 0.9372,
+      "step": 23465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003820450392762e-06,
+      "loss": 0.8871,
+      "step": 23466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003782720159606e-06,
+      "loss": 0.9081,
+      "step": 23467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0037451771261574e-06,
+      "loss": 0.8579,
+      "step": 23468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0037078212931173e-06,
+      "loss": 0.9145,
+      "step": 23469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036706526611874e-06,
+      "loss": 0.947,
+      "step": 23470
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0036336712310626e-06,
+      "loss": 0.9565,
+      "step": 23471
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035968770034353e-06,
+      "loss": 0.9154,
+      "step": 23472
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035602699789947e-06,
+      "loss": 0.867,
+      "step": 23473
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0035238501584276e-06,
+      "loss": 0.967,
+      "step": 23474
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0034876175424147e-06,
+      "loss": 0.9543,
+      "step": 23475
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003451572131634e-06,
+      "loss": 0.9393,
+      "step": 23476
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003415713926762e-06,
+      "loss": 0.9091,
+      "step": 23477
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0033800429284692e-06,
+      "loss": 0.9079,
+      "step": 23478
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003344559137424e-06,
+      "loss": 0.927,
+      "step": 23479
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003309262554292e-06,
+      "loss": 0.9121,
+      "step": 23480
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0032741531797308e-06,
+      "loss": 0.9184,
+      "step": 23481
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003239231014401e-06,
+      "loss": 1.0667,
+      "step": 23482
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0032044960589572e-06,
+      "loss": 0.9172,
+      "step": 23483
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031699483140468e-06,
+      "loss": 0.9113,
+      "step": 23484
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031355877803197e-06,
+      "loss": 0.9376,
+      "step": 23485
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0031014144584174e-06,
+      "loss": 0.9289,
+      "step": 23486
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0030674283489805e-06,
+      "loss": 0.8898,
+      "step": 23487
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0030336294526456e-06,
+      "loss": 0.9043,
+      "step": 23488
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.003000017770045e-06,
+      "loss": 0.9335,
+      "step": 23489
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029665933018097e-06,
+      "loss": 0.9631,
+      "step": 23490
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0029333560485644e-06,
+      "loss": 0.9581,
+      "step": 23491
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002900306010931e-06,
+      "loss": 0.8908,
+      "step": 23492
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0028674431895305e-06,
+      "loss": 0.8795,
+      "step": 23493
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002834767584976e-06,
+      "loss": 0.9697,
+      "step": 23494
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002802279197881e-06,
+      "loss": 0.9131,
+      "step": 23495
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0027699780288523e-06,
+      "loss": 0.9018,
+      "step": 23496
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0027378640784967e-06,
+      "loss": 0.927,
+      "step": 23497
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002705937347416e-06,
+      "loss": 0.8832,
+      "step": 23498
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0026741978362055e-06,
+      "loss": 0.9747,
+      "step": 23499
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002642645545461e-06,
+      "loss": 0.9555,
+      "step": 23500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002611280475773e-06,
+      "loss": 0.9175,
+      "step": 23501
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025801026277303e-06,
+      "loss": 1.0074,
+      "step": 23502
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002549112001915e-06,
+      "loss": 0.975,
+      "step": 23503
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0025183085989076e-06,
+      "loss": 0.9281,
+      "step": 23504
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002487692419286e-06,
+      "loss": 0.9019,
+      "step": 23505
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024572634636243e-06,
+      "loss": 0.931,
+      "step": 23506
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0024270217324893e-06,
+      "loss": 0.9642,
+      "step": 23507
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0023969672264502e-06,
+      "loss": 0.9188,
+      "step": 23508
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0023670999460687e-06,
+      "loss": 0.8447,
+      "step": 23509
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0023374198919034e-06,
+      "loss": 1.0056,
+      "step": 23510
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002307927064512e-06,
+      "loss": 0.9689,
+      "step": 23511
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0022786214644452e-06,
+      "loss": 0.9723,
+      "step": 23512
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002249503092252e-06,
+      "loss": 0.9505,
+      "step": 23513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002220571948478e-06,
+      "loss": 0.9444,
+      "step": 23514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021918280336662e-06,
+      "loss": 0.8778,
+      "step": 23515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002163271348352e-06,
+      "loss": 0.9104,
+      "step": 23516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0021349018930736e-06,
