diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,18604 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9998385273696109,
+  "eval_steps": 500,
+  "global_step": 3096,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 5.6197,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0584973660566424e-06,
+      "loss": 4.0379,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.847603633754206e-06,
+      "loss": 4.0265,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.116994732113285e-06,
+      "loss": 3.4186,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.1016109623859155e-06,
+      "loss": 3.0335,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.906100999810849e-06,
+      "loss": 2.9229,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.586287634699333e-06,
+      "loss": 2.8172,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.175492098169926e-06,
+      "loss": 2.4764,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.695207267508411e-06,
+      "loss": 2.6366,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.016010832844256e-05,
+      "loss": 2.5701,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0580662493655243e-05,
+      "loss": 2.2335,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0964598365867492e-05,
+      "loss": 2.9733,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1317785131185914e-05,
+      "loss": 2.7569,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1644785000755973e-05,
+      "loss": 2.4825,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1949214596140122e-05,
+      "loss": 2.3785,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.223398946422657e-05,
+      "loss": 2.5537,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2501494333818565e-05,
+      "loss": 2.113,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2753704633565055e-05,
+      "loss": 2.5507,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2992275111066747e-05,
+      "loss": 2.8315,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3218605694499201e-05,
+      "loss": 2.6407,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3433891268453539e-05,
+      "loss": 2.3256,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3639159859711886e-05,
+      "loss": 2.3308,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3835302327794407e-05,
+      "loss": 2.5138,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4023095731924135e-05,
+      "loss": 2.1783,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4203221924771831e-05,
+      "loss": 2.4013,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4376282497242557e-05,
+      "loss": 2.6894,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4542810901262617e-05,
+      "loss": 2.202,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4703282366812616e-05,
+      "loss": 2.3651,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4858122077950907e-05,
+      "loss": 2.1564,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5007711962196765e-05,
+      "loss": 2.2699,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5152396366245794e-05,
+      "loss": 2.01,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5292486830283214e-05,
+      "loss": 2.4627,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5428266127409445e-05,
+      "loss": 2.4766,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.555999169987521e-05,
+      "loss": 2.1383,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5687898597085248e-05,
+      "loss": 2.4817,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5812201999621698e-05,
+      "loss": 2.1342,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5933099397371056e-05,
+      "loss": 2.4917,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6050772477123392e-05,
+      "loss": 2.606,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.616538876494012e-05,
+      "loss": 2.8568,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6277103060555844e-05,
+      "loss": 2.329,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6386058694635895e-05,
+      "loss": 2.2347,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.649238863451018e-05,
+      "loss": 2.4219,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6596216459782358e-05,
+      "loss": 2.4639,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6697657225768528e-05,
+      "loss": 2.0298,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6796818229894328e-05,
+      "loss": 2.6644,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6893799693851048e-05,
+      "loss": 2.1338,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6988695372383644e-05,
+      "loss": 2.1853,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7081593097980777e-05,
+      "loss": 2.2014,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7172575269398667e-05,
+      "loss": 2.2149,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7261719290828474e-05,
+      "loss": 2.6644,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.734909796757277e-05,
+      "loss": 2.0282,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.74347798632992e-05,
+      "loss": 2.7916,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.751882962326351e-05,
+      "loss": 1.9409,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.760130826731926e-05,
+      "loss": 2.6301,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7682273456041158e-05,
+      "loss": 2.4751,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7761779732869263e-05,
+      "loss": 2.2532,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7839878744820955e-05,
+      "loss": 1.9501,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7916619444007548e-05,
+      "loss": 2.6455,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7992048271924608e-05,
+      "loss": 2.3038,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8066209328253407e-05,
+      "loss": 2.224,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.813914452570987e-05,
+      "loss": 2.2083,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8210893732302436e-05,
+      "loss": 2.4714,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8281494902207743e-05,
+      "loss": 2.1586,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8350984196339852e-05,
+      "loss": 2.312,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.841939609357183e-05,
+      "loss": 2.2592,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8486763493466087e-05,
+      "loss": 2.7085,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8553117811279765e-05,
+      "loss": 2.3921,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.861848906593185e-05,
+      "loss": 2.165,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8682905961548614e-05,
+      "loss": 2.5573,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8746395963141893e-05,
+      "loss": 2.5953,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8808985366919495e-05,
+      "loss": 2.6419,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.887069936567834e-05,
+      "loss": 2.37,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8931562109687253e-05,
+      "loss": 2.4942,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.89915967634277e-05,
+      "loss": 2.5282,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9050825558526037e-05,
+      "loss": 2.5213,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9109269843180034e-05,
+      "loss": 2.7033,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9166950128354576e-05,
+      "loss": 2.4289,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.922388613099676e-05,
+      "loss": 2.4382,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.928009681449824e-05,
+      "loss": 2.3433,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9335600426612486e-05,
+      "loss": 2.1485,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9390414535016823e-05,
+      "loss": 2.2484,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.944455606069254e-05,
+      "loss": 2.7444,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9498041309281806e-05,
+      "loss": 2.7103,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9550886000566822e-05,
+      "loss": 2.323,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.960310529620448e-05,
+      "loss": 2.2805,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9654713825839e-05,
+      "loss": 2.4226,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9705725711705114e-05,
+      "loss": 3.1575,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.975615459182517e-05,
+      "loss": 2.42,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.980601364189567e-05,
+      "loss": 2.8334,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.985531559595097e-05,
+      "loss": 2.1596,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9904072765885248e-05,
+      "loss": 2.297,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9952297059907693e-05,
+      "loss": 2.1634,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 2.1536,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 2.3292,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9993339993339997e-05,
+      "loss": 2.1803,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.998667998667999e-05,
+      "loss": 2.3258,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9980019980019983e-05,
+      "loss": 2.3052,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9973359973359975e-05,
+      "loss": 2.4744,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9966699966699967e-05,
+      "loss": 2.53,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9960039960039962e-05,
+      "loss": 2.2324,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9953379953379953e-05,
+      "loss": 2.2506,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994671994671995e-05,
+      "loss": 2.2623,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994005994005994e-05,
+      "loss": 2.1717,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9933399933399935e-05,
+      "loss": 2.2862,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.992673992673993e-05,
+      "loss": 2.655,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9920079920079922e-05,
+      "loss": 2.3577,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9913419913419917e-05,
+      "loss": 2.2228,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.990675990675991e-05,
+      "loss": 2.3564,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99000999000999e-05,
+      "loss": 1.9478,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9893439893439895e-05,
+      "loss": 2.3105,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9886779886779887e-05,
+      "loss": 2.1497,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9880119880119882e-05,
+      "loss": 2.324,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9873459873459874e-05,
+      "loss": 2.347,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.986679986679987e-05,
+      "loss": 2.5979,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.986013986013986e-05,
+      "loss": 2.1719,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9853479853479855e-05,
+      "loss": 2.364,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.984681984681985e-05,
+      "loss": 2.3204,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9840159840159842e-05,
+      "loss": 2.2438,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9833499833499834e-05,
+      "loss": 2.3866,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982683982683983e-05,
+      "loss": 2.3403,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.982017982017982e-05,
+      "loss": 2.5583,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9813519813519816e-05,
+      "loss": 2.4001,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9806859806859807e-05,
+      "loss": 2.0737,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9800199800199802e-05,
+      "loss": 2.6478,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9793539793539794e-05,
+      "loss": 2.3219,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.978687978687979e-05,
+      "loss": 2.4913,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9780219780219784e-05,
+      "loss": 2.5623,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9773559773559776e-05,
+      "loss": 2.1222,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9766899766899767e-05,
+      "loss": 2.2996,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9760239760239762e-05,
+      "loss": 2.188,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9753579753579754e-05,
+      "loss": 2.3078,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.974691974691975e-05,
+      "loss": 1.9336,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.974025974025974e-05,
+      "loss": 2.331,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9733599733599736e-05,
+      "loss": 2.2825,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9726939726939727e-05,
+      "loss": 2.3408,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9720279720279722e-05,
+      "loss": 2.5497,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9713619713619718e-05,
+      "loss": 2.5769,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.970695970695971e-05,
+      "loss": 2.2489,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.97002997002997e-05,
+      "loss": 2.4171,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9693639693639696e-05,
+      "loss": 2.1212,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9686979686979688e-05,
+      "loss": 2.1807,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9680319680319683e-05,
+      "loss": 2.5963,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9673659673659674e-05,
+      "loss": 2.2758,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.966699966699967e-05,
+      "loss": 2.1257,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.966033966033966e-05,
+      "loss": 2.0845,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9653679653679656e-05,
+      "loss": 2.3406,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.964701964701965e-05,
+      "loss": 2.7481,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9640359640359643e-05,
+      "loss": 2.62,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9633699633699634e-05,
+      "loss": 2.6909,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.962703962703963e-05,
+      "loss": 2.377,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.962037962037962e-05,
+      "loss": 2.589,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9613719613719616e-05,
+      "loss": 2.3372,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9607059607059608e-05,
+      "loss": 2.5401,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9600399600399603e-05,
+      "loss": 2.4326,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9593739593739595e-05,
+      "loss": 2.1091,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.958707958707959e-05,
+      "loss": 2.5741,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.958041958041958e-05,
+      "loss": 2.2289,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9573759573759573e-05,
+      "loss": 2.1936,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9567099567099568e-05,
+      "loss": 2.5029,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9560439560439563e-05,
+      "loss": 2.1305,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9553779553779555e-05,
+      "loss": 2.6174,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.954711954711955e-05,
+      "loss": 2.3183,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.954045954045954e-05,
+      "loss": 2.1334,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9533799533799536e-05,
+      "loss": 2.0724,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9527139527139528e-05,
+      "loss": 2.4142,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9520479520479523e-05,
+      "loss": 2.4914,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9513819513819515e-05,
+      "loss": 2.4262,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9507159507159506e-05,
+      "loss": 2.4177,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.95004995004995e-05,
+      "loss": 2.6596,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9493839493839493e-05,
+      "loss": 2.715,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9487179487179488e-05,
+      "loss": 2.2579,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9480519480519483e-05,
+      "loss": 2.3006,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9473859473859475e-05,
+      "loss": 2.468,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.946719946719947e-05,
+      "loss": 2.1376,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.946053946053946e-05,
+      "loss": 2.2508,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9453879453879457e-05,
+      "loss": 2.0564,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9447219447219448e-05,
+      "loss": 2.3014,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.944055944055944e-05,
+      "loss": 2.3618,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9433899433899435e-05,
+      "loss": 2.5365,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9427239427239427e-05,
+      "loss": 2.7618,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9420579420579422e-05,
+      "loss": 2.3436,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9413919413919417e-05,
+      "loss": 2.3323,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.940725940725941e-05,
+      "loss": 2.5585,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9400599400599403e-05,
+      "loss": 2.1312,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 2.2441,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.938727938727939e-05,
+      "loss": 2.39,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9380619380619382e-05,
+      "loss": 2.2908,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9373959373959373e-05,
+      "loss": 2.0086,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.936729936729937e-05,
+      "loss": 2.4001,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.936063936063936e-05,
+      "loss": 2.8553,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9353979353979355e-05,
+      "loss": 2.4966,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.934731934731935e-05,
+      "loss": 2.154,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9340659340659342e-05,
+      "loss": 2.3334,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9333999333999337e-05,
+      "loss": 2.4991,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.932733932733933e-05,
+      "loss": 2.8977,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9320679320679324e-05,
+      "loss": 2.3419,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9314019314019315e-05,
+      "loss": 2.5231,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9307359307359307e-05,
+      "loss": 2.1564,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9300699300699302e-05,
+      "loss": 2.4947,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9294039294039294e-05,
+      "loss": 2.2529,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.928737928737929e-05,
+      "loss": 2.1226,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9280719280719284e-05,
+      "loss": 2.3532,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9274059274059275e-05,
+      "loss": 2.0921,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.926739926739927e-05,
+      "loss": 2.4993,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9260739260739262e-05,
+      "loss": 2.3471,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9254079254079257e-05,
+      "loss": 2.0753,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.924741924741925e-05,
+      "loss": 2.2224,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.924075924075924e-05,
+      "loss": 2.5303,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9234099234099236e-05,
+      "loss": 2.4996,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9227439227439227e-05,
+      "loss": 2.6265,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9220779220779222e-05,
+      "loss": 2.6065,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9214119214119214e-05,
+      "loss": 2.3971,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.920745920745921e-05,
+      "loss": 2.1446,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9200799200799204e-05,
+      "loss": 2.5675,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9194139194139196e-05,
+      "loss": 2.3931,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.918747918747919e-05,
+      "loss": 2.4702,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9180819180819182e-05,
+      "loss": 2.1735,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9174159174159174e-05,
+      "loss": 2.4071,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.916749916749917e-05,
+      "loss": 1.8597,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.916083916083916e-05,
+      "loss": 1.8554,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9154179154179156e-05,
+      "loss": 2.0846,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9147519147519148e-05,
+      "loss": 2.0149,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9140859140859143e-05,
+      "loss": 2.3099,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9134199134199138e-05,
+      "loss": 2.0661,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.912753912753913e-05,
+      "loss": 2.6157,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9120879120879124e-05,
+      "loss": 2.1951,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9114219114219116e-05,
+      "loss": 2.3748,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9107559107559108e-05,
+      "loss": 2.4059,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9100899100899103e-05,
+      "loss": 2.1412,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9094239094239094e-05,
+      "loss": 2.3607,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.908757908757909e-05,
+      "loss": 2.1776,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.908091908091908e-05,
+      "loss": 2.203,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9074259074259076e-05,
+      "loss": 2.4362,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.906759906759907e-05,
+      "loss": 2.4494,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9060939060939063e-05,
+      "loss": 2.4537,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9054279054279058e-05,
+      "loss": 2.4323,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 2.2333,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.904095904095904e-05,
+      "loss": 2.0068,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9034299034299036e-05,
+      "loss": 2.3171,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9027639027639028e-05,
+      "loss": 2.3164,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9020979020979023e-05,
+      "loss": 2.3376,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9014319014319015e-05,
+      "loss": 2.3069,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.900765900765901e-05,
+      "loss": 2.3466,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9000999000999005e-05,
+      "loss": 2.4892,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8994338994338996e-05,
+      "loss": 2.6126,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.898767898767899e-05,
+      "loss": 2.3424,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8981018981018983e-05,
+      "loss": 2.4094,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8974358974358975e-05,
+      "loss": 2.6285,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.896769896769897e-05,
+      "loss": 2.1004,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.896103896103896e-05,
+      "loss": 2.388,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8954378954378956e-05,
+      "loss": 2.4717,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8947718947718948e-05,
+      "loss": 2.5347,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8941058941058943e-05,
+      "loss": 2.3411,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8934398934398935e-05,
+      "loss": 2.2643,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.892773892773893e-05,
+      "loss": 2.1589,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8921078921078925e-05,
+      "loss": 2.2632,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8914418914418917e-05,
+      "loss": 2.3147,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8907758907758908e-05,
+      "loss": 2.5631,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8901098901098903e-05,
+      "loss": 2.3611,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8894438894438895e-05,
+      "loss": 2.1864,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.888777888777889e-05,
+      "loss": 2.1575,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.888111888111888e-05,
+      "loss": 1.9505,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8874458874458877e-05,
+      "loss": 2.3675,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.886779886779887e-05,
+      "loss": 2.163,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8861138861138863e-05,
+      "loss": 2.4756,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.885447885447886e-05,
+      "loss": 2.3415,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8847818847818847e-05,
+      "loss": 2.2277,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8841158841158842e-05,
+      "loss": 2.3095,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8834498834498837e-05,
+      "loss": 2.2745,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.882783882783883e-05,
+      "loss": 2.394,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8821178821178824e-05,
+      "loss": 2.2613,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8814518814518815e-05,
+      "loss": 2.3459,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.880785880785881e-05,
+      "loss": 2.2457,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8801198801198802e-05,
+      "loss": 2.1834,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8794538794538797e-05,
+      "loss": 2.1588,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 1.8992,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.878121878121878e-05,
+      "loss": 2.0668,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8774558774558775e-05,
+      "loss": 2.4307,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.876789876789877e-05,
+      "loss": 2.3569,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8761238761238762e-05,
+      "loss": 2.4165,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8754578754578757e-05,
+      "loss": 2.4905,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.874791874791875e-05,
+      "loss": 2.2661,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8741258741258744e-05,
+      "loss": 2.0345,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8734598734598735e-05,
+      "loss": 2.2992,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.872793872793873e-05,
+      "loss": 2.4836,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8721278721278722e-05,
+      "loss": 2.3262,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8714618714618714e-05,
+      "loss": 2.153,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.870795870795871e-05,
+      "loss": 2.3403,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8701298701298704e-05,
+      "loss": 2.646,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8694638694638696e-05,
+      "loss": 2.2565,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.868797868797869e-05,
+      "loss": 2.2722,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8681318681318682e-05,
+      "loss": 2.3042,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8674658674658677e-05,
+      "loss": 2.1947,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.866799866799867e-05,
+      "loss": 2.3379,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8661338661338664e-05,
+      "loss": 2.2324,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8654678654678656e-05,
+      "loss": 2.3622,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8648018648018647e-05,
+      "loss": 2.3975,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8641358641358642e-05,
+      "loss": 2.4424,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8634698634698637e-05,
+      "loss": 2.3469,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.862803862803863e-05,
+      "loss": 2.3571,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8621378621378624e-05,
+      "loss": 2.3555,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8614718614718616e-05,
+      "loss": 2.163,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.860805860805861e-05,
+      "loss": 2.346,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8601398601398602e-05,
+      "loss": 2.2202,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8594738594738598e-05,
+      "loss": 2.1152,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.858807858807859e-05,
+      "loss": 2.2279,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.858141858141858e-05,
+      "loss": 2.2075,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8574758574758576e-05,
+      "loss": 2.2101,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8568098568098568e-05,
+      "loss": 2.3263,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8561438561438563e-05,
+      "loss": 2.3575,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8554778554778558e-05,
+      "loss": 2.4605,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.854811854811855e-05,
+      "loss": 2.2948,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8541458541458544e-05,
+      "loss": 2.2603,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8534798534798536e-05,
+      "loss": 2.3462,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.852813852813853e-05,
+      "loss": 2.2318,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8521478521478523e-05,
+      "loss": 2.6032,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8514818514818514e-05,
+      "loss": 2.2655,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.850815850815851e-05,
+      "loss": 2.3818,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.85014985014985e-05,
+      "loss": 2.0516,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8494838494838496e-05,
+      "loss": 2.1365,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.848817848817849e-05,
+      "loss": 2.2405,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8481518481518483e-05,
+      "loss": 2.2483,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8474858474858478e-05,
+      "loss": 2.2947,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.846819846819847e-05,
+      "loss": 2.6142,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 2.3921,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8454878454878456e-05,
+      "loss": 2.1345,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8448218448218448e-05,
+      "loss": 2.4003,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8441558441558443e-05,
+      "loss": 2.2519,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8434898434898435e-05,
+      "loss": 2.3748,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.842823842823843e-05,
+      "loss": 2.3626,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8421578421578425e-05,
+      "loss": 2.1233,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8414918414918416e-05,
+      "loss": 2.2667,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.840825840825841e-05,
+      "loss": 2.4423,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8401598401598403e-05,
+      "loss": 2.5582,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8394938394938398e-05,
+      "loss": 2.1841,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.838827838827839e-05,
+      "loss": 2.165,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.838161838161838e-05,
+      "loss": 2.3087,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8374958374958377e-05,
+      "loss": 1.9477,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8368298368298368e-05,
+      "loss": 2.1811,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8361638361638363e-05,
+      "loss": 2.2253,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8354978354978358e-05,
+      "loss": 2.5806,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.834831834831835e-05,
+      "loss": 2.5688,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8341658341658345e-05,
+      "loss": 2.2876,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8334998334998337e-05,
+      "loss": 2.3799,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.832833832833833e-05,
+      "loss": 2.512,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8321678321678323e-05,
+      "loss": 2.468,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8315018315018315e-05,
+      "loss": 2.3685,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.830835830835831e-05,
+      "loss": 2.6111,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8301698301698302e-05,
+      "loss": 2.4662,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8295038295038297e-05,
+      "loss": 2.3753,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.828837828837829e-05,
+      "loss": 2.4121,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8281718281718283e-05,
+      "loss": 2.3655,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.827505827505828e-05,
+      "loss": 2.088,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.826839826839827e-05,
+      "loss": 2.1913,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8261738261738262e-05,
+      "loss": 2.707,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8255078255078257e-05,
+      "loss": 2.5546,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.824841824841825e-05,
+      "loss": 2.7323,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8241758241758244e-05,
+      "loss": 2.3275,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8235098235098235e-05,
+      "loss": 2.1817,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.822843822843823e-05,
+      "loss": 2.2056,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8221778221778222e-05,
+      "loss": 2.257,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8215118215118217e-05,
+      "loss": 2.414,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8208458208458212e-05,
+      "loss": 2.0458,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8201798201798204e-05,
+      "loss": 2.3306,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8195138195138195e-05,
+      "loss": 1.9361,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.818847818847819e-05,
+      "loss": 2.0643,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 2.2792,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8175158175158177e-05,
+      "loss": 2.2135,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.816849816849817e-05,
+      "loss": 2.5029,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8161838161838164e-05,
+      "loss": 2.329,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8155178155178155e-05,
+      "loss": 2.4359,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.814851814851815e-05,
+      "loss": 2.3799,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8141858141858146e-05,
+      "loss": 2.2492,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8135198135198137e-05,
+      "loss": 2.0776,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.812853812853813e-05,
+      "loss": 2.1338,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8121878121878124e-05,
+      "loss": 2.1495,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8115218115218116e-05,
+      "loss": 2.1781,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.810855810855811e-05,
+      "loss": 2.1408,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8101898101898102e-05,
+      "loss": 2.4071,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 1.9914,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.808857808857809e-05,
+      "loss": 2.2891,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8081918081918084e-05,
+      "loss": 2.69,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.807525807525808e-05,
+      "loss": 2.0886,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.806859806859807e-05,
+      "loss": 2.4311,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8061938061938062e-05,
+      "loss": 2.2259,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.8055278055278057e-05,
+      "loss": 2.1621,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.804861804861805e-05,
+      "loss": 2.1084,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8041958041958044e-05,
+      "loss": 2.6351,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8035298035298036e-05,
+      "loss": 2.2655,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.802863802863803e-05,
+      "loss": 2.2025,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8021978021978023e-05,
+      "loss": 2.2536,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8015318015318018e-05,
+      "loss": 2.4942,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.800865800865801e-05,
+      "loss": 2.4491,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.8001998001998004e-05,
+      "loss": 2.0134,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7995337995337996e-05,
+      "loss": 2.5289,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.798867798867799e-05,
+      "loss": 1.8824,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7982017982017983e-05,
+      "loss": 2.0628,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7975357975357978e-05,
+      "loss": 2.5476,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.796869796869797e-05,
+      "loss": 2.2608,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7962037962037964e-05,
+      "loss": 2.4513,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7955377955377956e-05,
+      "loss": 1.9845,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.794871794871795e-05,
+      "loss": 2.5729,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7942057942057943e-05,
+      "loss": 2.2361,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7935397935397938e-05,
+      "loss": 2.4256,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.792873792873793e-05,
+      "loss": 2.5642,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.792207792207792e-05,
+      "loss": 2.3534,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7915417915417916e-05,
+      "loss": 2.2842,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.790875790875791e-05,
+      "loss": 2.4754,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7902097902097903e-05,
+      "loss": 2.2847,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7895437895437898e-05,
+      "loss": 2.2094,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.788877788877789e-05,
+      "loss": 2.177,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7882117882117885e-05,
+      "loss": 2.1124,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7875457875457876e-05,
+      "loss": 2.2123,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7868797868797868e-05,
+      "loss": 2.3913,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7862137862137863e-05,
+      "loss": 2.0855,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7855477855477855e-05,
+      "loss": 2.4815,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.784881784881785e-05,
+      "loss": 2.3416,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7842157842157845e-05,
+      "loss": 2.4372,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7835497835497836e-05,
+      "loss": 2.1983,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.782883782883783e-05,
+      "loss": 2.1572,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7822177822177823e-05,
+      "loss": 2.1836,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7815517815517818e-05,
+      "loss": 2.476,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.780885780885781e-05,
+      "loss": 2.1913,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.78021978021978e-05,
+      "loss": 2.0628,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7795537795537797e-05,
+      "loss": 2.52,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7788877788877788e-05,
+      "loss": 2.2527,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7782217782217783e-05,
+      "loss": 2.5086,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.777555777555778e-05,
+      "loss": 1.9078,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.776889776889777e-05,
+      "loss": 2.4022,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7762237762237765e-05,
+      "loss": 2.1668,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7755577755577757e-05,
+      "loss": 2.1554,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7748917748917752e-05,
+      "loss": 2.2628,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7742257742257743e-05,
+      "loss": 2.3566,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7735597735597735e-05,
+      "loss": 1.9286,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.772893772893773e-05,
+      "loss": 2.1282,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7722277722277722e-05,
+      "loss": 2.1971,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7715617715617717e-05,
+      "loss": 2.2017,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7708957708957712e-05,
+      "loss": 2.3941,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7702297702297704e-05,
+      "loss": 2.238,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.76956376956377e-05,
+      "loss": 2.2168,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.768897768897769e-05,
+      "loss": 2.633,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7682317682317685e-05,
+      "loss": 2.2253,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7675657675657677e-05,
+      "loss": 2.3759,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.766899766899767e-05,
+      "loss": 2.1857,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7662337662337664e-05,
+      "loss": 1.8779,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7655677655677655e-05,
+      "loss": 2.3284,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.764901764901765e-05,
+      "loss": 2.3726,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.7642357642357642e-05,
+      "loss": 2.3571,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7635697635697637e-05,
+      "loss": 2.234,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7629037629037632e-05,
+      "loss": 2.4575,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7622377622377624e-05,
+      "loss": 2.4493,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.761571761571762e-05,
+      "loss": 2.5949,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.760905760905761e-05,
+      "loss": 2.1611,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7602397602397602e-05,
+      "loss": 2.098,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7595737595737597e-05,
+      "loss": 2.5338,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.758907758907759e-05,
+      "loss": 2.1937,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7582417582417584e-05,
+      "loss": 1.9971,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 2.5965,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.756909756909757e-05,
+      "loss": 2.5391,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7562437562437566e-05,
+      "loss": 2.2621,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7555777555777557e-05,
+      "loss": 2.6587,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7549117549117552e-05,
+      "loss": 2.1076,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7542457542457544e-05,
+      "loss": 2.2218,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7535797535797536e-05,
+      "loss": 2.1344,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.752913752913753e-05,
+      "loss": 2.1141,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7522477522477522e-05,
+      "loss": 2.2861,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7515817515817517e-05,
+      "loss": 2.0521,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.750915750915751e-05,
+      "loss": 2.2225,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7502497502497504e-05,
+      "loss": 2.2175,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.74958374958375e-05,
+      "loss": 2.2166,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.748917748917749e-05,
+      "loss": 2.2954,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7482517482517486e-05,
+      "loss": 2.304,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7475857475857478e-05,
+      "loss": 2.1468,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.746919746919747e-05,
+      "loss": 2.086,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7462537462537464e-05,
+      "loss": 2.2171,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7455877455877456e-05,
+      "loss": 1.9691,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.744921744921745e-05,
+      "loss": 2.2249,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7442557442557443e-05,
+      "loss": 2.2322,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7435897435897438e-05,
+      "loss": 2.0625,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7429237429237433e-05,
+      "loss": 2.207,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7422577422577424e-05,
+      "loss": 1.9785,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.741591741591742e-05,
+      "loss": 2.3575,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.740925740925741e-05,
+      "loss": 2.4298,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7402597402597403e-05,
+      "loss": 2.0367,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7395937395937398e-05,
+      "loss": 2.106,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.738927738927739e-05,
+      "loss": 2.2274,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7382617382617384e-05,
+      "loss": 2.5543,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7375957375957376e-05,
+      "loss": 2.1349,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.736929736929737e-05,
+      "loss": 2.7839,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7362637362637363e-05,
+      "loss": 2.4834,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7355977355977358e-05,
+      "loss": 2.2922,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7349317349317353e-05,
+      "loss": 1.9395,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7342657342657345e-05,
+      "loss": 2.1922,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7335997335997336e-05,
+      "loss": 2.6027,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.732933732933733e-05,
+      "loss": 2.224,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7322677322677323e-05,
+      "loss": 2.3803,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7316017316017318e-05,
+      "loss": 2.1903,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.730935730935731e-05,
+      "loss": 2.2367,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7302697302697305e-05,
+      "loss": 2.1596,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7296037296037296e-05,
+      "loss": 2.0781,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.728937728937729e-05,
+      "loss": 2.114,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7282717282717286e-05,
+      "loss": 1.8946,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7276057276057275e-05,
+      "loss": 2.435,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.726939726939727e-05,
+      "loss": 2.6339,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7262737262737265e-05,
+      "loss": 2.3442,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7256077256077257e-05,
+      "loss": 1.9465,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.724941724941725e-05,
+      "loss": 2.1176,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7242757242757243e-05,
+      "loss": 2.1839,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7236097236097238e-05,
+      "loss": 2.3516,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.722943722943723e-05,
+      "loss": 2.5197,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7222777222777225e-05,
+      "loss": 2.2095,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.721611721611722e-05,
+      "loss": 2.0084,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7209457209457208e-05,
+      "loss": 2.2814,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7202797202797203e-05,
+      "loss": 2.7024,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.71961371961372e-05,
+      "loss": 2.066,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.718947718947719e-05,
+      "loss": 2.1891,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7182817182817185e-05,
+      "loss": 2.4965,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7176157176157177e-05,
+      "loss": 2.0252,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7169497169497172e-05,
+      "loss": 2.1452,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7162837162837163e-05,
+      "loss": 2.3493,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.715617715617716e-05,
+      "loss": 2.3903,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7149517149517154e-05,
+      "loss": 2.1646,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 2.0877,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7136197136197137e-05,
+      "loss": 1.9676,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7129537129537132e-05,
+      "loss": 2.041,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7122877122877124e-05,
+      "loss": 2.2661,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.711621711621712e-05,
+      "loss": 1.9324,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.710955710955711e-05,
+      "loss": 2.2885,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7102897102897105e-05,
+      "loss": 2.1132,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7096237096237097e-05,
+      "loss": 2.3448,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7089577089577092e-05,
+      "loss": 2.5355,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7082917082917084e-05,
+      "loss": 2.2138,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7076257076257075e-05,
+      "loss": 2.3053,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.706959706959707e-05,
+      "loss": 2.2341,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7062937062937065e-05,
+      "loss": 2.0239,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7056277056277057e-05,
+      "loss": 2.3606,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7049617049617052e-05,
+      "loss": 2.0661,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7042957042957044e-05,
+      "loss": 2.3162,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.703629703629704e-05,
+      "loss": 2.5458,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.702963702963703e-05,
+      "loss": 2.2829,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.7022977022977026e-05,
+      "loss": 2.334,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7016317016317017e-05,
+      "loss": 2.5688,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.700965700965701e-05,
+      "loss": 2.2628,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.7002997002997004e-05,
+      "loss": 2.4533,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6996336996336996e-05,
+      "loss": 2.2987,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.698967698967699e-05,
+      "loss": 2.2345,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6983016983016986e-05,
+      "loss": 2.2184,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6976356976356977e-05,
+      "loss": 2.3339,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 2.2084,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6963036963036964e-05,
+      "loss": 1.9902,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.695637695637696e-05,
+      "loss": 2.4545,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.694971694971695e-05,
+      "loss": 2.6325,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6943056943056942e-05,
+      "loss": 2.3287,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6936396936396937e-05,
+      "loss": 2.0722,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.692973692973693e-05,
+      "loss": 2.2756,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 2.3275,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.691641691641692e-05,
+      "loss": 2.0434,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.690975690975691e-05,
+      "loss": 2.2929,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6903096903096906e-05,
+      "loss": 2.4747,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6896436896436898e-05,
+      "loss": 2.64,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6889776889776893e-05,
+      "loss": 2.5052,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6883116883116884e-05,
+      "loss": 2.084,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6876456876456876e-05,
+      "loss": 2.3138,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.686979686979687e-05,
+      "loss": 2.4343,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6863136863136863e-05,
+      "loss": 2.0991,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6856476856476858e-05,
+      "loss": 2.1944,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6849816849816853e-05,
+      "loss": 2.0603,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6843156843156844e-05,
+      "loss": 1.9135,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.683649683649684e-05,
+      "loss": 1.9686,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.682983682983683e-05,
+      "loss": 2.1508,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6823176823176826e-05,
+      "loss": 2.5039,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.6816516816516818e-05,
+      "loss": 2.1426,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.680985680985681e-05,
+      "loss": 2.0784,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6803196803196805e-05,
+      "loss": 2.0423,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6796536796536796e-05,
+      "loss": 2.0142,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.678987678987679e-05,
+      "loss": 2.1758,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6783216783216786e-05,
+      "loss": 2.1645,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6776556776556778e-05,
+      "loss": 2.3659,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6769896769896773e-05,
+      "loss": 2.0987,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6763236763236765e-05,
+      "loss": 2.0613,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.675657675657676e-05,
+      "loss": 1.7434,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.674991674991675e-05,
+      "loss": 2.0355,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6743256743256743e-05,
+      "loss": 2.0541,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6736596736596738e-05,
+      "loss": 2.1092,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.672993672993673e-05,
+      "loss": 1.9518,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6723276723276725e-05,
+      "loss": 2.6304,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6716616716616716e-05,
+      "loss": 2.2874,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.670995670995671e-05,
+      "loss": 2.4463,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6703296703296707e-05,
+      "loss": 2.0073,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6696636696636698e-05,
+      "loss": 2.1739,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6689976689976693e-05,
+      "loss": 2.1862,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6683316683316685e-05,
+      "loss": 2.3884,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6676656676656677e-05,
+      "loss": 2.3039,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.666999666999667e-05,
+      "loss": 2.3983,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6663336663336663e-05,
+      "loss": 2.1567,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.665667665667666e-05,
+      "loss": 2.2221,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.665001665001665e-05,
+      "loss": 2.5055,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6643356643356645e-05,
+      "loss": 2.3958,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.663669663669664e-05,
+      "loss": 2.2816,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6630036630036632e-05,
+      "loss": 2.0743,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.6623376623376627e-05,
+      "loss": 2.4132,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.661671661671662e-05,
+      "loss": 2.0607,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.661005661005661e-05,
+      "loss": 2.2404,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6603396603396605e-05,
+      "loss": 2.2776,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6596736596736597e-05,
+      "loss": 2.2286,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6590076590076592e-05,
+      "loss": 2.4273,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6583416583416584e-05,
+      "loss": 2.3809,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.657675657675658e-05,
+      "loss": 2.2514,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6570096570096574e-05,
+      "loss": 2.094,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6563436563436565e-05,
+      "loss": 2.0031,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6556776556776557e-05,
+      "loss": 1.9058,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6550116550116552e-05,
+      "loss": 2.3659,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6543456543456544e-05,
+      "loss": 2.2941,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.653679653679654e-05,
+      "loss": 2.3532,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.653013653013653e-05,
+      "loss": 1.8604,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6523476523476525e-05,
+      "loss": 2.0395,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6516816516816517e-05,
+      "loss": 2.0781,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6510156510156512e-05,
+      "loss": 2.4494,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6503496503496507e-05,
+      "loss": 2.4592,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.64968364968365e-05,
+      "loss": 2.1272,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.649017649017649e-05,
+      "loss": 2.1638,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6483516483516486e-05,
+      "loss": 2.3392,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6476856476856477e-05,
+      "loss": 2.5994,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6470196470196472e-05,
+      "loss": 2.2454,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6463536463536464e-05,
+      "loss": 2.2388,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.645687645687646e-05,
+      "loss": 2.2044,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.645021645021645e-05,
+      "loss": 2.1871,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6443556443556446e-05,
+      "loss": 2.0402,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6436896436896437e-05,
+      "loss": 2.0985,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6430236430236432e-05,
+      "loss": 1.9078,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6423576423576424e-05,
+      "loss": 2.2103,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.641691641691642e-05,
+      "loss": 2.5373,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.641025641025641e-05,
+      "loss": 2.1592,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.6403596403596406e-05,
+      "loss": 2.1464,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6396936396936397e-05,
+      "loss": 2.0688,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6390276390276392e-05,
+      "loss": 2.3607,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6383616383616384e-05,
+      "loss": 1.9685,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.637695637695638e-05,
+      "loss": 2.4353,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.637029637029637e-05,
+      "loss": 2.2482,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 2.2851,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6356976356976358e-05,
+      "loss": 2.1299,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.635031635031635e-05,
+      "loss": 2.0393,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6343656343656344e-05,
+      "loss": 2.1979,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.633699633699634e-05,
+      "loss": 2.1997,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.633033633033633e-05,
+      "loss": 2.1314,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6323676323676326e-05,
+      "loss": 2.1146,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6317016317016318e-05,
+      "loss": 2.4859,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6310356310356313e-05,
+      "loss": 2.4153,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6303696303696304e-05,
+      "loss": 2.2779,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.62970362970363e-05,
+      "loss": 2.1539,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.629037629037629e-05,
+      "loss": 2.5078,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6283716283716283e-05,
+      "loss": 2.1404,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6277056277056278e-05,
+      "loss": 2.2307,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6270396270396273e-05,