+      "loss": 0.8714,
+      "step": 23517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00210671966836e-06,
+      "loss": 0.9444,
+      "step": 23518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.002078724674739e-06,
+      "loss": 0.8861,
+      "step": 23519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020509169127357e-06,
+      "loss": 0.9305,
+      "step": 23520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0020232963828715e-06,
+      "loss": 0.954,
+      "step": 23521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001995863085661e-06,
+      "loss": 0.9771,
+      "step": 23522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0019686170216214e-06,
+      "loss": 0.9282,
+      "step": 23523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0019415581912596e-06,
+      "loss": 0.9389,
+      "step": 23524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0019146865950847e-06,
+      "loss": 0.9539,
+      "step": 23525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018880022336e-06,
+      "loss": 0.9418,
+      "step": 23526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018615051073033e-06,
+      "loss": 0.9618,
+      "step": 23527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018351952166917e-06,
+      "loss": 0.9715,
+      "step": 23528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0018090725622587e-06,
+      "loss": 0.9228,
+      "step": 23529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001783137144493e-06,
+      "loss": 0.9591,
+      "step": 23530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.00175738896388e-06,
+      "loss": 0.9969,
+      "step": 23531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017318280209023e-06,
+      "loss": 0.9794,
+      "step": 23532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0017064543160383e-06,
+      "loss": 0.9752,
+      "step": 23533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016812678497635e-06,
+      "loss": 0.9019,
+      "step": 23534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0016562686225485e-06,
+      "loss": 0.9273,
+      "step": 23535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001631456634863e-06,
+      "loss": 0.9509,
+      "step": 23536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001606831887171e-06,
+      "loss": 0.9384,
+      "step": 23537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001582394379933e-06,
+      "loss": 0.9964,
+      "step": 23538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001558144113607e-06,
+      "loss": 0.9482,
+      "step": 23539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0015340810886476e-06,
+      "loss": 1.0058,
+      "step": 23540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001510205305506e-06,
+      "loss": 0.9269,
+      "step": 23541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001486516764628e-06,
+      "loss": 0.9554,
+      "step": 23542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014630154664572e-06,
+      "loss": 0.861,
+      "step": 23543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001439701411435e-06,
+      "loss": 0.9042,
+      "step": 23544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0014165745999973e-06,
+      "loss": 1.036,
+      "step": 23545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.001393635032576e-06,
+      "loss": 0.9555,
+      "step": 23546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013708827096023e-06,
+      "loss": 0.9418,
+      "step": 23547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013483176315033e-06,
+      "loss": 0.8659,
+      "step": 23548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013259397986987e-06,
+      "loss": 0.8839,
+      "step": 23549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0013037492116085e-06,
+      "loss": 0.9184,
+      "step": 23550
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001281745870649e-06,
+      "loss": 0.9379,
+      "step": 23551
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001259929776233e-06,
+      "loss": 0.9353,
+      "step": 23552
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012383009287674e-06,
+      "loss": 0.9333,
+      "step": 23553
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0012168593286575e-06,
+      "loss": 0.9063,
+      "step": 23554
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001195604976306e-06,
+      "loss": 0.898,
+      "step": 23555
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011745378721096e-06,
+      "loss": 0.9607,
+      "step": 23556
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001153658016464e-06,
+      "loss": 0.9339,
+      "step": 23557
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001132965409759e-06,
+      "loss": 0.9127,
+      "step": 23558
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0011124600523827e-06,
+      "loss": 0.9157,
+      "step": 23559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010921419447195e-06,
+      "loss": 0.9697,
+      "step": 23560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010720110871487e-06,
+      "loss": 0.9224,
+      "step": 23561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010520674800496e-06,
+      "loss": 0.907,
+      "step": 23562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.001032311123793e-06,
+      "loss": 0.9,
+      "step": 23563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0010127420187498e-06,
+      "loss": 0.8542,
+      "step": 23564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000993360165288e-06,
+      "loss": 0.9275,
+      "step": 23565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009741655637684e-06,
+      "loss": 0.9328,
+      "step": 23566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009551582145515e-06,
+      "loss": 0.9626,
+      "step": 23567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0009363381179926e-06,
+      "loss": 0.9406,