+      "loss": 2.3537,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6263736263736265e-05,
+      "loss": 2.2911,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.625707625707626e-05,
+      "loss": 2.3069,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.625041625041625e-05,
+      "loss": 1.9134,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6243756243756246e-05,
+      "loss": 2.0207,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6237096237096238e-05,
+      "loss": 2.1953,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6230436230436233e-05,
+      "loss": 2.3933,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6223776223776225e-05,
+      "loss": 2.5901,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6217116217116216e-05,
+      "loss": 2.1731,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.621045621045621e-05,
+      "loss": 2.0747,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6203796203796206e-05,
+      "loss": 2.4054,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.6197136197136198e-05,
+      "loss": 2.3329,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 2.3597,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6183816183816185e-05,
+      "loss": 2.1394,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.617715617715618e-05,
+      "loss": 2.0528,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.617049617049617e-05,
+      "loss": 2.039,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6163836163836163e-05,
+      "loss": 2.0619,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6157176157176158e-05,
+      "loss": 2.1944,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.615051615051615e-05,
+      "loss": 1.9134,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6143856143856145e-05,
+      "loss": 2.1542,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.613719613719614e-05,
+      "loss": 1.9389,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.613053613053613e-05,
+      "loss": 2.0178,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6123876123876127e-05,
+      "loss": 2.1461,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6117216117216118e-05,
+      "loss": 2.285,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6110556110556113e-05,
+      "loss": 2.1752,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6103896103896105e-05,
+      "loss": 2.2729,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6097236097236097e-05,
+      "loss": 2.1157,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6090576090576092e-05,
+      "loss": 2.3058,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6083916083916083e-05,
+      "loss": 2.0585,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.607725607725608e-05,
+      "loss": 2.2369,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.607059607059607e-05,
+      "loss": 2.0021,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6063936063936065e-05,
+      "loss": 2.0729,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.605727605727606e-05,
+      "loss": 2.3256,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6050616050616052e-05,
+      "loss": 2.0678,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6043956043956047e-05,
+      "loss": 2.1566,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.603729603729604e-05,
+      "loss": 1.9684,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.603063603063603e-05,
+      "loss": 2.303,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6023976023976025e-05,
+      "loss": 2.4482,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6017316017316017e-05,
+      "loss": 2.2525,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6010656010656012e-05,
+      "loss": 2.2937,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.6003996003996004e-05,
+      "loss": 2.2038,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5997335997336e-05,
+      "loss": 2.0672,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.5990675990675994e-05,
+      "loss": 1.9896,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5984015984015985e-05,
+      "loss": 2.5553,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.597735597735598e-05,
+      "loss": 2.4546,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5970695970695972e-05,
+      "loss": 2.5075,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5964035964035964e-05,
+      "loss": 2.3779,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.595737595737596e-05,
+      "loss": 2.0646,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.595071595071595e-05,
+      "loss": 2.5372,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5944055944055945e-05,
+      "loss": 2.525,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5937395937395937e-05,
+      "loss": 2.2203,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5930735930735932e-05,
+      "loss": 2.0597,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5924075924075927e-05,
+      "loss": 2.1772,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.591741591741592e-05,
+      "loss": 2.2899,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5910755910755914e-05,
+      "loss": 1.9592,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5904095904095906e-05,
+      "loss": 1.8019,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5897435897435897e-05,
+      "loss": 2.2747,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5890775890775892e-05,
+      "loss": 2.2167,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5884115884115884e-05,
+      "loss": 2.372,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.587745587745588e-05,
+      "loss": 2.2488,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.587079587079587e-05,
+      "loss": 1.7996,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5864135864135866e-05,
+      "loss": 2.0765,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.585747585747586e-05,
+      "loss": 2.3668,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5850815850815852e-05,
+      "loss": 2.1249,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5844155844155847e-05,
+      "loss": 2.1965,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.583749583749584e-05,
+      "loss": 2.1551,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.583083583083583e-05,
+      "loss": 2.1249,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5824175824175826e-05,
+      "loss": 2.4733,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5817515817515818e-05,
+      "loss": 2.2126,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5810855810855813e-05,
+      "loss": 1.9863,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5804195804195804e-05,
+      "loss": 1.9921,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.57975357975358e-05,
+      "loss": 2.1582,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.579087579087579e-05,
+      "loss": 2.2654,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5784215784215786e-05,
+      "loss": 2.5253,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.577755577755578e-05,
+      "loss": 2.4391,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5770895770895773e-05,
+      "loss": 2.2587,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5764235764235764e-05,
+      "loss": 2.4558,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 2.4022,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.575091575091575e-05,
+      "loss": 2.3447,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5744255744255746e-05,
+      "loss": 2.3007,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5737595737595738e-05,
+      "loss": 2.2316,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5730935730935733e-05,
+      "loss": 2.1006,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5724275724275724e-05,
+      "loss": 2.2904,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.571761571761572e-05,
+      "loss": 2.2783,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5710955710955715e-05,
+      "loss": 1.9613,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5704295704295703e-05,
+      "loss": 2.1841,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5697635697635698e-05,
+      "loss": 2.2845,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5690975690975693e-05,
+      "loss": 2.4037,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5684315684315685e-05,
+      "loss": 2.2033,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.567765567765568e-05,
+      "loss": 2.4562,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.567099567099567e-05,
+      "loss": 2.4934,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5664335664335666e-05,
+      "loss": 1.9538,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5657675657675658e-05,
+      "loss": 2.3749,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5651015651015653e-05,
+      "loss": 2.168,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5644355644355648e-05,
+      "loss": 1.8635,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5637695637695636e-05,
+      "loss": 2.2389,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.563103563103563e-05,
+      "loss": 2.1747,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5624375624375626e-05,
+      "loss": 2.1833,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5617715617715618e-05,
+      "loss": 2.4881,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5611055611055613e-05,
+      "loss": 2.204,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5604395604395605e-05,
+      "loss": 2.4888,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.55977355977356e-05,
+      "loss": 1.9936,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.559107559107559e-05,
+      "loss": 2.3733,
+      "step": 756
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5584415584415587e-05,
+      "loss": 2.0777,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.557775557775558e-05,
+      "loss": 2.407,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.557109557109557e-05,
+      "loss": 2.2239,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5564435564435565e-05,
+      "loss": 2.4847,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.555777555777556e-05,
+      "loss": 2.0636,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.555111555111555e-05,
+      "loss": 2.212,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5544455544455547e-05,
+      "loss": 2.4005,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.553779553779554e-05,
+      "loss": 2.0357,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5531135531135533e-05,
+      "loss": 2.2405,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5524475524475525e-05,
+      "loss": 2.1692,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.551781551781552e-05,
+      "loss": 1.9529,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5511155511155512e-05,
+      "loss": 2.2127,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5504495504495503e-05,
+      "loss": 2.0401,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.54978354978355e-05,
+      "loss": 2.3622,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5491175491175494e-05,
+      "loss": 2.472,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5484515484515485e-05,
+      "loss": 2.2086,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.547785547785548e-05,
+      "loss": 2.3907,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5471195471195472e-05,
+      "loss": 2.1379,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5464535464535467e-05,
+      "loss": 2.15,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.545787545787546e-05,
+      "loss": 2.2539,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5451215451215454e-05,
+      "loss": 1.9994,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5444555444555445e-05,
+      "loss": 2.1626,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5437895437895437e-05,
+      "loss": 2.3279,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5431235431235432e-05,
+      "loss": 1.8714,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5424575424575424e-05,
+      "loss": 2.2571,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.541791541791542e-05,
+      "loss": 2.0141,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5411255411255414e-05,
+      "loss": 2.3353,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5404595404595405e-05,
+      "loss": 2.3837,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.53979353979354e-05,
+      "loss": 2.0441,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5391275391275392e-05,
+      "loss": 2.1453,
+      "step": 786
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 2.4309,
+      "step": 787
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.537795537795538e-05,
+      "loss": 2.3969,
+      "step": 788
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.537129537129537e-05,
+      "loss": 2.2167,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5364635364635366e-05,
+      "loss": 2.2847,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5357975357975357e-05,
+      "loss": 2.0822,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5351315351315352e-05,
+      "loss": 2.11,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5344655344655347e-05,
+      "loss": 2.3892,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.533799533799534e-05,
+      "loss": 2.397,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5331335331335334e-05,
+      "loss": 2.4139,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5324675324675326e-05,
+      "loss": 2.2879,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.531801531801532e-05,
+      "loss": 1.9324,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5311355311355312e-05,
+      "loss": 2.0442,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5304695304695304e-05,
+      "loss": 2.3671,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.52980352980353e-05,
+      "loss": 2.4714,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.529137529137529e-05,
+      "loss": 2.4649,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5284715284715286e-05,
+      "loss": 2.3489,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.527805527805528e-05,
+      "loss": 2.3305,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5271395271395272e-05,
+      "loss": 2.379,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5264735264735268e-05,
+      "loss": 2.4106,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.525807525807526e-05,
+      "loss": 2.3545,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5251415251415254e-05,
+      "loss": 2.4417,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5244755244755244e-05,
+      "loss": 2.4266,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 1.8821,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5231435231435233e-05,
+      "loss": 1.9677,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5224775224775226e-05,
+      "loss": 2.1928,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.521811521811522e-05,
+      "loss": 2.1587,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5211455211455213e-05,
+      "loss": 2.2686,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5204795204795206e-05,
+      "loss": 2.4397,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.51981351981352e-05,
+      "loss": 2.3053,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5191475191475193e-05,
+      "loss": 2.0649,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5184815184815188e-05,
+      "loss": 2.3123,
+      "step": 817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5178155178155178e-05,
+      "loss": 2.1642,
+      "step": 818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5171495171495171e-05,
+      "loss": 2.4853,
+      "step": 819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.5164835164835166e-05,
+      "loss": 2.5845,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.515817515817516e-05,
+      "loss": 2.4331,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 2.2714,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5144855144855146e-05,
+      "loss": 2.4564,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.513819513819514e-05,
+      "loss": 1.9726,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5131535131535133e-05,
+      "loss": 2.6197,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5124875124875126e-05,
+      "loss": 2.1596,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.511821511821512e-05,
+      "loss": 2.168,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5111555111555111e-05,
+      "loss": 2.5404,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5104895104895105e-05,
+      "loss": 2.043,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.50982350982351e-05,
+      "loss": 1.9904,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5091575091575093e-05,
+      "loss": 2.1586,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5084915084915086e-05,
+      "loss": 2.1772,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.507825507825508e-05,
+      "loss": 2.3707,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5071595071595073e-05,
+      "loss": 2.0147,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5064935064935066e-05,
+      "loss": 2.2129,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.505827505827506e-05,
+      "loss": 2.718,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5051615051615053e-05,
+      "loss": 2.2389,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5044955044955045e-05,
+      "loss": 2.4735,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5038295038295038e-05,
+      "loss": 2.1387,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5031635031635032e-05,
+      "loss": 2.4229,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5024975024975027e-05,
+      "loss": 2.3892,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.501831501831502e-05,
+      "loss": 2.21,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5011655011655013e-05,
+      "loss": 1.9403,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.5004995004995007e-05,
+      "loss": 2.1753,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4998334998335e-05,
+      "loss": 2.1643,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4991674991674993e-05,
+      "loss": 2.0698,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4985014985014987e-05,
+      "loss": 2.1225,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4978354978354978e-05,
+      "loss": 1.8875,
+      "step": 848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4971694971694972e-05,
+      "loss": 2.132,
+      "step": 849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.4965034965034965e-05,
+      "loss": 2.3513,
+      "step": 850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.495837495837496e-05,
+      "loss": 2.0729,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4951714951714953e-05,
+      "loss": 2.2722,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4945054945054947e-05,
+      "loss": 2.2044,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.493839493839494e-05,
+      "loss": 1.91,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4931734931734934e-05,
+      "loss": 2.2084,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4925074925074927e-05,
+      "loss": 2.23,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.491841491841492e-05,
+      "loss": 2.0392,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4911754911754912e-05,
+      "loss": 2.1344,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4905094905094905e-05,
+      "loss": 2.4532,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4898434898434899e-05,
+      "loss": 2.0617,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4891774891774892e-05,
+      "loss": 2.5575,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4885114885114887e-05,
+      "loss": 1.7669,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.487845487845488e-05,
+      "loss": 2.1625,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4871794871794874e-05,
+      "loss": 2.0812,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4865134865134867e-05,
+      "loss": 2.5612,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.485847485847486e-05,
+      "loss": 2.122,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4851814851814852e-05,
+      "loss": 2.1065,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4845154845154845e-05,
+      "loss": 2.202,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4838494838494839e-05,
+      "loss": 2.4651,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4831834831834832e-05,
+      "loss": 2.3268,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4825174825174825e-05,
+      "loss": 2.0439,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.481851481851482e-05,
+      "loss": 2.0657,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4811854811854814e-05,
+      "loss": 2.1371,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4805194805194807e-05,
+      "loss": 1.8312,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.47985347985348e-05,
+      "loss": 2.4989,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4791874791874794e-05,
+      "loss": 2.2903,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4785214785214786e-05,
+      "loss": 2.1106,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4778554778554779e-05,
+      "loss": 1.9148,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4771894771894772e-05,
+      "loss": 2.175,
+      "step": 879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4765234765234766e-05,
+      "loss": 2.1513,
+      "step": 880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4758574758574759e-05,
+      "loss": 2.2099,
+      "step": 881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4751914751914752e-05,
+      "loss": 2.3896,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4745254745254747e-05,
+      "loss": 2.1548,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.473859473859474e-05,
+      "loss": 1.9231,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4731934731934734e-05,
+      "loss": 2.4171,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4725274725274727e-05,
+      "loss": 1.9173,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4718614718614719e-05,
+      "loss": 1.9534,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4711954711954712e-05,
+      "loss": 2.2358,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4705294705294706e-05,
+      "loss": 2.3021,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.46986346986347e-05,
+      "loss": 2.3526,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4691974691974693e-05,
+      "loss": 2.2938,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4685314685314686e-05,
+      "loss": 2.1548,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4678654678654681e-05,
+      "loss": 2.2523,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4671994671994674e-05,
+      "loss": 2.5749,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4665334665334668e-05,
+      "loss": 1.9229,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4658674658674661e-05,
+      "loss": 2.1494,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4652014652014653e-05,
+      "loss": 2.2894,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4645354645354646e-05,
+      "loss": 1.6319,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.463869463869464e-05,
+      "loss": 2.3689,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4632034632034633e-05,
+      "loss": 2.4005,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4625374625374626e-05,
+      "loss": 1.9144,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.461871461871462e-05,
+      "loss": 2.1846,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4612054612054613e-05,
+      "loss": 2.0404,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4605394605394608e-05,
+      "loss": 2.3889,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4598734598734601e-05,
+      "loss": 2.2112,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4592074592074595e-05,
+      "loss": 2.5558,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4585414585414586e-05,
+      "loss": 2.3258,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.457875457875458e-05,
+      "loss": 2.2957,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4572094572094573e-05,
+      "loss": 2.3933,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4565434565434566e-05,
+      "loss": 2.3,
+      "step": 910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.455877455877456e-05,
+      "loss": 2.1413,
+      "step": 911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4552114552114553e-05,
+      "loss": 2.2554,
+      "step": 912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 2.2615,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4538794538794541e-05,
+      "loss": 1.9527,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4532134532134535e-05,
+      "loss": 2.2875,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4525474525474528e-05,
+      "loss": 2.3463,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.451881451881452e-05,
+      "loss": 1.9386,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4512154512154513e-05,
+      "loss": 2.333,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4505494505494506e-05,
+      "loss": 2.1749,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.44988344988345e-05,
+      "loss": 2.363,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4492174492174493e-05,
+      "loss": 2.1788,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4485514485514487e-05,
+      "loss": 2.0537,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.447885447885448e-05,
+      "loss": 2.0775,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4472194472194473e-05,
+      "loss": 1.9196,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4465534465534468e-05,
+      "loss": 2.4373,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4458874458874458e-05,
+      "loss": 2.1422,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4452214452214453e-05,
+      "loss": 2.3787,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4445554445554447e-05,
+      "loss": 2.2295,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.443889443889444e-05,
+      "loss": 2.3708,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4432234432234433e-05,
+      "loss": 2.0595,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4425574425574427e-05,
+      "loss": 2.0183,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.441891441891442e-05,
+      "loss": 2.3705,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4412254412254413e-05,
+      "loss": 2.4185,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4405594405594407e-05,
+      "loss": 1.849,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4398934398934402e-05,
+      "loss": 2.0143,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4392274392274392e-05,
+      "loss": 2.4094,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4385614385614385e-05,
+      "loss": 1.9504,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.437895437895438e-05,
+      "loss": 2.0889,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4372294372294374e-05,
+      "loss": 2.1043,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4365634365634367e-05,
+      "loss": 2.1244,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.435897435897436e-05,
+      "loss": 2.1411,
+      "step": 941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4352314352314354e-05,
+      "loss": 2.0072,
+      "step": 942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.4345654345654347e-05,
+      "loss": 2.2938,
+      "step": 943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.433899433899434e-05,
+      "loss": 1.983,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4332334332334334e-05,
+      "loss": 2.3703,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4325674325674325e-05,
+      "loss": 2.2363,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4319014319014319e-05,
+      "loss": 2.1675,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4312354312354314e-05,
+      "loss": 2.2989,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4305694305694307e-05,
+      "loss": 2.4679,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.42990342990343e-05,
+      "loss": 2.2002,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4292374292374294e-05,
+      "loss": 2.1691,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 2.1063,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.427905427905428e-05,
+      "loss": 1.831,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4272394272394274e-05,
+      "loss": 2.2524,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4265734265734267e-05,
+      "loss": 2.3467,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4259074259074259e-05,
+      "loss": 2.0781,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4252414252414252e-05,
+      "loss": 2.1983,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4245754245754246e-05,
+      "loss": 2.3916,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.423909423909424e-05,
+      "loss": 1.8103,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4232434232434234e-05,
+      "loss": 1.8787,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4225774225774227e-05,
+      "loss": 2.1913,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.421911421911422e-05,
+      "loss": 2.3899,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4212454212454214e-05,
+      "loss": 2.0981,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4205794205794207e-05,
+      "loss": 2.2867,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.41991341991342e-05,
+      "loss": 2.048,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4192474192474192e-05,
+      "loss": 1.9008,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4185814185814186e-05,
+      "loss": 2.5543,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4179154179154179e-05,
+      "loss": 2.0468,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4172494172494174e-05,
+      "loss": 2.2178,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4165834165834167e-05,
+      "loss": 2.0409,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.415917415917416e-05,
+      "loss": 1.9854,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4152514152514154e-05,
+      "loss": 2.4807,
+      "step": 972
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4145854145854148e-05,
+      "loss": 1.9608,
+      "step": 973
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4139194139194141e-05,
+      "loss": 2.2175,
+      "step": 974
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.4132534132534134e-05,
+      "loss": 2.3748,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4125874125874126e-05,
+      "loss": 1.973,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.411921411921412e-05,
+      "loss": 2.2337,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4112554112554113e-05,
+      "loss": 2.3202,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4105894105894106e-05,
+      "loss": 1.6411,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4099234099234101e-05,
+      "loss": 1.7744,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4092574092574094e-05,
+      "loss": 2.1421,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4085914085914088e-05,
+      "loss": 2.1492,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4079254079254081e-05,
+      "loss": 2.1704,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4072594072594074e-05,
+      "loss": 2.1422,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4065934065934068e-05,
+      "loss": 2.3283,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.405927405927406e-05,
+      "loss": 2.1311,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4052614052614053e-05,
+      "loss": 2.4541,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4045954045954046e-05,
+      "loss": 2.0056,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.403929403929404e-05,
+      "loss": 2.4167,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4032634032634035e-05,
+      "loss": 2.0308,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4025974025974028e-05,
+      "loss": 2.0927,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4019314019314021e-05,
+      "loss": 2.208,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4012654012654015e-05,
+      "loss": 2.2109,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.4005994005994008e-05,
+      "loss": 1.6871,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3999333999334e-05,
+      "loss": 2.1865,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3992673992673993e-05,
+      "loss": 2.1017,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3986013986013986e-05,
+      "loss": 2.3476,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.397935397935398e-05,
+      "loss": 2.3577,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3972693972693973e-05,
+      "loss": 1.9452,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3966033966033966e-05,
+      "loss": 2.4596,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3959373959373961e-05,
+      "loss": 2.2069,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3952713952713955e-05,
+      "loss": 2.6863,
+      "step": 1002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3946053946053948e-05,
+      "loss": 2.1351,
+      "step": 1003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 2.0845,
+      "step": 1004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3932733932733933e-05,
+      "loss": 2.3763,
+      "step": 1005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.3926073926073927e-05,
+      "loss": 2.3115,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.391941391941392e-05,
+      "loss": 2.3597,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3912753912753913e-05,
+      "loss": 2.0739,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3906093906093907e-05,
+      "loss": 1.986,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.38994338994339e-05,
+      "loss": 2.3764,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3892773892773895e-05,
+      "loss": 2.3915,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3886113886113888e-05,
+      "loss": 2.4691,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3879453879453882e-05,
+      "loss": 1.9923,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3872793872793875e-05,
+      "loss": 1.912,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3866133866133867e-05,
+      "loss": 2.3449,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.385947385947386e-05,
+      "loss": 1.8845,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3852813852813853e-05,
+      "loss": 2.2558,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 2.1044,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.383949383949384e-05,
+      "loss": 2.0271,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3832833832833833e-05,
+      "loss": 2.0045,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3826173826173827e-05,
+      "loss": 2.3164,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3819513819513822e-05,
+      "loss": 1.9561,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3812853812853815e-05,
+      "loss": 2.3725,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3806193806193809e-05,
+      "loss": 2.0202,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.37995337995338e-05,
+      "loss": 2.2589,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3792873792873794e-05,
+      "loss": 2.0411,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3786213786213787e-05,
+      "loss": 2.1022,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.377955377955378e-05,
+      "loss": 2.0929,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3772893772893774e-05,
+      "loss": 2.1247,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3766233766233767e-05,
+      "loss": 2.248,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.375957375957376e-05,
+      "loss": 2.2473,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3752913752913755e-05,
+      "loss": 1.9959,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3746253746253749e-05,
+      "loss": 1.9993,
+      "step": 1033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3739593739593742e-05,
+      "loss": 2.0104,
+      "step": 1034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3732933732933734e-05,
+      "loss": 2.3667,
+      "step": 1035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.3726273726273727e-05,
+      "loss": 2.4507,
+      "step": 1036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.371961371961372e-05,
+      "loss": 2.2942,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3712953712953714e-05,
+      "loss": 2.2761,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3706293706293707e-05,
+      "loss": 1.7803,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.36996336996337e-05,
+      "loss": 2.5945,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3692973692973694e-05,
+      "loss": 2.2486,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3686313686313687e-05,
+      "loss": 2.0633,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3679653679653682e-05,
+      "loss": 2.4316,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3672993672993676e-05,
+      "loss": 2.2027,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3666333666333667e-05,
+      "loss": 2.1528,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.365967365967366e-05,
+      "loss": 2.2221,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3653013653013654e-05,
+      "loss": 1.9451,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3646353646353647e-05,
+      "loss": 1.9863,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.363969363969364e-05,
+      "loss": 2.1379,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3633033633033634e-05,
+      "loss": 2.2561,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3626373626373627e-05,
+      "loss": 2.0788,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.361971361971362e-05,
+      "loss": 2.2973,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3613053613053616e-05,
+      "loss": 2.2747,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3606393606393606e-05,
+      "loss": 1.9616,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3599733599733599e-05,
+      "loss": 2.29,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3593073593073594e-05,
+      "loss": 2.1506,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3586413586413588e-05,
+      "loss": 2.2742,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3579753579753581e-05,
+      "loss": 1.9203,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3573093573093574e-05,
+      "loss": 2.2063,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3566433566433568e-05,
+      "loss": 2.0703,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3559773559773561e-05,
+      "loss": 1.9804,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3553113553113554e-05,
+      "loss": 1.9916,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3546453546453548e-05,
+      "loss": 2.1919,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.353979353979354e-05,
+      "loss": 2.3887,
+      "step": 1064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3533133533133533e-05,
+      "loss": 2.2464,
+      "step": 1065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3526473526473528e-05,
+      "loss": 2.0859,
+      "step": 1066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3519813519813521e-05,
+      "loss": 1.9709,
+      "step": 1067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.3513153513153514e-05,
+      "loss": 2.2976,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3506493506493508e-05,
+      "loss": 2.1672,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3499833499833501e-05,
+      "loss": 2.1375,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3493173493173495e-05,
+      "loss": 2.2589,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3486513486513488e-05,
+      "loss": 2.3123,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3479853479853481e-05,
+      "loss": 2.0027,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3473193473193473e-05,
+      "loss": 2.0061,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3466533466533466e-05,
+      "loss": 2.3684,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.345987345987346e-05,
+      "loss": 1.9675,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3453213453213455e-05,
+      "loss": 2.1472,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3446553446553448e-05,
+      "loss": 1.8693,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3439893439893441e-05,
+      "loss": 2.2502,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3433233433233435e-05,
+      "loss": 2.2306,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3426573426573428e-05,
+      "loss": 2.2384,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3419913419913421e-05,
+      "loss": 2.0701,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3413253413253415e-05,
+      "loss": 1.9537,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3406593406593406e-05,
+      "loss": 2.2193,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.33999333999334e-05,
+      "loss": 2.0794,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3393273393273393e-05,
+      "loss": 2.3901,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3386613386613388e-05,
+      "loss": 2.0586,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3379953379953382e-05,
+      "loss": 2.1949,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3373293373293375e-05,
+      "loss": 2.3174,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3366633366633368e-05,
+      "loss": 2.2018,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3359973359973362e-05,
+      "loss": 1.8817,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3353313353313355e-05,
+      "loss": 2.6603,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3346653346653348e-05,
+      "loss": 1.9038,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.333999333999334e-05,
+      "loss": 2.1959,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 2.2264,
+      "step": 1095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3326673326673327e-05,
+      "loss": 1.9471,
+      "step": 1096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.332001332001332e-05,
+      "loss": 2.4133,
+      "step": 1097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3313353313353315e-05,
+      "loss": 2.156,
+      "step": 1098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3306693306693308e-05,
+      "loss": 1.7461,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3300033300033302e-05,
+      "loss": 2.1475,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3293373293373295e-05,
+      "loss": 2.4761,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3286713286713288e-05,
+      "loss": 2.297,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3280053280053282e-05,
+      "loss": 2.386,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3273393273393273e-05,
+      "loss": 2.1474,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3266733266733267e-05,
+      "loss": 2.0815,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.326007326007326e-05,
+      "loss": 2.176,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3253413253413254e-05,
+      "loss": 2.2993,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3246753246753249e-05,
+      "loss": 2.2363,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3240093240093242e-05,
+      "loss": 2.1168,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3233433233433235e-05,
+      "loss": 2.6152,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3226773226773229e-05,
+      "loss": 2.5099,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3220113220113222e-05,
+      "loss": 2.5464,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3213453213453215e-05,
+      "loss": 2.4192,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3206793206793207e-05,
+      "loss": 2.2192,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.32001332001332e-05,
+      "loss": 2.0045,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3193473193473194e-05,
+      "loss": 1.8347,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3186813186813187e-05,
+      "loss": 2.2567,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.318015318015318e-05,
+      "loss": 2.035,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3173493173493175e-05,
+      "loss": 2.2267,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3166833166833169e-05,
+      "loss": 2.2397,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3160173160173162e-05,
+      "loss": 2.3785,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3153513153513156e-05,
+      "loss": 1.7226,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3146853146853147e-05,
+      "loss": 1.8212,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.314019314019314e-05,
+      "loss": 1.9724,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3133533133533134e-05,
+      "loss": 2.1946,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3126873126873127e-05,
+      "loss": 2.5406,
+      "step": 1126
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.312021312021312e-05,
+      "loss": 2.3117,
+      "step": 1127
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3113553113553114e-05,
+      "loss": 2.0079,
+      "step": 1128
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3106893106893109e-05,
+      "loss": 2.1064,
+      "step": 1129
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.3100233100233102e-05,
+      "loss": 2.464,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3093573093573096e-05,
+      "loss": 2.2763,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3086913086913089e-05,
+      "loss": 2.1824,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.308025308025308e-05,
+      "loss": 2.014,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3073593073593074e-05,
+      "loss": 2.2574,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3066933066933067e-05,
+      "loss": 2.165,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.306027306027306e-05,
+      "loss": 2.1618,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3053613053613054e-05,
+      "loss": 2.2093,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3046953046953047e-05,
+      "loss": 2.233,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3040293040293041e-05,
+      "loss": 2.0023,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3033633033633036e-05,
+      "loss": 2.3294,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.302697302697303e-05,
+      "loss": 2.2855,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3020313020313023e-05,
+      "loss": 2.0542,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3013653013653014e-05,
+      "loss": 2.0647,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3006993006993008e-05,
+      "loss": 1.9396,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3000333000333001e-05,
+      "loss": 2.1364,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2993672993672994e-05,
+      "loss": 2.159,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2987012987012988e-05,
+      "loss": 2.2494,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2980352980352981e-05,
+      "loss": 2.168,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2973692973692974e-05,
+      "loss": 2.2766,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.296703296703297e-05,
+      "loss": 2.1034,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2960372960372963e-05,
+      "loss": 2.3177,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2953712953712956e-05,
+      "loss": 2.0554,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2947052947052948e-05,
+      "loss": 1.8022,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2940392940392941e-05,
+      "loss": 2.1322,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2933732933732935e-05,
+      "loss": 2.3892,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2927072927072928e-05,
+      "loss": 1.8758,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2920412920412921e-05,
+      "loss": 2.2379,
+      "step": 1157
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2913752913752915e-05,
+      "loss": 2.0984,
+      "step": 1158
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2907092907092908e-05,
+      "loss": 2.6114,
+      "step": 1159
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2900432900432901e-05,
+      "loss": 2.246,
+      "step": 1160
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.2893772893772896e-05,
+      "loss": 2.1017,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.288711288711289e-05,
+      "loss": 1.9166,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2880452880452881e-05,
+      "loss": 2.3935,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2873792873792875e-05,
+      "loss": 2.2243,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2867132867132868e-05,
+      "loss": 2.0931,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2860472860472861e-05,
+      "loss": 2.0937,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2853812853812855e-05,
+      "loss": 2.2896,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2847152847152848e-05,
+      "loss": 2.1438,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2840492840492841e-05,
+      "loss": 2.3407,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2833832833832835e-05,
+      "loss": 2.1894,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.282717282717283e-05,
+      "loss": 1.7256,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2820512820512823e-05,
+      "loss": 1.7673,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2813852813852813e-05,
+      "loss": 2.1501,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2807192807192808e-05,
+      "loss": 1.8488,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2800532800532802e-05,
+      "loss": 2.3453,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2793872793872795e-05,
+      "loss": 2.0686,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2787212787212788e-05,
+      "loss": 2.6198,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2780552780552782e-05,
+      "loss": 2.215,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2773892773892775e-05,
+      "loss": 2.1061,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2767232767232768e-05,
+      "loss": 2.3177,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2760572760572762e-05,
+      "loss": 2.3494,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2753912753912753e-05,
+      "loss": 2.5845,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2747252747252747e-05,
+      "loss": 2.11,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2740592740592742e-05,
+      "loss": 2.0759,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2733932733932735e-05,
+      "loss": 2.0447,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 2.022,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2720612720612722e-05,
+      "loss": 2.0507,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2713952713952715e-05,
+      "loss": 2.141,
+      "step": 1188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2707292707292709e-05,
+      "loss": 1.797,
+      "step": 1189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2700632700632702e-05,
+      "loss": 2.291,
+      "step": 1190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2693972693972695e-05,
+      "loss": 2.2256,
+      "step": 1191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.2687312687312687e-05,
+      "loss": 2.0305,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.268065268065268e-05,
+      "loss": 2.2179,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2673992673992674e-05,
+      "loss": 2.334,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2667332667332669e-05,
+      "loss": 2.0295,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2660672660672662e-05,
+      "loss": 2.3466,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2654012654012655e-05,
+      "loss": 1.991,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2647352647352649e-05,
+      "loss": 1.8826,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2640692640692642e-05,
+      "loss": 2.3201,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2634032634032635e-05,
+      "loss": 2.4909,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2627372627372629e-05,
+      "loss": 2.398,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.262071262071262e-05,
+      "loss": 1.9449,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2614052614052614e-05,
+      "loss": 2.1944,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2607392607392607e-05,
+      "loss": 2.5672,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2600732600732602e-05,
+      "loss": 2.1281,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2594072594072596e-05,
+      "loss": 1.989,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2587412587412589e-05,
+      "loss": 1.8355,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2580752580752582e-05,
+      "loss": 2.2391,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2574092574092576e-05,
+      "loss": 2.0018,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2567432567432569e-05,
+      "loss": 2.1906,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2560772560772562e-05,
+      "loss": 2.2617,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2554112554112554e-05,
+      "loss": 1.8165,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2547452547452547e-05,
+      "loss": 2.2907,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.254079254079254e-05,
+      "loss": 1.81,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2534132534132534e-05,
+      "loss": 2.2708,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2527472527472529e-05,
+      "loss": 1.9221,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2520812520812522e-05,
+      "loss": 1.9883,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2514152514152516e-05,
+      "loss": 2.1006,
+      "step": 1218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2507492507492509e-05,
+      "loss": 1.9644,
+      "step": 1219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2500832500832502e-05,
+      "loss": 2.2955,
+      "step": 1220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2494172494172496e-05,
+      "loss": 2.303,
+      "step": 1221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2487512487512488e-05,
+      "loss": 2.4287,
+      "step": 1222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.2480852480852481e-05,
+      "loss": 2.2955,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2474192474192474e-05,
+      "loss": 2.3142,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2467532467532468e-05,
+      "loss": 2.1359,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2460872460872463e-05,
+      "loss": 2.3016,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2454212454212456e-05,
+      "loss": 1.8438,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.244755244755245e-05,
+      "loss": 1.8841,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2440892440892443e-05,
+      "loss": 2.3419,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2434232434232436e-05,
+      "loss": 2.0053,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.242757242757243e-05,
+      "loss": 2.1556,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2420912420912421e-05,
+      "loss": 2.1308,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2414252414252414e-05,
+      "loss": 2.2188,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2407592407592408e-05,
+      "loss": 2.0338,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2400932400932401e-05,
+      "loss": 1.8144,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2394272394272394e-05,
+      "loss": 2.0513,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.238761238761239e-05,
+      "loss": 2.0329,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 2.1761,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2374292374292376e-05,
+      "loss": 2.0918,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.236763236763237e-05,
+      "loss": 2.2067,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2360972360972363e-05,
+      "loss": 1.8156,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2354312354312355e-05,
+      "loss": 1.8132,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2347652347652348e-05,
+      "loss": 2.4586,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2340992340992341e-05,
+      "loss": 2.2911,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2334332334332335e-05,
+      "loss": 2.047,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2327672327672328e-05,
+      "loss": 2.2068,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2321012321012323e-05,
+      "loss": 2.2328,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2314352314352316e-05,
+      "loss": 2.1429,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 2.1343,
+      "step": 1249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2301032301032303e-05,
+      "loss": 1.971,
+      "step": 1250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2294372294372295e-05,
+      "loss": 2.2726,
+      "step": 1251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2287712287712288e-05,
+      "loss": 2.2892,
+      "step": 1252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2281052281052281e-05,
+      "loss": 2.1933,
+      "step": 1253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.2274392274392275e-05,
+      "loss": 2.2571,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2267732267732268e-05,
+      "loss": 2.2024,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2261072261072262e-05,
+      "loss": 2.1359,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2254412254412255e-05,
+      "loss": 2.3302,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.224775224775225e-05,
+      "loss": 2.0554,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2241092241092243e-05,
+      "loss": 2.1431,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2234432234432237e-05,
+      "loss": 1.8926,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2227772227772228e-05,
+      "loss": 1.8939,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2221112221112222e-05,
+      "loss": 2.0437,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2214452214452215e-05,
+      "loss": 2.0073,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2207792207792208e-05,
+      "loss": 2.2154,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2201132201132202e-05,
+      "loss": 2.2428,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2194472194472195e-05,
+      "loss": 2.563,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2187812187812188e-05,
+      "loss": 2.3632,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2181152181152183e-05,
+      "loss": 2.24,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2174492174492177e-05,
+      "loss": 2.2675,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.216783216783217e-05,
+      "loss": 2.6083,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2161172161172162e-05,
+      "loss": 2.1922,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2154512154512155e-05,
+      "loss": 2.1364,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2147852147852149e-05,
+      "loss": 2.1506,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2141192141192142e-05,
+      "loss": 2.3915,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2134532134532135e-05,
+      "loss": 1.9927,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2127872127872129e-05,
+      "loss": 2.2943,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 1.5995,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2114552114552115e-05,
+      "loss": 2.2979,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.210789210789211e-05,
+      "loss": 2.2628,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2101232101232104e-05,
+      "loss": 2.1433,
+      "step": 1280
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2094572094572095e-05,
+      "loss": 1.8365,
+      "step": 1281
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2087912087912089e-05,
+      "loss": 2.3128,
+      "step": 1282
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2081252081252082e-05,
+      "loss": 2.2763,
+      "step": 1283
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2074592074592075e-05,
+      "loss": 2.1279,
+      "step": 1284
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.2067932067932069e-05,
+      "loss": 2.038,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2061272061272062e-05,
+      "loss": 2.1118,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2054612054612055e-05,
+      "loss": 1.9955,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2047952047952049e-05,
+      "loss": 2.0149,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2041292041292044e-05,
+      "loss": 1.7371,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2034632034632037e-05,
+      "loss": 2.2568,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2027972027972027e-05,
+      "loss": 1.9867,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2021312021312022e-05,
+      "loss": 2.2436,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2014652014652016e-05,
+      "loss": 1.8882,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2007992007992009e-05,
+      "loss": 2.0004,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2001332001332002e-05,
+      "loss": 2.1119,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1994671994671996e-05,