+      "step": 23568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000917705274446e-06,
+      "loss": 0.8966,
+      "step": 23569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008992596842584e-06,
+      "loss": 1.0122,
+      "step": 23570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008810013477754e-06,
+      "loss": 0.8918,
+      "step": 23571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0008629302653405e-06,
+      "loss": 0.8959,
+      "step": 23572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000845046437291e-06,
+      "loss": 0.9064,
+      "step": 23573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000827349863962e-06,
+      "loss": 0.9287,
+      "step": 23574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000809840545686e-06,
+      "loss": 0.9043,
+      "step": 23575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007925184827894e-06,
+      "loss": 1.0045,
+      "step": 23576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000775383675596e-06,
+      "loss": 0.9368,
+      "step": 23577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007584361244295e-06,
+      "loss": 0.9321,
+      "step": 23578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007416758296042e-06,
+      "loss": 0.9556,
+      "step": 23579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007251027914358e-06,
+      "loss": 0.9043,
+      "step": 23580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0007087170102338e-06,
+      "loss": 0.9058,
+      "step": 23581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000692518486306e-06,
+      "loss": 0.9009,
+      "step": 23582
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0006765072199543e-06,
+      "loss": 0.9301,
+      "step": 23583
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000660683211481e-06,
+      "loss": 0.9084,
+      "step": 23584
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0006450464611795e-06,
+      "loss": 0.8968,
+      "step": 23585
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000629596969344e-06,
+      "loss": 0.9385,
+      "step": 23586
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000614334736264e-06,
+      "loss": 0.9115,
+      "step": 23587
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005992597622262e-06,
+      "loss": 0.9657,
+      "step": 23588
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005843720475105e-06,
+      "loss": 0.9395,
+      "step": 23589
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0005696715923972e-06,
+      "loss": 0.87,
+      "step": 23590
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000555158397162e-06,
+      "loss": 0.9636,
+      "step": 23591
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000540832462076e-06,
+      "loss": 0.9658,
+      "step": 23592
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000526693787407e-06,
+      "loss": 0.8642,
+      "step": 23593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000512742373421e-06,
+      "loss": 0.9569,
+      "step": 23594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000498978220379e-06,
+      "loss": 0.9483,
+      "step": 23595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000485401328537e-06,
+      "loss": 0.9094,
+      "step": 23596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004720116981516e-06,
+      "loss": 0.9813,
+      "step": 23597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000458809329473e-06,
+      "loss": 0.9656,
+      "step": 23598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004457942227475e-06,
+      "loss": 0.9631,
+      "step": 23599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00043296637822e-06,
+      "loss": 0.9025,
+      "step": 23600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00042032579613e-06,
+      "loss": 0.9754,
+      "step": 23601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0004078724767132e-06,
+      "loss": 0.9459,
+      "step": 23602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000395606420205e-06,
+      "loss": 0.9185,
+      "step": 23603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003835276268336e-06,
+      "loss": 1.0237,
+      "step": 23604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003716360968264e-06,
+      "loss": 0.9341,
+      "step": 23605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000359931830405e-06,
+      "loss": 0.9757,
+      "step": 23606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003484148277886e-06,
+      "loss": 0.886,
+      "step": 23607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003370850891937e-06,
+      "loss": 0.9406,
+      "step": 23608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003259426148316e-06,
+      "loss": 0.9068,
+      "step": 23609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003149874049114e-06,
+      "loss": 1.0217,
+      "step": 23610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000304219459638e-06,
+      "loss": 0.9784,
+      "step": 23611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000293638779214e-06,
+      "loss": 0.9162,
+      "step": 23612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000283245363837e-06,
+      "loss": 1.0041,
+      "step": 23613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002730392137008e-06,
+      "loss": 0.9319,
+      "step": 23614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002630203289963e-06,
+      "loss": 0.91,
+      "step": 23615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002531887099133e-06,
+      "loss": 0.9682,
+      "step": 23616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000243544356635e-06,
+      "loss": 0.8741,
+      "step": 23617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000234087269341e-06,
+      "loss": 0.9106,
+      "step": 23618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000224817448209e-06,
+      "loss": 0.8963,
+      "step": 23619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002157348934127e-06,
+      "loss": 1.0143,