+      "loss": 2.0837,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1988011988011989e-05,
+      "loss": 2.3973,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1981351981351982e-05,
+      "loss": 2.362,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1974691974691976e-05,
+      "loss": 2.4077,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.196803196803197e-05,
+      "loss": 1.8011,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.196137196137196e-05,
+      "loss": 2.0254,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1954711954711956e-05,
+      "loss": 2.0712,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1948051948051949e-05,
+      "loss": 2.2674,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1941391941391942e-05,
+      "loss": 2.0957,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1934731934731936e-05,
+      "loss": 2.215,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.192807192807193e-05,
+      "loss": 2.2936,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1921411921411923e-05,
+      "loss": 1.9042,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1914751914751916e-05,
+      "loss": 1.9411,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.190809190809191e-05,
+      "loss": 2.0871,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1901431901431901e-05,
+      "loss": 2.2656,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1894771894771894e-05,
+      "loss": 2.5247,
+      "step": 1311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1888111888111888e-05,
+      "loss": 2.0572,
+      "step": 1312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1881451881451883e-05,
+      "loss": 2.2074,
+      "step": 1313
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1874791874791876e-05,
+      "loss": 2.1339,
+      "step": 1314
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.186813186813187e-05,
+      "loss": 2.2609,
+      "step": 1315
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.1861471861471863e-05,
+      "loss": 2.4527,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1854811854811856e-05,
+      "loss": 2.2356,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.184815184815185e-05,
+      "loss": 1.8799,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1841491841491843e-05,
+      "loss": 2.1835,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1834831834831834e-05,
+      "loss": 2.0912,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1828171828171828e-05,
+      "loss": 2.2099,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1821511821511821e-05,
+      "loss": 2.2068,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1814851814851816e-05,
+      "loss": 2.3415,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.180819180819181e-05,
+      "loss": 1.924,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1801531801531803e-05,
+      "loss": 1.9962,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1794871794871796e-05,
+      "loss": 2.2788,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.178821178821179e-05,
+      "loss": 1.8041,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1781551781551783e-05,
+      "loss": 1.973,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1774891774891776e-05,
+      "loss": 2.0802,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1768231768231768e-05,
+      "loss": 2.539,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1761571761571761e-05,
+      "loss": 2.1723,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1754911754911755e-05,
+      "loss": 2.1162,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1748251748251748e-05,
+      "loss": 2.0692,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1741591741591743e-05,
+      "loss": 1.9945,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1734931734931736e-05,
+      "loss": 2.1231,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.172827172827173e-05,
+      "loss": 2.0236,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1721611721611723e-05,
+      "loss": 2.3007,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1714951714951717e-05,
+      "loss": 2.1058,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.170829170829171e-05,
+      "loss": 2.3902,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1701631701631702e-05,
+      "loss": 2.1444,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1694971694971695e-05,
+      "loss": 1.8472,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1688311688311688e-05,
+      "loss": 2.2286,
+      "step": 1342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1681651681651682e-05,
+      "loss": 1.8942,
+      "step": 1343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1674991674991677e-05,
+      "loss": 2.4263,
+      "step": 1344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.166833166833167e-05,
+      "loss": 2.3779,
+      "step": 1345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.1661671661671663e-05,
+      "loss": 1.9359,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1655011655011657e-05,
+      "loss": 2.1994,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.164835164835165e-05,
+      "loss": 2.1203,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1641691641691643e-05,
+      "loss": 1.8326,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1635031635031635e-05,
+      "loss": 2.2201,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1628371628371628e-05,
+      "loss": 2.2446,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1621711621711622e-05,
+      "loss": 2.1797,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1615051615051615e-05,
+      "loss": 2.2613,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1608391608391608e-05,
+      "loss": 2.1367,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1601731601731604e-05,
+      "loss": 2.3304,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1595071595071597e-05,
+      "loss": 1.6237,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.158841158841159e-05,
+      "loss": 2.2526,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1581751581751584e-05,
+      "loss": 2.2389,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1575091575091577e-05,
+      "loss": 2.3868,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1568431568431569e-05,
+      "loss": 2.2093,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1561771561771562e-05,
+      "loss": 2.0288,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1555111555111555e-05,
+      "loss": 2.3129,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1548451548451549e-05,
+      "loss": 2.3318,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1541791541791542e-05,
+      "loss": 2.0441,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1535131535131537e-05,
+      "loss": 2.181,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.152847152847153e-05,
+      "loss": 1.9566,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1521811521811524e-05,
+      "loss": 2.1167,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 1.7443,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.150849150849151e-05,
+      "loss": 1.8582,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1501831501831502e-05,
+      "loss": 2.0387,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1495171495171495e-05,
+      "loss": 2.2773,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1488511488511489e-05,
+      "loss": 2.0942,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1481851481851482e-05,
+      "loss": 2.0868,
+      "step": 1373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1475191475191476e-05,
+      "loss": 1.9966,
+      "step": 1374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1468531468531469e-05,
+      "loss": 1.9732,
+      "step": 1375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1461871461871464e-05,
+      "loss": 2.1093,
+      "step": 1376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.1455211455211457e-05,
+      "loss": 2.3416,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.144855144855145e-05,
+      "loss": 2.5861,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1441891441891442e-05,
+      "loss": 1.9947,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1435231435231436e-05,
+      "loss": 1.8078,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.8856,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1421911421911422e-05,
+      "loss": 2.0693,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1415251415251416e-05,
+      "loss": 1.7312,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1408591408591409e-05,
+      "loss": 1.7028,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1401931401931402e-05,
+      "loss": 2.2813,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1395271395271397e-05,
+      "loss": 2.2063,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.138861138861139e-05,
+      "loss": 2.0004,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1381951381951384e-05,
+      "loss": 2.1554,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1375291375291376e-05,
+      "loss": 2.5723,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.136863136863137e-05,
+      "loss": 2.0047,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1361971361971363e-05,
+      "loss": 2.0277,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1355311355311356e-05,
+      "loss": 2.1265,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.134865134865135e-05,
+      "loss": 1.8693,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1341991341991343e-05,
+      "loss": 2.2237,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1335331335331336e-05,
+      "loss": 2.123,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.132867132867133e-05,
+      "loss": 2.4905,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1322011322011324e-05,
+      "loss": 2.4723,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1315351315351318e-05,
+      "loss": 1.9856,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.130869130869131e-05,
+      "loss": 1.8603,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1302031302031303e-05,
+      "loss": 2.576,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1295371295371296e-05,
+      "loss": 2.2682,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.128871128871129e-05,
+      "loss": 1.8814,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1282051282051283e-05,
+      "loss": 1.6653,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1275391275391276e-05,
+      "loss": 2.6116,
+      "step": 1404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.126873126873127e-05,
+      "loss": 2.2326,
+      "step": 1405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1262071262071263e-05,
+      "loss": 2.2363,
+      "step": 1406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1255411255411258e-05,
+      "loss": 1.8594,
+      "step": 1407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1248751248751251e-05,
+      "loss": 1.9456,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1242091242091241e-05,
+      "loss": 2.0055,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1235431235431236e-05,
+      "loss": 2.2969,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.122877122877123e-05,
+      "loss": 2.0125,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1222111222111223e-05,
+      "loss": 2.5628,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1215451215451216e-05,
+      "loss": 2.2179,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.120879120879121e-05,
+      "loss": 2.1653,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1202131202131203e-05,
+      "loss": 2.1464,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1195471195471196e-05,
+      "loss": 2.0812,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.118881118881119e-05,
+      "loss": 2.0977,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1182151182151185e-05,
+      "loss": 2.455,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1175491175491175e-05,
+      "loss": 2.2992,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.116883116883117e-05,
+      "loss": 2.4595,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1162171162171163e-05,
+      "loss": 2.1294,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1155511155511157e-05,
+      "loss": 1.652,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.114885114885115e-05,
+      "loss": 1.8387,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1142191142191143e-05,
+      "loss": 2.1524,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1135531135531137e-05,
+      "loss": 1.8971,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.112887112887113e-05,
+      "loss": 2.0648,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1122211122211123e-05,
+      "loss": 2.0216,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1115551115551118e-05,
+      "loss": 2.0277,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1108891108891108e-05,
+      "loss": 2.0744,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1102231102231102e-05,
+      "loss": 2.3353,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1095571095571097e-05,
+      "loss": 2.2377,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.108891108891109e-05,
+      "loss": 2.6385,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1082251082251083e-05,
+      "loss": 1.7911,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1075591075591077e-05,
+      "loss": 1.9705,
+      "step": 1434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.106893106893107e-05,
+      "loss": 2.0622,
+      "step": 1435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1062271062271063e-05,
+      "loss": 2.2829,
+      "step": 1436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1055611055611057e-05,
+      "loss": 1.8338,
+      "step": 1437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1048951048951048e-05,
+      "loss": 1.9204,
+      "step": 1438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1042291042291042e-05,
+      "loss": 2.0787,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1035631035631035e-05,
+      "loss": 2.0396,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.102897102897103e-05,
+      "loss": 2.3461,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1022311022311024e-05,
+      "loss": 2.1079,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1015651015651017e-05,
+      "loss": 2.0394,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.100899100899101e-05,
+      "loss": 2.1323,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1002331002331004e-05,
+      "loss": 1.715,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0995670995670997e-05,
+      "loss": 2.5268,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.098901098901099e-05,
+      "loss": 2.045,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0982350982350982e-05,
+      "loss": 2.1916,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0975690975690975e-05,
+      "loss": 2.0818,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0969030969030969e-05,
+      "loss": 2.0022,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0962370962370962e-05,
+      "loss": 2.2838,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0955710955710957e-05,
+      "loss": 1.9916,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.094905094905095e-05,
+      "loss": 1.9904,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0942390942390944e-05,
+      "loss": 2.0291,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0935730935730937e-05,
+      "loss": 2.1923,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.092907092907093e-05,
+      "loss": 2.0492,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0922410922410924e-05,
+      "loss": 2.1624,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0915750915750916e-05,
+      "loss": 1.975,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.9666,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0902430902430902e-05,
+      "loss": 2.141,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0895770895770896e-05,
+      "loss": 2.662,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.088911088911089e-05,
+      "loss": 1.8713,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0882450882450884e-05,
+      "loss": 2.0826,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0875790875790877e-05,
+      "loss": 2.3436,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.086913086913087e-05,
+      "loss": 2.203,
+      "step": 1465
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0862470862470864e-05,
+      "loss": 1.9594,
+      "step": 1466
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0855810855810857e-05,
+      "loss": 1.9079,
+      "step": 1467
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0849150849150849e-05,
+      "loss": 1.7283,
+      "step": 1468
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0842490842490842e-05,
+      "loss": 1.8342,
+      "step": 1469
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.0835830835830836e-05,
+      "loss": 2.022,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0829170829170829e-05,
+      "loss": 2.1569,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0822510822510823e-05,
+      "loss": 2.1092,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0815850815850818e-05,
+      "loss": 2.0548,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0809190809190811e-05,
+      "loss": 1.9096,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0802530802530804e-05,
+      "loss": 1.9234,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0795870795870798e-05,
+      "loss": 1.943,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0789210789210791e-05,
+      "loss": 1.9661,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0782550782550783e-05,
+      "loss": 2.2025,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0775890775890776e-05,
+      "loss": 2.5082,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 1.9282,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0762570762570763e-05,
+      "loss": 2.0934,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0755910755910756e-05,
+      "loss": 1.9589,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0749250749250751e-05,
+      "loss": 1.9494,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0742590742590744e-05,
+      "loss": 1.9719,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0735930735930738e-05,
+      "loss": 1.9635,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0729270729270731e-05,
+      "loss": 2.4118,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0722610722610724e-05,
+      "loss": 2.1163,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0715950715950716e-05,
+      "loss": 2.0423,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.070929070929071e-05,
+      "loss": 2.1263,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0702630702630703e-05,
+      "loss": 2.1037,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0695970695970696e-05,
+      "loss": 1.8924,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.068931068931069e-05,
+      "loss": 2.0489,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0682650682650683e-05,
+      "loss": 2.3232,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0675990675990678e-05,
+      "loss": 1.9777,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0669330669330671e-05,
+      "loss": 2.0857,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0662670662670665e-05,
+      "loss": 2.131,
+      "step": 1496
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0656010656010658e-05,
+      "loss": 1.9619,
+      "step": 1497
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.064935064935065e-05,
+      "loss": 2.0753,
+      "step": 1498
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0642690642690643e-05,
+      "loss": 2.0794,
+      "step": 1499
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0636030636030636e-05,
+      "loss": 2.1306,
+      "step": 1500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.062937062937063e-05,
+      "loss": 2.0501,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0622710622710623e-05,
+      "loss": 1.8552,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0616050616050616e-05,
+      "loss": 1.9746,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0609390609390612e-05,
+      "loss": 1.7596,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0602730602730605e-05,
+      "loss": 2.0527,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0596070596070598e-05,
+      "loss": 2.2045,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.058941058941059e-05,
+      "loss": 2.403,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0582750582750583e-05,
+      "loss": 2.0032,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0576090576090577e-05,
+      "loss": 2.0957,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.056943056943057e-05,
+      "loss": 2.0223,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0562770562770563e-05,
+      "loss": 2.2442,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0556110556110557e-05,
+      "loss": 2.2231,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.054945054945055e-05,
+      "loss": 2.1746,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0542790542790543e-05,
+      "loss": 2.1007,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0536130536130538e-05,
+      "loss": 2.0811,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0529470529470532e-05,
+      "loss": 2.1658,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0522810522810523e-05,
+      "loss": 1.5915,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0516150516150517e-05,
+      "loss": 2.0572,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.050949050949051e-05,
+      "loss": 2.0452,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0502830502830503e-05,
+      "loss": 1.831,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0496170496170497e-05,
+      "loss": 2.1943,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.048951048951049e-05,
+      "loss": 2.2153,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0482850482850484e-05,
+      "loss": 2.0994,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 1.9792,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0469530469530472e-05,
+      "loss": 2.1825,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0462870462870465e-05,
+      "loss": 2.0904,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0456210456210455e-05,
+      "loss": 2.5742,
+      "step": 1527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.044955044955045e-05,
+      "loss": 1.8866,
+      "step": 1528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0442890442890444e-05,
+      "loss": 2.1725,
+      "step": 1529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0436230436230437e-05,
+      "loss": 2.2476,
+      "step": 1530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.042957042957043e-05,
+      "loss": 2.121,
+      "step": 1531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0422910422910424e-05,
+      "loss": 2.326,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0416250416250417e-05,
+      "loss": 2.3644,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040959040959041e-05,
+      "loss": 2.1097,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402930402930404e-05,
+      "loss": 2.0434,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396270396270399e-05,
+      "loss": 1.792,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389610389610389e-05,
+      "loss": 2.0806,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382950382950384e-05,
+      "loss": 2.1305,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376290376290377e-05,
+      "loss": 2.306,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036963036963037e-05,
+      "loss": 1.9748,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0362970362970364e-05,
+      "loss": 2.2127,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0356310356310357e-05,
+      "loss": 2.1115,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.034965034965035e-05,
+      "loss": 1.8751,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342990342990344e-05,
+      "loss": 2.0844,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0336330336330337e-05,
+      "loss": 1.8827,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329670329670332e-05,
+      "loss": 2.1729,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0323010323010322e-05,
+      "loss": 1.8843,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0316350316350316e-05,
+      "loss": 2.2355,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.030969030969031e-05,
+      "loss": 2.0293,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 1.9933,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0296370296370297e-05,
+      "loss": 2.216,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.028971028971029e-05,
+      "loss": 2.1696,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0283050283050284e-05,
+      "loss": 2.1192,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0276390276390277e-05,
+      "loss": 2.2361,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.026973026973027e-05,
+      "loss": 1.9504,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0263070263070264e-05,
+      "loss": 2.4274,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0256410256410256e-05,
+      "loss": 2.3444,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.024975024975025e-05,
+      "loss": 1.9472,
+      "step": 1558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0243090243090244e-05,
+      "loss": 2.265,
+      "step": 1559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0236430236430238e-05,
+      "loss": 2.4456,
+      "step": 1560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0229770229770231e-05,
+      "loss": 1.8011,
+      "step": 1561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0223110223110224e-05,
+      "loss": 1.636,
+      "step": 1562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0216450216450218e-05,
+      "loss": 2.1791,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209790209790211e-05,
+      "loss": 2.0571,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203130203130204e-05,
+      "loss": 2.0932,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196470196470196e-05,
+      "loss": 2.1403,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.018981018981019e-05,
+      "loss": 2.2295,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0183150183150183e-05,
+      "loss": 2.0444,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0176490176490176e-05,
+      "loss": 2.297,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0169830169830171e-05,
+      "loss": 2.0774,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0163170163170165e-05,
+      "loss": 1.9375,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0156510156510158e-05,
+      "loss": 1.8427,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149850149850151e-05,
+      "loss": 2.439,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143190143190145e-05,
+      "loss": 1.9373,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136530136530138e-05,
+      "loss": 1.8929,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012987012987013e-05,
+      "loss": 2.5367,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123210123210123e-05,
+      "loss": 2.1269,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116550116550116e-05,
+      "loss": 2.0459,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010989010989011e-05,
+      "loss": 2.0522,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103230103230105e-05,
+      "loss": 2.1583,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096570096570098e-05,
+      "loss": 2.2738,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089910089910091e-05,
+      "loss": 1.8103,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083250083250085e-05,
+      "loss": 2.032,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076590076590078e-05,
+      "loss": 2.083,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069930069930071e-05,
+      "loss": 1.8969,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063270063270063e-05,
+      "loss": 1.9028,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056610056610056e-05,
+      "loss": 2.0198,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004995004995005e-05,
+      "loss": 2.1802,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043290043290043e-05,
+      "loss": 2.0096,
+      "step": 1589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036630036630037e-05,
+      "loss": 1.9165,
+      "step": 1590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0029970029970032e-05,
+      "loss": 1.8645,
+      "step": 1591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0023310023310025e-05,
+      "loss": 2.1475,
+      "step": 1592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0016650016650018e-05,
+      "loss": 2.2697,
+      "step": 1593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009990009990012e-05,
+      "loss": 1.5257,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0003330003330005e-05,
+      "loss": 1.7076,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996669996669998e-06,
+      "loss": 2.1426,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990009990009992e-06,
+      "loss": 2.4466,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.983349983349983e-06,
+      "loss": 2.0488,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.976689976689977e-06,
+      "loss": 1.7435,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97002997002997e-06,
+      "loss": 2.4566,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963369963369965e-06,
+      "loss": 2.0896,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956709956709958e-06,
+      "loss": 2.3736,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95004995004995e-06,
+      "loss": 2.1591,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943389943389943e-06,
+      "loss": 2.029,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936729936729937e-06,
+      "loss": 1.8237,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93006993006993e-06,
+      "loss": 2.2001,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923409923409925e-06,
+      "loss": 2.0129,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916749916749917e-06,
+      "loss": 2.1062,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91008991008991e-06,
+      "loss": 2.363,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903429903429904e-06,
+      "loss": 1.9817,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896769896769897e-06,
+      "loss": 1.9983,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890109890109892e-06,
+      "loss": 2.4078,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883449883449884e-06,
+      "loss": 1.982,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876789876789877e-06,
+      "loss": 2.1889,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87012987012987e-06,
+      "loss": 1.8707,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863469863469864e-06,
+      "loss": 2.2914,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856809856809859e-06,
+      "loss": 2.0469,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85014985014985e-06,
+      "loss": 1.7255,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.843489843489844e-06,
+      "loss": 2.2199,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.836829836829837e-06,
+      "loss": 2.3623,
+      "step": 1620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.83016983016983e-06,
+      "loss": 1.9568,
+      "step": 1621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.823509823509826e-06,
+      "loss": 2.2277,
+      "step": 1622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.816849816849817e-06,
+      "loss": 2.1325,
+      "step": 1623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.81018981018981e-06,
+      "loss": 1.936,
+      "step": 1624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803529803529804e-06,
+      "loss": 2.096,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.796869796869797e-06,
+      "loss": 2.1472,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.79020979020979e-06,
+      "loss": 1.9745,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.783549783549784e-06,
+      "loss": 2.0098,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.776889776889777e-06,
+      "loss": 2.2874,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.77022977022977e-06,
+      "loss": 2.0592,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.763569763569764e-06,
+      "loss": 1.8628,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.756909756909757e-06,
+      "loss": 2.3414,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.75024975024975e-06,
+      "loss": 1.9691,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 2.1297,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.736929736929737e-06,
+      "loss": 2.4334,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.73026973026973e-06,
+      "loss": 1.8556,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.723609723609724e-06,
+      "loss": 2.127,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.716949716949717e-06,
+      "loss": 2.0858,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.710289710289711e-06,
+      "loss": 2.3815,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.703629703629704e-06,
+      "loss": 1.9915,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 1.9445,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.690309690309691e-06,
+      "loss": 2.0394,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.683649683649684e-06,
+      "loss": 2.1391,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676989676989678e-06,
+      "loss": 2.195,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670329670329671e-06,
+      "loss": 2.1066,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663669663669664e-06,
+      "loss": 2.1345,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657009657009658e-06,
+      "loss": 1.918,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.650349650349651e-06,
+      "loss": 2.1256,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.643689643689644e-06,
+      "loss": 2.0144,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637029637029638e-06,
+      "loss": 2.039,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630369630369631e-06,
+      "loss": 2.0765,
+      "step": 1651
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623709623709624e-06,
+      "loss": 2.1114,
+      "step": 1652
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617049617049618e-06,
+      "loss": 1.9351,
+      "step": 1653
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610389610389611e-06,
+      "loss": 2.0967,
+      "step": 1654
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603729603729605e-06,
+      "loss": 1.8287,
+      "step": 1655
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597069597069598e-06,
+      "loss": 1.9316,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.590409590409591e-06,
+      "loss": 2.3989,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.583749583749585e-06,
+      "loss": 1.8451,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.577089577089578e-06,
+      "loss": 2.2964,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.570429570429571e-06,
+      "loss": 1.9967,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.563769563769565e-06,
+      "loss": 1.8196,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.557109557109558e-06,
+      "loss": 2.119,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.550449550449551e-06,
+      "loss": 2.3902,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.543789543789545e-06,
+      "loss": 2.0843,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.537129537129538e-06,
+      "loss": 2.1566,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.530469530469531e-06,
+      "loss": 2.2454,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 2.1896,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.517149517149518e-06,
+      "loss": 2.0342,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.510489510489511e-06,
+      "loss": 1.9779,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.503829503829505e-06,
+      "loss": 2.2324,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.497169497169498e-06,
+      "loss": 1.8206,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.490509490509492e-06,
+      "loss": 1.9678,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.483849483849485e-06,
+      "loss": 2.1228,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.477189477189478e-06,
+      "loss": 1.986,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.470529470529472e-06,
+      "loss": 2.2215,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.463869463869465e-06,
+      "loss": 2.0035,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.457209457209458e-06,
+      "loss": 2.1121,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.450549450549452e-06,
+      "loss": 2.1819,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.443889443889445e-06,
+      "loss": 1.9543,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.437229437229438e-06,
+      "loss": 2.0793,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.430569430569432e-06,
+      "loss": 2.4374,
+      "step": 1681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.423909423909423e-06,
+      "loss": 2.261,
+      "step": 1682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.417249417249418e-06,
+      "loss": 2.3722,
+      "step": 1683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.410589410589412e-06,
+      "loss": 2.2755,
+      "step": 1684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.403929403929405e-06,
+      "loss": 2.0948,
+      "step": 1685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.397269397269398e-06,
+      "loss": 1.6642,
+      "step": 1686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.39060939060939e-06,
+      "loss": 1.8549,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.383949383949385e-06,
+      "loss": 2.4488,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.377289377289379e-06,
+      "loss": 2.3024,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.370629370629372e-06,
+      "loss": 1.98,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.363969363969365e-06,
+      "loss": 2.1093,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.357309357309357e-06,
+      "loss": 1.7813,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.350649350649352e-06,
+      "loss": 1.9949,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.343989343989345e-06,
+      "loss": 2.1844,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.337329337329339e-06,
+      "loss": 1.7842,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.330669330669332e-06,
+      "loss": 1.9444,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.324009324009324e-06,
+      "loss": 1.9221,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.317349317349319e-06,
+      "loss": 2.1786,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.310689310689312e-06,
+      "loss": 2.0656,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.304029304029305e-06,
+      "loss": 2.1657,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.297369297369299e-06,
+      "loss": 1.7895,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.29070929070929e-06,
+      "loss": 2.1137,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.284049284049284e-06,
+      "loss": 2.5451,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.277389277389279e-06,
+      "loss": 2.2624,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.270729270729272e-06,
+      "loss": 2.3661,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.264069264069266e-06,
+      "loss": 2.0289,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.257409257409257e-06,
+      "loss": 1.8541,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.25074925074925e-06,
+      "loss": 1.891,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.244089244089246e-06,
+      "loss": 1.916,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.237429237429239e-06,
+      "loss": 2.111,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 2.086,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.224109224109224e-06,
+      "loss": 2.0985,
+      "step": 1712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.217449217449217e-06,
+      "loss": 2.1109,
+      "step": 1713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.210789210789212e-06,
+      "loss": 2.0429,
+      "step": 1714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.204129204129206e-06,
+      "loss": 2.2983,
+      "step": 1715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.197469197469199e-06,
+      "loss": 2.0742,
+      "step": 1716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.19080919080919e-06,
+      "loss": 1.9567,
+      "step": 1717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.184149184149184e-06,
+      "loss": 2.5652,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.177489177489179e-06,
+      "loss": 2.2763,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.170829170829172e-06,
+      "loss": 1.8635,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.164169164169166e-06,
+      "loss": 2.1494,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.157509157509158e-06,
+      "loss": 2.2028,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.150849150849151e-06,
+      "loss": 2.2006,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.144189144189144e-06,
+      "loss": 2.1821,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.13752913752914e-06,
+      "loss": 1.9682,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.130869130869131e-06,
+      "loss": 1.6484,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.124209124209124e-06,
+      "loss": 2.2031,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.117549117549118e-06,
+      "loss": 2.2507,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.110889110889111e-06,
+      "loss": 2.1434,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.104229104229106e-06,
+      "loss": 1.7589,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.097569097569098e-06,
+      "loss": 2.371,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 2.4206,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.084249084249084e-06,
+      "loss": 1.8061,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.077589077589078e-06,
+      "loss": 2.2299,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.070929070929073e-06,
+      "loss": 1.9105,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.064269064269064e-06,
+      "loss": 1.9142,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.057609057609058e-06,
+      "loss": 2.477,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.050949050949051e-06,
+      "loss": 2.3917,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.044289044289045e-06,
+      "loss": 2.1284,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.03762903762904e-06,
+      "loss": 2.1435,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.030969030969031e-06,
+      "loss": 1.7644,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.024309024309025e-06,
+      "loss": 2.1091,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.017649017649018e-06,
+      "loss": 2.3801,
+      "step": 1743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.010989010989011e-06,
+      "loss": 2.2018,
+      "step": 1744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.004329004329005e-06,
+      "loss": 2.1205,
+      "step": 1745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.997668997668998e-06,
+      "loss": 2.0873,
+      "step": 1746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.991008991008991e-06,
+      "loss": 1.7479,
+      "step": 1747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.984348984348985e-06,
+      "loss": 2.3178,
+      "step": 1748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.977688977688978e-06,
+      "loss": 2.0438,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.971028971028971e-06,
+      "loss": 2.2407,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.964368964368965e-06,
+      "loss": 2.0516,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.957708957708958e-06,
+      "loss": 2.267,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.951048951048951e-06,
+      "loss": 2.1322,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.944388944388945e-06,
+      "loss": 2.1086,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.937728937728938e-06,
+      "loss": 1.8868,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.931068931068932e-06,
+      "loss": 2.2892,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.924408924408925e-06,
+      "loss": 2.1755,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.917748917748918e-06,
+      "loss": 1.8424,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.911088911088912e-06,
+      "loss": 2.0821,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.904428904428905e-06,
+      "loss": 2.049,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.897768897768898e-06,
+      "loss": 2.4319,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.891108891108892e-06,
+      "loss": 1.7835,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.884448884448885e-06,
+      "loss": 2.0279,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.877788877788878e-06,
+      "loss": 2.1372,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.871128871128872e-06,
+      "loss": 2.1454,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.864468864468865e-06,
+      "loss": 2.0046,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.857808857808858e-06,
+      "loss": 1.9477,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.851148851148852e-06,
+      "loss": 2.2396,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.844488844488845e-06,
+      "loss": 1.8351,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.837828837828838e-06,
+      "loss": 1.4372,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.831168831168832e-06,
+      "loss": 2.41,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.824508824508825e-06,
+      "loss": 2.1526,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.817848817848819e-06,
+      "loss": 1.8931,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.811188811188812e-06,
+      "loss": 2.0453,
+      "step": 1774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.804528804528805e-06,
+      "loss": 1.8744,
+      "step": 1775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.797868797868799e-06,
+      "loss": 1.8456,
+      "step": 1776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.791208791208792e-06,
+      "loss": 2.3725,
+      "step": 1777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.784548784548785e-06,
+      "loss": 2.1805,
+      "step": 1778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.777888777888779e-06,
+      "loss": 2.1454,
+      "step": 1779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.771228771228772e-06,
+      "loss": 1.6537,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.764568764568765e-06,
+      "loss": 1.9107,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.757908757908759e-06,
+      "loss": 1.8195,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.751248751248752e-06,
+      "loss": 2.2631,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.744588744588745e-06,
+      "loss": 2.1095,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.737928737928739e-06,
+      "loss": 2.2307,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.731268731268732e-06,
+      "loss": 1.8325,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.724608724608725e-06,
+      "loss": 1.8182,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.9485,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.711288711288712e-06,
+      "loss": 2.5901,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.704628704628706e-06,
+      "loss": 2.116,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.697968697968699e-06,
+      "loss": 1.9233,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.691308691308692e-06,
+      "loss": 1.6485,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.684648684648686e-06,
+      "loss": 2.0888,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.677988677988679e-06,
+      "loss": 1.8531,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.671328671328672e-06,
+      "loss": 1.7037,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.664668664668666e-06,
+      "loss": 1.8387,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.658008658008659e-06,
+      "loss": 2.1438,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.651348651348652e-06,
+      "loss": 2.2822,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.644688644688646e-06,
+      "loss": 2.5913,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.638028638028637e-06,
+      "loss": 2.2545,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.631368631368632e-06,
+      "loss": 2.0146,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.624708624708626e-06,
+      "loss": 2.3756,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.618048618048619e-06,
+      "loss": 2.2038,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.611388611388612e-06,
+      "loss": 2.0386,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.604728604728604e-06,
+      "loss": 2.0614,
+      "step": 1805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.5980685980686e-06,
+      "loss": 1.9987,
+      "step": 1806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.591408591408593e-06,
+      "loss": 2.2711,
+      "step": 1807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.584748584748586e-06,
+      "loss": 2.083,
+      "step": 1808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.57808857808858e-06,
+      "loss": 1.6978,
+      "step": 1809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.8692,
+      "step": 1810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.564768564768566e-06,
+      "loss": 1.9946,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.55810855810856e-06,
+      "loss": 1.9898,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.551448551448553e-06,
+      "loss": 1.7614,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.544788544788546e-06,
+      "loss": 1.8388,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.538128538128538e-06,
+      "loss": 1.9065,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.531468531468533e-06,
+      "loss": 1.9913,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.524808524808526e-06,
+      "loss": 1.9819,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.51814851814852e-06,
+      "loss": 2.1861,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.511488511488513e-06,
+      "loss": 1.7676,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.504828504828504e-06,
+      "loss": 2.2753,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.498168498168498e-06,
+      "loss": 2.0206,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.491508491508493e-06,
+      "loss": 1.9448,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.8874,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.47818847818848e-06,
+      "loss": 1.8022,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.471528471528471e-06,
+      "loss": 2.341,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.464868464868465e-06,
+      "loss": 1.7421,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.45820845820846e-06,
+      "loss": 2.2535,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.451548451548453e-06,
+      "loss": 2.4063,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.444888444888446e-06,
+      "loss": 2.1698,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.438228438228438e-06,
+      "loss": 2.2852,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.431568431568431e-06,
+      "loss": 1.9725,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.424908424908426e-06,
+      "loss": 2.067,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.41824841824842e-06,
+      "loss": 2.0827,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.411588411588413e-06,
+      "loss": 2.1847,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.404928404928405e-06,
+      "loss": 1.7281,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.398268398268398e-06,
+      "loss": 2.1471,
+      "step": 1836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.391608391608393e-06,
+      "loss": 1.9428,
+      "step": 1837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.384948384948387e-06,
+      "loss": 2.2487,
+      "step": 1838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.37828837828838e-06,
+      "loss": 2.1359,
+      "step": 1839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.371628371628372e-06,
+      "loss": 2.2877,
+      "step": 1840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.364968364968365e-06,
+      "loss": 1.928,
+      "step": 1841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.358308358308358e-06,
+      "loss": 2.0802,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.351648351648353e-06,
+      "loss": 2.1018,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.344988344988347e-06,
+      "loss": 2.0342,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.338328338328338e-06,
+      "loss": 1.6845,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.331668331668332e-06,
+      "loss": 2.0924,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.325008325008325e-06,
+      "loss": 1.8375,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.31834831834832e-06,
+      "loss": 2.4988,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.311688311688313e-06,
+      "loss": 1.8094,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.305028305028305e-06,
+      "loss": 2.0692,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.298368298368298e-06,
+      "loss": 2.0697,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.291708291708292e-06,
+      "loss": 1.9989,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.285048285048287e-06,
+      "loss": 2.0785,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.278388278388278e-06,
+      "loss": 2.1774,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.271728271728272e-06,
+      "loss": 2.298,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.265068265068265e-06,
+      "loss": 1.9716,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.258408258408259e-06,
+      "loss": 1.9825,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.251748251748254e-06,
+      "loss": 1.7394,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.245088245088245e-06,
+      "loss": 2.1175,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.238428238428239e-06,
+      "loss": 2.0039,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.231768231768232e-06,
+      "loss": 2.2883,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.225108225108225e-06,
+      "loss": 1.897,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.218448218448219e-06,
+      "loss": 1.8254,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.211788211788212e-06,
+      "loss": 2.5904,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.8959,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.198468198468199e-06,
+      "loss": 2.2237,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.191808191808192e-06,
+      "loss": 1.7984,
+      "step": 1867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.185148185148185e-06,
+      "loss": 2.3891,
+      "step": 1868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.178488178488179e-06,
+      "loss": 1.9485,
+      "step": 1869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.171828171828172e-06,
+      "loss": 1.9702,
+      "step": 1870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.165168165168165e-06,
+      "loss": 2.0711,
+      "step": 1871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.158508158508159e-06,
+      "loss": 1.9726,
+      "step": 1872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.151848151848152e-06,
+      "loss": 2.4355,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.145188145188146e-06,
+      "loss": 2.0423,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.138528138528139e-06,
+      "loss": 2.1993,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.131868131868132e-06,
+      "loss": 2.1565,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.125208125208126e-06,
+      "loss": 2.5899,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.118548118548119e-06,
+      "loss": 2.0221,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.111888111888112e-06,
+      "loss": 1.8381,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.105228105228106e-06,
+      "loss": 2.2077,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.098568098568099e-06,
+      "loss": 2.2171,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.091908091908092e-06,
+      "loss": 2.1034,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.085248085248086e-06,
+      "loss": 2.24,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.078588078588079e-06,
+      "loss": 2.0161,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.071928071928072e-06,
+      "loss": 2.3173,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.065268065268066e-06,
+      "loss": 2.1335,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.058608058608059e-06,
+      "loss": 1.9346,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.051948051948052e-06,
+      "loss": 1.9104,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.045288045288046e-06,
+      "loss": 1.9039,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.03862803862804e-06,
+      "loss": 1.9744,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.031968031968033e-06,
+      "loss": 2.1639,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.025308025308026e-06,
+      "loss": 2.1034,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.01864801864802e-06,
+      "loss": 2.0815,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.011988011988013e-06,
+      "loss": 2.1652,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 8.005328005328006e-06,
+      "loss": 2.2216,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.998667998668e-06,
+      "loss": 2.0791,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.992007992007993e-06,
+      "loss": 2.1338,
+      "step": 1897
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.985347985347986e-06,
+      "loss": 2.1649,
+      "step": 1898
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.97868797868798e-06,
+      "loss": 2.0304,
+      "step": 1899
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.972027972027973e-06,
+      "loss": 1.8956,
+      "step": 1900
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.965367965367966e-06,
+      "loss": 1.6164,
+      "step": 1901
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.95870795870796e-06,
+      "loss": 2.1785,
+      "step": 1902
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.952047952047953e-06,
+      "loss": 2.0962,
+      "step": 1903
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.945387945387946e-06,
+      "loss": 2.1847,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.93872793872794e-06,
+      "loss": 2.1237,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.932067932067933e-06,
+      "loss": 1.9569,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.925407925407926e-06,
+      "loss": 1.9581,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.91874791874792e-06,
+      "loss": 2.3031,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.912087912087913e-06,
+      "loss": 2.1474,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.905427905427906e-06,
+      "loss": 1.8613,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.8987678987679e-06,
+      "loss": 1.8432,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.892107892107893e-06,
+      "loss": 2.1363,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.885447885447886e-06,
+      "loss": 2.2437,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 1.949,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.872127872127873e-06,
+      "loss": 1.7083,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.865467865467866e-06,
+      "loss": 1.6696,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.85880785880786e-06,
+      "loss": 2.0141,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.852147852147851e-06,
+      "loss": 2.1223,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.845487845487846e-06,
+      "loss": 2.2926,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.83882783882784e-06,
+      "loss": 1.8707,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.832167832167833e-06,
+      "loss": 2.0217,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.825507825507827e-06,
+      "loss": 1.9453,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.818847818847818e-06,
+      "loss": 2.0473,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.812187812187813e-06,
+      "loss": 2.0643,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.805527805527807e-06,
+      "loss": 2.0029,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.7988677988678e-06,
+      "loss": 2.225,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.792207792207793e-06,
+      "loss": 1.7514,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.785547785547785e-06,
+      "loss": 2.0127,
+      "step": 1928
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.77888777888778e-06,
+      "loss": 2.4357,
+      "step": 1929
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.772227772227773e-06,
+      "loss": 2.3409,
+      "step": 1930