+      "step": 23620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0002068396051215e-06,
+      "loss": 0.9529,
+      "step": 23621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001981315835032e-06,
+      "loss": 0.8958,
+      "step": 23622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001896108287202e-06,
+      "loss": 0.9032,
+      "step": 23623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001812773409325e-06,
+      "loss": 0.9231,
+      "step": 23624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001731311202956e-06,
+      "loss": 0.9757,
+      "step": 23625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001651721669627e-06,
+      "loss": 0.9208,
+      "step": 23626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001574004810812e-06,
+      "loss": 0.9093,
+      "step": 23627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001498160627986e-06,
+      "loss": 0.9424,
+      "step": 23628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001424189122567e-06,
+      "loss": 0.9883,
+      "step": 23629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001352090295926e-06,
+      "loss": 0.9614,
+      "step": 23630
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001281864149425e-06,
+      "loss": 0.9509,
+      "step": 23631
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001213510684383e-06,
+      "loss": 0.9958,
+      "step": 23632
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000114702990207e-06,
+      "loss": 0.9136,
+      "step": 23633
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0001082421803733e-06,
+      "loss": 0.8959,
+      "step": 23634
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000101968639059e-06,
+      "loss": 0.9666,
+      "step": 23635
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000958823663797e-06,
+      "loss": 0.9795,
+      "step": 23636
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000899833624512e-06,
+      "loss": 0.9233,
+      "step": 23637
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000084271627384e-06,
+      "loss": 0.9862,
+      "step": 23638
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000787471612838e-06,
+      "loss": 0.8891,
+      "step": 23639
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000734099642552e-06,
+      "loss": 0.942,
+      "step": 23640
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000682600363963e-06,
+      "loss": 0.9426,
+      "step": 23641
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000632973778058e-06,
+      "loss": 0.9084,
+      "step": 23642
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000585219885763e-06,
+      "loss": 0.9516,
+      "step": 23643
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000539338687964e-06,
+      "loss": 0.8872,
+      "step": 23644
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000495330185525e-06,
+      "loss": 0.8993,
+      "step": 23645
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000045319437926e-06,
+      "loss": 0.9403,
+      "step": 23646
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000412931269966e-06,
+      "loss": 0.886,
+      "step": 23647
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000374540858405e-06,
+      "loss": 0.9329,
+      "step": 23648
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000338023145287e-06,
+      "loss": 0.9723,
+      "step": 23649
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000030337813128e-06,
+      "loss": 0.9572,
+      "step": 23650
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000270605817077e-06,
+      "loss": 0.9272,
+      "step": 23651
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000023970620324e-06,
+      "loss": 0.9055,
+      "step": 23652
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000021067929039e-06,
+      "loss": 0.953,
+      "step": 23653
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000183525079042e-06,
+      "loss": 1.0211,
+      "step": 23654
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000158243569724e-06,
+      "loss": 0.8972,
+      "step": 23655
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000134834762885e-06,
+      "loss": 0.9459,
+      "step": 23656
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000113298659e-06,
+      "loss": 0.9075,
+      "step": 23657
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000093635258437e-06,
+      "loss": 0.9193,
+      "step": 23658
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000075844561585e-06,
+      "loss": 0.9209,
+      "step": 23659
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000059926568762e-06,
+      "loss": 0.9273,
+      "step": 23660
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000045881280286e-06,
+      "loss": 0.9417,
+      "step": 23661
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.00000337086964e-06,
+      "loss": 0.9202,
+      "step": 23662
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.000002340881735e-06,
+      "loss": 0.969,
+      "step": 23663
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000014981643312e-06,
+      "loss": 0.995,
+      "step": 23664
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000008427174456e-06,
+      "loss": 0.9495,
+      "step": 23665
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000003745410904e-06,
+      "loss": 0.8801,
+      "step": 23666
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0000000936352727e-06,
+      "loss": 0.9272,
+      "step": 23667
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2e-06,
+      "loss": 0.9008,
+      "step": 23668
+    },
+    {
+      "epoch": 1.0,
+      "step": 23668,
+      "total_flos": 1.1417813710346702e+20,
+      "train_loss": 1.0369043331080159,
+      "train_runtime": 40554.162,
+      "train_samples_per_second": 74.704,
+      "train_steps_per_second": 0.584
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 23668,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100000,
+  "total_flos": 1.1417813710346702e+20,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}