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.765567765567767e-06,
+      "loss": 1.9882,
+      "step": 1931
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.75890775890776e-06,
+      "loss": 1.8991,
+      "step": 1932
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.752247752247752e-06,
+      "loss": 2.0804,
+      "step": 1933
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.745587745587747e-06,
+      "loss": 1.9466,
+      "step": 1934
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.73892773892774e-06,
+      "loss": 1.9829,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.732267732267733e-06,
+      "loss": 1.9736,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.725607725607727e-06,
+      "loss": 1.8735,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.718947718947718e-06,
+      "loss": 1.9332,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.712287712287712e-06,
+      "loss": 2.0701,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.705627705627707e-06,
+      "loss": 2.2052,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.6989676989677e-06,
+      "loss": 1.773,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.8403,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.685647685647685e-06,
+      "loss": 2.1394,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.678987678987679e-06,
+      "loss": 2.3473,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.672327672327674e-06,
+      "loss": 2.28,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.665667665667667e-06,
+      "loss": 1.799,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.65900765900766e-06,
+      "loss": 2.4687,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.652347652347652e-06,
+      "loss": 1.9487,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.645687645687645e-06,
+      "loss": 2.3072,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.63902763902764e-06,
+      "loss": 1.8621,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.632367632367634e-06,
+      "loss": 2.059,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.625707625707627e-06,
+      "loss": 2.1722,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 2.2433,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.612387612387613e-06,
+      "loss": 2.1816,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.605727605727606e-06,
+      "loss": 2.2429,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5990675990676e-06,
+      "loss": 2.2073,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.592407592407594e-06,
+      "loss": 2.4764,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5857475857475856e-06,
+      "loss": 2.0856,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.57908757908758e-06,
+      "loss": 2.0753,
+      "step": 1959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.572427572427573e-06,
+      "loss": 1.8964,
+      "step": 1960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5657675657675665e-06,
+      "loss": 1.9931,
+      "step": 1961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.55910755910756e-06,
+      "loss": 2.1012,
+      "step": 1962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.552447552447552e-06,
+      "loss": 2.2147,
+      "step": 1963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.5457875457875465e-06,
+      "loss": 1.7878,
+      "step": 1964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.53912753912754e-06,
+      "loss": 2.1167,
+      "step": 1965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.532467532467533e-06,
+      "loss": 2.5965,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.525807525807527e-06,
+      "loss": 1.8384,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.519147519147519e-06,
+      "loss": 1.8408,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.512487512487513e-06,
+      "loss": 1.9863,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.505827505827507e-06,
+      "loss": 2.5059,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.4991674991675e-06,
+      "loss": 2.243,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.492507492507493e-06,
+      "loss": 2.0347,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.485847485847486e-06,
+      "loss": 2.0064,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.47918747918748e-06,
+      "loss": 1.8515,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.472527472527473e-06,
+      "loss": 2.0183,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.465867465867467e-06,
+      "loss": 1.9937,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.45920745920746e-06,
+      "loss": 2.3394,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.452547452547453e-06,
+      "loss": 2.0402,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.445887445887446e-06,
+      "loss": 2.2708,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.43922743922744e-06,
+      "loss": 1.8693,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.4325674325674335e-06,
+      "loss": 2.1657,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.425907425907426e-06,
+      "loss": 2.6521,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.419247419247419e-06,
+      "loss": 2.0531,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.412587412587413e-06,
+      "loss": 1.6797,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.405927405927407e-06,
+      "loss": 1.8407,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.3992673992674e-06,
+      "loss": 1.9251,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.392607392607393e-06,
+      "loss": 1.8448,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.385947385947386e-06,
+      "loss": 1.9959,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.3792873792873795e-06,
+      "loss": 2.38,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.372627372627374e-06,
+      "loss": 2.0544,
+      "step": 1990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.365967365967367e-06,
+      "loss": 2.1017,
+      "step": 1991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.3593073593073596e-06,
+      "loss": 2.2032,
+      "step": 1992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.352647352647353e-06,
+      "loss": 2.0122,
+      "step": 1993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.345987345987346e-06,
+      "loss": 1.9974,
+      "step": 1994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.3393273393273405e-06,
+      "loss": 2.3256,
+      "step": 1995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.332667332667334e-06,
+      "loss": 2.1078,
+      "step": 1996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.326007326007326e-06,
+      "loss": 2.0276,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.31934731934732e-06,
+      "loss": 2.1209,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.312687312687313e-06,
+      "loss": 1.8243,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.306027306027306e-06,
+      "loss": 1.7992,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.299367299367301e-06,
+      "loss": 2.0845,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.292707292707293e-06,
+      "loss": 1.9788,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.2860472860472865e-06,
+      "loss": 2.2982,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.27938727938728e-06,
+      "loss": 2.0549,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 2.2485,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.266067266067267e-06,
+      "loss": 2.2766,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.25940725940726e-06,
+      "loss": 2.1436,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.252747252747253e-06,
+      "loss": 2.075,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.246087246087247e-06,
+      "loss": 1.7549,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.23942723942724e-06,
+      "loss": 1.8957,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.232767232767234e-06,
+      "loss": 2.3391,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.226107226107227e-06,
+      "loss": 1.8357,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.21944721944722e-06,
+      "loss": 2.2902,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.212787212787213e-06,
+      "loss": 1.9766,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.206127206127207e-06,
+      "loss": 1.9861,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.199467199467201e-06,
+      "loss": 2.0153,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.1928071928071926e-06,
+      "loss": 2.0646,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.186147186147187e-06,
+      "loss": 2.0568,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 2.096,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.1728271728271735e-06,
+      "loss": 2.0116,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.166167166167167e-06,
+      "loss": 1.9549,
+      "step": 2021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.159507159507159e-06,
+      "loss": 2.0594,
+      "step": 2022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.1528471528471535e-06,
+      "loss": 1.6001,
+      "step": 2023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.146187146187147e-06,
+      "loss": 2.1137,
+      "step": 2024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.13952713952714e-06,
+      "loss": 2.1748,
+      "step": 2025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.132867132867134e-06,
+      "loss": 1.8813,
+      "step": 2026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.126207126207126e-06,
+      "loss": 2.2047,
+      "step": 2027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.11954711954712e-06,
+      "loss": 2.0842,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.112887112887114e-06,
+      "loss": 2.1593,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.106227106227107e-06,
+      "loss": 2.1596,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.0995670995671e-06,
+      "loss": 1.9912,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.092907092907093e-06,
+      "loss": 1.9336,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.086247086247087e-06,
+      "loss": 2.0142,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.07958707958708e-06,
+      "loss": 2.0556,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.072927072927074e-06,
+      "loss": 2.0597,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.066267066267067e-06,
+      "loss": 1.9089,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.05960705960706e-06,
+      "loss": 1.8458,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.052947052947053e-06,
+      "loss": 1.9534,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.046287046287047e-06,
+      "loss": 1.9466,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.0396270396270405e-06,
+      "loss": 1.9784,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.032967032967034e-06,
+      "loss": 2.084,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.026307026307026e-06,
+      "loss": 1.885,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.01964701964702e-06,
+      "loss": 2.1857,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.012987012987014e-06,
+      "loss": 1.9195,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.006327006327007e-06,
+      "loss": 2.0401,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.999666999667e-06,
+      "loss": 1.8581,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.993006993006993e-06,
+      "loss": 1.8309,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.9863469863469865e-06,
+      "loss": 2.0749,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.979686979686981e-06,
+      "loss": 1.94,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.973026973026974e-06,
+      "loss": 2.1301,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.966366966366967e-06,
+      "loss": 2.065,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.95970695970696e-06,
+      "loss": 2.2027,
+      "step": 2052
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.953046953046953e-06,
+      "loss": 2.2333,
+      "step": 2053
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.9463869463869475e-06,
+      "loss": 2.0112,
+      "step": 2054
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.939726939726941e-06,
+      "loss": 1.9574,
+      "step": 2055
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.933066933066933e-06,
+      "loss": 1.9878,
+      "step": 2056
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.926406926406927e-06,
+      "loss": 1.9587,
+      "step": 2057
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.91974691974692e-06,
+      "loss": 2.3912,
+      "step": 2058
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 6.913086913086913e-06,
+      "loss": 2.2797,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.906426906426908e-06,
+      "loss": 1.9756,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.8997668997669e-06,
+      "loss": 2.0954,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.8931068931068935e-06,
+      "loss": 2.1925,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.886446886446887e-06,
+      "loss": 2.0042,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.87978687978688e-06,
+      "loss": 1.8559,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.873126873126874e-06,
+      "loss": 2.1691,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.866466866466867e-06,
+      "loss": 2.1827,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.85980685980686e-06,
+      "loss": 1.9781,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.853146853146854e-06,
+      "loss": 2.0568,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.846486846486847e-06,
+      "loss": 1.8702,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.839826839826841e-06,
+      "loss": 2.2786,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.833166833166834e-06,
+      "loss": 2.0314,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.826506826506827e-06,
+      "loss": 1.9545,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.81984681984682e-06,
+      "loss": 1.7995,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.813186813186814e-06,
+      "loss": 1.6801,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.806526806526808e-06,
+      "loss": 2.5632,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7998667998667996e-06,
+      "loss": 2.0709,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.793206793206794e-06,
+      "loss": 1.8679,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.786546786546787e-06,
+      "loss": 1.9178,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7798867798867805e-06,
+      "loss": 2.2005,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.773226773226774e-06,
+      "loss": 1.8751,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.766566766566766e-06,
+      "loss": 2.2392,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.7599067599067605e-06,
+      "loss": 1.7492,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.753246753246754e-06,
+      "loss": 2.1646,
+      "step": 2083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.746586746586747e-06,
+      "loss": 2.1599,
+      "step": 2084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.739926739926741e-06,
+      "loss": 2.0334,
+      "step": 2085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.733266733266733e-06,
+      "loss": 2.0097,
+      "step": 2086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.726606726606727e-06,
+      "loss": 2.2176,
+      "step": 2087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.719946719946721e-06,
+      "loss": 2.2678,
+      "step": 2088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.713286713286714e-06,
+      "loss": 2.1614,
+      "step": 2089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 6.706626706626707e-06,
+      "loss": 1.9663,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.6999666999667e-06,
+      "loss": 2.1363,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.693306693306694e-06,
+      "loss": 2.293,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.6866466866466874e-06,
+      "loss": 1.8329,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.679986679986681e-06,
+      "loss": 1.8585,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.673326673326674e-06,
+      "loss": 2.1456,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 2.3208,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.66000666000666e-06,
+      "loss": 1.7839,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.653346653346654e-06,
+      "loss": 2.1332,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.6466866466866476e-06,
+      "loss": 1.9718,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.640026640026641e-06,
+      "loss": 1.7325,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.633366633366633e-06,
+      "loss": 1.8306,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.626706626706627e-06,
+      "loss": 1.9684,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.620046620046621e-06,
+      "loss": 2.0042,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.613386613386614e-06,
+      "loss": 2.2076,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.606726606726608e-06,
+      "loss": 2.44,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.6000666000666e-06,
+      "loss": 1.7097,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.5934065934065935e-06,
+      "loss": 1.8,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.586746586746588e-06,
+      "loss": 2.2629,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.580086580086581e-06,
+      "loss": 2.1363,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.573426573426574e-06,
+      "loss": 2.441,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.566766566766567e-06,
+      "loss": 2.0833,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.56010656010656e-06,
+      "loss": 2.0845,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.5534465534465545e-06,
+      "loss": 2.0359,
+      "step": 2113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.546786546786548e-06,
+      "loss": 1.9299,
+      "step": 2114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.54012654012654e-06,
+      "loss": 1.8535,
+      "step": 2115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.533466533466534e-06,
+      "loss": 2.3031,
+      "step": 2116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.526806526806527e-06,
+      "loss": 2.1565,
+      "step": 2117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.5201465201465204e-06,
+      "loss": 2.2567,
+      "step": 2118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.513486513486515e-06,
+      "loss": 2.1103,
+      "step": 2119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.506826506826507e-06,
+      "loss": 1.88,
+      "step": 2120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 6.5001665001665005e-06,
+      "loss": 1.7858,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.493506493506494e-06,
+      "loss": 2.1052,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.486846486846487e-06,
+      "loss": 2.206,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.480186480186481e-06,
+      "loss": 1.8526,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.473526473526474e-06,
+      "loss": 2.0491,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.466866466866467e-06,
+      "loss": 1.7028,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.460206460206461e-06,
+      "loss": 1.733,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.453546453546454e-06,
+      "loss": 2.3488,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.446886446886448e-06,
+      "loss": 1.4992,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.440226440226441e-06,
+      "loss": 1.7451,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.433566433566434e-06,
+      "loss": 2.0005,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.426906426906427e-06,
+      "loss": 1.8925,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.420246420246421e-06,
+      "loss": 2.2209,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.413586413586415e-06,
+      "loss": 2.218,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.406926406926407e-06,
+      "loss": 2.1306,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.400266400266401e-06,
+      "loss": 1.9708,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.393606393606394e-06,
+      "loss": 2.1596,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.3869463869463875e-06,
+      "loss": 2.1702,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.380286380286381e-06,
+      "loss": 2.0098,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.373626373626373e-06,
+      "loss": 1.9598,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.3669663669663676e-06,
+      "loss": 2.1186,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.360306360306361e-06,
+      "loss": 2.1761,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.353646353646354e-06,
+      "loss": 2.1453,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.346986346986348e-06,
+      "loss": 1.8039,
+      "step": 2144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.34032634032634e-06,
+      "loss": 2.0849,
+      "step": 2145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.333666333666334e-06,
+      "loss": 2.296,
+      "step": 2146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.327006327006328e-06,
+      "loss": 2.0184,
+      "step": 2147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.320346320346321e-06,
+      "loss": 1.9179,
+      "step": 2148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.313686313686314e-06,
+      "loss": 2.26,
+      "step": 2149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.307026307026307e-06,
+      "loss": 2.0973,
+      "step": 2150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.300366300366301e-06,
+      "loss": 2.1311,
+      "step": 2151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 6.2937062937062944e-06,
+      "loss": 1.9604,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.287046287046288e-06,
+      "loss": 2.1199,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.280386280386281e-06,
+      "loss": 1.7957,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.273726273726274e-06,
+      "loss": 2.1513,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.267066267066267e-06,
+      "loss": 1.9202,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.260406260406261e-06,
+      "loss": 2.1138,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2537462537462546e-06,
+      "loss": 1.8976,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.247086247086248e-06,
+      "loss": 1.8855,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2404262404262404e-06,
+      "loss": 1.5838,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.233766233766234e-06,
+      "loss": 1.9422,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.227106227106228e-06,
+      "loss": 2.1285,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.220446220446221e-06,
+      "loss": 1.8745,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.213786213786215e-06,
+      "loss": 1.7102,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.207126207126207e-06,
+      "loss": 2.0188,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.2004662004662005e-06,
+      "loss": 1.9087,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.193806193806195e-06,
+      "loss": 2.0753,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.187146187146188e-06,
+      "loss": 1.9719,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.1804861804861815e-06,
+      "loss": 1.6798,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.173826173826174e-06,
+      "loss": 1.9022,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.167166167166167e-06,
+      "loss": 1.9739,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.1605061605061615e-06,
+      "loss": 1.651,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.9403,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.147186147186147e-06,
+      "loss": 2.1967,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.140526140526141e-06,
+      "loss": 2.161,
+      "step": 2175
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.133866133866134e-06,
+      "loss": 1.8383,
+      "step": 2176
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.1272061272061274e-06,
+      "loss": 2.407,
+      "step": 2177
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.120546120546122e-06,
+      "loss": 2.128,
+      "step": 2178
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.113886113886114e-06,
+      "loss": 2.1574,
+      "step": 2179
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.1072261072261075e-06,
+      "loss": 1.8036,
+      "step": 2180
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.100566100566101e-06,
+      "loss": 2.0534,
+      "step": 2181
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.093906093906094e-06,
+      "loss": 1.8293,
+      "step": 2182
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 6.087246087246088e-06,
+      "loss": 1.848,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.080586080586081e-06,
+      "loss": 2.0488,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.073926073926074e-06,
+      "loss": 1.789,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.067266067266068e-06,
+      "loss": 2.261,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.7746,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.053946053946055e-06,
+      "loss": 2.4117,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.047286047286048e-06,
+      "loss": 1.683,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.040626040626041e-06,
+      "loss": 1.7051,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.033966033966034e-06,
+      "loss": 1.9012,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.027306027306028e-06,
+      "loss": 1.9108,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.020646020646022e-06,
+      "loss": 2.1442,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.013986013986014e-06,
+      "loss": 2.0447,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.007326007326008e-06,
+      "loss": 2.0611,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 6.000666000666001e-06,
+      "loss": 2.3313,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9940059940059945e-06,
+      "loss": 2.0571,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.987345987345988e-06,
+      "loss": 1.8655,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.98068598068598e-06,
+      "loss": 1.9532,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9740259740259746e-06,
+      "loss": 1.896,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.967365967365968e-06,
+      "loss": 1.874,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.960705960705961e-06,
+      "loss": 1.9077,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.954045954045955e-06,
+      "loss": 1.9595,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.947385947385947e-06,
+      "loss": 2.1298,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.940725940725941e-06,
+      "loss": 1.7677,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.934065934065935e-06,
+      "loss": 2.0829,
+      "step": 2206
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.927405927405928e-06,
+      "loss": 2.279,
+      "step": 2207
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.920745920745921e-06,
+      "loss": 2.033,
+      "step": 2208
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914085914085914e-06,
+      "loss": 2.2869,
+      "step": 2209
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.907425907425908e-06,
+      "loss": 1.9999,
+      "step": 2210
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9007659007659015e-06,
+      "loss": 1.9304,
+      "step": 2211
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.894105894105895e-06,
+      "loss": 2.4465,
+      "step": 2212
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.887445887445888e-06,
+      "loss": 1.9757,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.880785880785881e-06,
+      "loss": 2.2639,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.874125874125874e-06,
+      "loss": 1.9605,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.867465867465868e-06,
+      "loss": 2.1627,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.860805860805862e-06,
+      "loss": 1.8691,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.854145854145855e-06,
+      "loss": 1.6944,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8474858474858474e-06,
+      "loss": 1.8352,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.840825840825841e-06,
+      "loss": 2.0595,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.834165834165835e-06,
+      "loss": 2.317,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.827505827505828e-06,
+      "loss": 1.8542,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.820845820845822e-06,
+      "loss": 2.3744,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.814185814185814e-06,
+      "loss": 2.0869,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.8075258075258076e-06,
+      "loss": 2.1283,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.800865800865802e-06,
+      "loss": 2.0397,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.794205794205795e-06,
+      "loss": 2.1671,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7875457875457885e-06,
+      "loss": 1.9534,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.780885780885781e-06,
+      "loss": 1.6984,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.774225774225774e-06,
+      "loss": 2.1189,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7675657675657685e-06,
+      "loss": 1.569,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.760905760905762e-06,
+      "loss": 2.0562,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.754245754245755e-06,
+      "loss": 1.9993,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.747585747585748e-06,
+      "loss": 2.2558,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.740925740925741e-06,
+      "loss": 2.5334,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7342657342657345e-06,
+      "loss": 1.951,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.727605727605729e-06,
+      "loss": 2.278,
+      "step": 2237
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.720945720945721e-06,
+      "loss": 2.0663,
+      "step": 2238
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.898,
+      "step": 2239
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.707625707625708e-06,
+      "loss": 2.206,
+      "step": 2240
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.700965700965701e-06,
+      "loss": 2.2643,
+      "step": 2241
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.694305694305695e-06,
+      "loss": 2.0389,
+      "step": 2242
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.687645687645688e-06,
+      "loss": 2.0684,
+      "step": 2243
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.680985680985681e-06,
+      "loss": 2.2746,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.674325674325675e-06,
+      "loss": 2.091,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.667665667665668e-06,
+      "loss": 1.995,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.661005661005662e-06,
+      "loss": 1.8354,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.654345654345655e-06,
+      "loss": 1.8657,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.647685647685648e-06,
+      "loss": 2.1556,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.9797,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.634365634365635e-06,
+      "loss": 2.0407,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.627705627705629e-06,
+      "loss": 2.0123,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.621045621045621e-06,
+      "loss": 1.9787,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.614385614385615e-06,
+      "loss": 1.9224,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.607725607725608e-06,
+      "loss": 2.1307,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.6010656010656015e-06,
+      "loss": 2.0825,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.594405594405595e-06,
+      "loss": 2.0003,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.587745587745587e-06,
+      "loss": 1.8542,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.581085581085582e-06,
+      "loss": 1.8863,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.574425574425575e-06,
+      "loss": 2.0508,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.567765567765568e-06,
+      "loss": 1.7526,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.561105561105562e-06,
+      "loss": 1.9392,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.554445554445554e-06,
+      "loss": 1.9412,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.547785547785548e-06,
+      "loss": 1.9923,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.541125541125542e-06,
+      "loss": 2.1186,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.534465534465535e-06,
+      "loss": 2.1561,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.527805527805528e-06,
+      "loss": 1.6991,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.521145521145521e-06,
+      "loss": 2.0223,
+      "step": 2268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.514485514485515e-06,
+      "loss": 2.2821,
+      "step": 2269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.5078255078255085e-06,
+      "loss": 1.638,
+      "step": 2270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.501165501165502e-06,
+      "loss": 2.4105,
+      "step": 2271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.494505494505495e-06,
+      "loss": 2.1378,
+      "step": 2272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.487845487845488e-06,
+      "loss": 1.7653,
+      "step": 2273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.481185481185481e-06,
+      "loss": 1.9301,
+      "step": 2274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.474525474525475e-06,
+      "loss": 2.0913,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.467865467865469e-06,
+      "loss": 1.8653,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.461205461205462e-06,
+      "loss": 1.8543,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.8076,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.447885447885448e-06,
+      "loss": 1.9196,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.441225441225442e-06,
+      "loss": 2.137,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.434565434565435e-06,
+      "loss": 1.9041,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.427905427905429e-06,
+      "loss": 1.9378,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.421245421245421e-06,
+      "loss": 2.0634,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.4145854145854146e-06,
+      "loss": 2.0872,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.407925407925409e-06,
+      "loss": 2.0858,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.401265401265402e-06,
+      "loss": 1.7555,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3946053946053955e-06,
+      "loss": 2.0148,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.387945387945388e-06,
+      "loss": 2.1661,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.381285381285381e-06,
+      "loss": 2.1426,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3746253746253755e-06,
+      "loss": 1.9885,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.367965367965369e-06,
+      "loss": 1.9381,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.361305361305362e-06,
+      "loss": 1.7976,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.354645354645355e-06,
+      "loss": 2.0816,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.347985347985348e-06,
+      "loss": 2.3565,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3413253413253415e-06,
+      "loss": 2.1333,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.334665334665336e-06,
+      "loss": 2.0349,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.328005328005329e-06,
+      "loss": 2.0465,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3213453213453215e-06,
+      "loss": 1.7555,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.314685314685315e-06,
+      "loss": 2.1167,
+      "step": 2299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.308025308025308e-06,
+      "loss": 2.0523,
+      "step": 2300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.3013653013653024e-06,
+      "loss": 1.9275,
+      "step": 2301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.294705294705295e-06,
+      "loss": 1.85,
+      "step": 2302
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.288045288045288e-06,
+      "loss": 2.0824,
+      "step": 2303
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.281385281385282e-06,
+      "loss": 2.4094,
+      "step": 2304
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.274725274725275e-06,
+      "loss": 1.7739,
+      "step": 2305
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.268065268065269e-06,
+      "loss": 1.8625,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.261405261405262e-06,
+      "loss": 2.2302,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.254745254745255e-06,
+      "loss": 2.3216,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.248085248085248e-06,
+      "loss": 2.0063,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.241425241425242e-06,
+      "loss": 2.0326,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.234765234765236e-06,
+      "loss": 1.8826,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.228105228105228e-06,
+      "loss": 1.8761,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.221445221445222e-06,
+      "loss": 2.1178,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.214785214785215e-06,
+      "loss": 1.9868,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.2081252081252085e-06,
+      "loss": 2.1863,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.201465201465202e-06,
+      "loss": 2.0349,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.194805194805194e-06,
+      "loss": 2.4442,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.188145188145189e-06,
+      "loss": 1.9202,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.181485181485182e-06,
+      "loss": 2.0252,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.174825174825175e-06,
+      "loss": 2.0409,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.168165168165169e-06,
+      "loss": 1.6357,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.161505161505161e-06,
+      "loss": 1.8885,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.154845154845155e-06,
+      "loss": 2.2196,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.148185148185149e-06,
+      "loss": 1.8671,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.141525141525142e-06,
+      "loss": 2.0019,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.134865134865135e-06,
+      "loss": 2.3511,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.7949,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.121545121545122e-06,
+      "loss": 1.8298,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.1148851148851155e-06,
+      "loss": 1.7743,
+      "step": 2329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.108225108225109e-06,
+      "loss": 2.2628,
+      "step": 2330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.101565101565102e-06,
+      "loss": 1.9894,
+      "step": 2331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.094905094905095e-06,
+      "loss": 1.8921,
+      "step": 2332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.088245088245088e-06,
+      "loss": 1.985,
+      "step": 2333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.081585081585082e-06,
+      "loss": 1.9645,
+      "step": 2334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.074925074925076e-06,
+      "loss": 2.0354,
+      "step": 2335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.068265068265069e-06,
+      "loss": 2.0854,
+      "step": 2336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.0616050616050615e-06,
+      "loss": 1.9656,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.054945054945055e-06,
+      "loss": 2.0819,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.048285048285049e-06,
+      "loss": 2.0647,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.041625041625042e-06,
+      "loss": 1.9489,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.034965034965036e-06,
+      "loss": 1.837,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.028305028305028e-06,
+      "loss": 1.8193,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.021645021645022e-06,
+      "loss": 2.0884,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.014985014985016e-06,
+      "loss": 1.8961,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.008325008325009e-06,
+      "loss": 1.6885,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.0016650016650025e-06,
+      "loss": 2.1458,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.995004995004996e-06,
+      "loss": 2.2248,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.988344988344988e-06,
+      "loss": 1.8958,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9816849816849826e-06,
+      "loss": 1.8703,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.975024975024975e-06,
+      "loss": 2.0145,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.968364968364968e-06,
+      "loss": 1.9357,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.961704961704963e-06,
+      "loss": 2.0348,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.955044955044955e-06,
+      "loss": 2.1498,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.9483849483849485e-06,
+      "loss": 1.9887,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.941724941724942e-06,
+      "loss": 2.1061,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.935064935064935e-06,
+      "loss": 2.4069,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.928404928404929e-06,
+      "loss": 2.1858,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.921744921744922e-06,
+      "loss": 2.1184,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.915084915084915e-06,
+      "loss": 2.562,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.908424908424909e-06,
+      "loss": 1.6186,
+      "step": 2360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.901764901764902e-06,
+      "loss": 1.9495,
+      "step": 2361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.895104895104895e-06,
+      "loss": 2.3476,
+      "step": 2362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.888444888444889e-06,
+      "loss": 1.9215,
+      "step": 2363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.881784881784882e-06,
+      "loss": 1.9019,
+      "step": 2364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.875124875124875e-06,
+      "loss": 2.0684,
+      "step": 2365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.868464868464869e-06,
+      "loss": 2.2997,
+      "step": 2366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.861804861804862e-06,
+      "loss": 2.3335,
+      "step": 2367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.8551448551448554e-06,
+      "loss": 2.2816,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 2.2674,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.841824841824842e-06,
+      "loss": 1.918,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8351648351648355e-06,
+      "loss": 1.9943,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.828504828504829e-06,
+      "loss": 1.7397,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.821844821844822e-06,
+      "loss": 2.0342,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.8151848151848155e-06,
+      "loss": 1.8352,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.808524808524809e-06,
+      "loss": 1.9368,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.801864801864802e-06,
+      "loss": 2.1893,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.795204795204796e-06,
+      "loss": 1.556,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.788544788544789e-06,
+      "loss": 1.9594,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.781884781884782e-06,
+      "loss": 2.4279,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.775224775224776e-06,
+      "loss": 1.8802,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.768564768564769e-06,
+      "loss": 2.2348,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 1.9829,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.755244755244756e-06,
+      "loss": 1.9067,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.748584748584749e-06,
+      "loss": 2.1772,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7419247419247424e-06,
+      "loss": 1.9385,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.735264735264736e-06,
+      "loss": 1.8643,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.728604728604729e-06,
+      "loss": 2.1412,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7219447219447225e-06,
+      "loss": 2.0924,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.715284715284716e-06,
+      "loss": 1.8704,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.708624708624709e-06,
+      "loss": 2.4107,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.7019647019647026e-06,
+      "loss": 1.9162,
+      "step": 2391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.695304695304695e-06,
+      "loss": 2.053,
+      "step": 2392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.688644688644689e-06,
+      "loss": 1.6727,
+      "step": 2393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.681984681984683e-06,
+      "loss": 2.0194,
+      "step": 2394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.675324675324676e-06,
+      "loss": 2.1364,
+      "step": 2395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.668664668664669e-06,
+      "loss": 1.9427,
+      "step": 2396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.662004662004662e-06,
+      "loss": 2.0949,
+      "step": 2397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.655344655344656e-06,
+      "loss": 2.0292,
+      "step": 2398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.648684648684649e-06,
+      "loss": 2.105,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.642024642024642e-06,
+      "loss": 1.8231,
+      "step": 2400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.635364635364636e-06,
+      "loss": 2.0108,
+      "step": 2401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.628704628704629e-06,
+      "loss": 2.0015,
+      "step": 2402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.622044622044623e-06,
+      "loss": 2.1838,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.9184,
+      "step": 2404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.608724608724609e-06,
+      "loss": 1.8252,
+      "step": 2405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.602064602064603e-06,
+      "loss": 1.8455,
+      "step": 2406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.595404595404595e-06,
+      "loss": 1.8186,
+      "step": 2407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5887445887445896e-06,
+      "loss": 1.9899,
+      "step": 2408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.582084582084583e-06,
+      "loss": 1.7368,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5754245754245754e-06,
+      "loss": 2.093,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.56876456876457e-06,
+      "loss": 1.602,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.562104562104562e-06,
+      "loss": 1.7121,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.5554445554445555e-06,
+      "loss": 2.3097,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.548784548784549e-06,
+      "loss": 1.8326,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.542124542124542e-06,
+      "loss": 2.0767,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.535464535464536e-06,
+      "loss": 1.8893,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.528804528804529e-06,
+      "loss": 2.0782,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.522144522144522e-06,
+      "loss": 2.375,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.515484515484516e-06,
+      "loss": 1.9786,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.508824508824509e-06,
+      "loss": 1.9985,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.502164502164502e-06,
+      "loss": 1.8467,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.495504495504496e-06,
+      "loss": 1.8237,
+      "step": 2422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.488844488844489e-06,
+      "loss": 1.8864,
+      "step": 2423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.482184482184482e-06,
+      "loss": 1.7577,
+      "step": 2424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.475524475524476e-06,
+      "loss": 2.0857,
+      "step": 2425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.468864468864469e-06,
+      "loss": 2.0299,
+      "step": 2426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4622044622044624e-06,
+      "loss": 1.5737,
+      "step": 2427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.455544455544456e-06,
+      "loss": 2.3694,
+      "step": 2428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.448884448884449e-06,
+      "loss": 2.0983,
+      "step": 2429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.4422244422244425e-06,
+      "loss": 1.9221,
+      "step": 2430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.435564435564436e-06,
+      "loss": 1.9558,
+      "step": 2431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.428904428904429e-06,
+      "loss": 2.3498,
+      "step": 2432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.4222444222444226e-06,
+      "loss": 1.6716,
+      "step": 2433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.415584415584416e-06,
+      "loss": 1.8944,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.408924408924409e-06,
+      "loss": 1.7774,
+      "step": 2435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.402264402264403e-06,
+      "loss": 1.9062,
+      "step": 2436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.395604395604396e-06,
+      "loss": 2.0755,
+      "step": 2437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.388944388944389e-06,
+      "loss": 1.6631,
+      "step": 2438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.382284382284383e-06,
+      "loss": 1.8427,
+      "step": 2439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.375624375624376e-06,
+      "loss": 1.9336,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.368964368964369e-06,
+      "loss": 1.9683,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.362304362304363e-06,
+      "loss": 2.1741,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.355644355644356e-06,
+      "loss": 2.328,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3489843489843494e-06,
+      "loss": 1.9623,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.342324342324343e-06,
+      "loss": 1.7309,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.335664335664336e-06,
+      "loss": 2.2485,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3290043290043295e-06,
+      "loss": 1.8919,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.322344322344323e-06,
+      "loss": 1.8334,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.315684315684316e-06,
+      "loss": 1.7148,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.3090243090243096e-06,
+      "loss": 2.0322,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.302364302364302e-06,
+      "loss": 1.9411,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.295704295704296e-06,
+      "loss": 1.7611,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.28904428904429e-06,
+      "loss": 1.8153,
+      "step": 2453
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.282384282384283e-06,
+      "loss": 1.8867,
+      "step": 2454
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.275724275724276e-06,
+      "loss": 1.9901,
+      "step": 2455
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.269064269064269e-06,
+      "loss": 2.2522,
+      "step": 2456
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.262404262404263e-06,
+      "loss": 1.7209,
+      "step": 2457
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.255744255744256e-06,
+      "loss": 2.0681,
+      "step": 2458
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.249084249084249e-06,
+      "loss": 1.8103,
+      "step": 2459
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 1.8775,
+      "step": 2460
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.235764235764236e-06,
+      "loss": 2.3159,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.22910422910423e-06,
+      "loss": 1.972,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.222444222444223e-06,
+      "loss": 1.9358,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.215784215784216e-06,
+      "loss": 2.0829,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.20912420912421e-06,
+      "loss": 1.9253,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.202464202464202e-06,
+      "loss": 2.2245,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.195804195804197e-06,
+      "loss": 2.0114,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.18914418914419e-06,
+      "loss": 1.8095,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1824841824841824e-06,
+      "loss": 1.9379,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.175824175824177e-06,
+      "loss": 2.2052,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.169164169164169e-06,
+      "loss": 2.2886,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.1625041625041625e-06,
+      "loss": 1.7751,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.155844155844157e-06,
+      "loss": 2.111,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.149184149184149e-06,
+      "loss": 2.0025,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.142524142524143e-06,
+      "loss": 1.7097,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.135864135864136e-06,
+      "loss": 2.1519,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.129204129204129e-06,
+      "loss": 1.7076,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.122544122544123e-06,
+      "loss": 1.728,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.115884115884116e-06,
+      "loss": 2.2057,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.109224109224109e-06,
+      "loss": 1.9799,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 2.0463,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.095904095904096e-06,
+      "loss": 2.052,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.089244089244089e-06,
+      "loss": 1.9911,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.082584082584083e-06,
+      "loss": 1.8057,
+      "step": 2484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.075924075924076e-06,
+      "loss": 2.1282,
+      "step": 2485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0692640692640695e-06,
+      "loss": 2.1138,
+      "step": 2486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.062604062604063e-06,
+      "loss": 1.8738,
+      "step": 2487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.055944055944056e-06,
+      "loss": 2.3204,
+      "step": 2488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0492840492840495e-06,
+      "loss": 1.8132,
+      "step": 2489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.042624042624043e-06,
+      "loss": 1.8549,
+      "step": 2490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.035964035964036e-06,
+      "loss": 2.0156,
+      "step": 2491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.0293040293040296e-06,
+      "loss": 2.1087,
+      "step": 2492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.022644022644023e-06,
+      "loss": 1.8989,
+      "step": 2493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.015984015984016e-06,
+      "loss": 2.3012,
+      "step": 2494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.00932400932401e-06,
+      "loss": 2.1079,
+      "step": 2495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.002664002664003e-06,
+      "loss": 2.2914,
+      "step": 2496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.996003996003996e-06,
+      "loss": 2.003,
+      "step": 2497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.98934398934399e-06,
+      "loss": 2.06,
+      "step": 2498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.982683982683983e-06,
+      "loss": 2.3096,
+      "step": 2499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.976023976023976e-06,
+      "loss": 1.8004,
+      "step": 2500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.96936396936397e-06,
+      "loss": 2.059,
+      "step": 2501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.962703962703963e-06,
+      "loss": 2.0652,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9560439560439565e-06,
+      "loss": 2.0426,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.94938394938395e-06,
+      "loss": 1.8907,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.942723942723943e-06,
+      "loss": 1.9472,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.9360639360639365e-06,
+      "loss": 1.7858,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.92940392940393e-06,
+      "loss": 2.0687,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.922743922743923e-06,
+      "loss": 1.997,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.916083916083917e-06,
+      "loss": 2.0426,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.909423909423909e-06,
+      "loss": 1.9327,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.902763902763903e-06,
+      "loss": 1.9627,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.896103896103897e-06,
+      "loss": 1.742,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.88944388944389e-06,
+      "loss": 1.7129,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.882783882783883e-06,
+      "loss": 1.5815,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.876123876123876e-06,
+      "loss": 1.923,
+      "step": 2515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.86946386946387e-06,
+      "loss": 2.1315,
+      "step": 2516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.862803862803863e-06,
+      "loss": 2.1605,
+      "step": 2517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.856143856143856e-06,
+      "loss": 1.826,
+      "step": 2518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.84948384948385e-06,
+      "loss": 2.2541,
+      "step": 2519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.842823842823843e-06,
+      "loss": 1.9891,
+      "step": 2520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.836163836163837e-06,
+      "loss": 1.9125,
+      "step": 2521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.82950382950383e-06,
+      "loss": 2.1137,
+      "step": 2522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.822843822843823e-06,
+      "loss": 2.0506,
+      "step": 2523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.816183816183817e-06,
+      "loss": 2.065,
+      "step": 2524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 2.1046,
+      "step": 2525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.802863802863803e-06,
+      "loss": 2.2896,
+      "step": 2526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.796203796203797e-06,
+      "loss": 2.1674,
+      "step": 2527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.78954378954379e-06,
+      "loss": 1.7385,
+      "step": 2528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7828837828837832e-06,
+      "loss": 2.2641,
+      "step": 2529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.776223776223776e-06,
+      "loss": 2.1348,
+      "step": 2530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.76956376956377e-06,
+      "loss": 2.3509,
+      "step": 2531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7629037629037633e-06,
+      "loss": 1.8688,
+      "step": 2532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7562437562437566e-06,
+      "loss": 2.0158,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.74958374958375e-06,
+      "loss": 1.8767,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.742923742923743e-06,
+      "loss": 2.1849,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7362637362637367e-06,
+      "loss": 1.698,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.72960372960373e-06,
+      "loss": 2.2736,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.722943722943723e-06,
+      "loss": 1.8388,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7162837162837168e-06,
+      "loss": 1.9894,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7096237096237097e-06,
+      "loss": 1.8732,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7029637029637035e-06,
+      "loss": 2.2632,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6963036963036964e-06,
+      "loss": 1.8411,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6896436896436898e-06,
+      "loss": 2.2588,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6829836829836835e-06,
+      "loss": 1.791,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6763236763236765e-06,
+      "loss": 1.8097,
+      "step": 2545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6696636696636702e-06,
+      "loss": 2.2426,
+      "step": 2546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.663003663003663e-06,
+      "loss": 1.957,
+      "step": 2547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6563436563436565e-06,
+      "loss": 1.9077,
+      "step": 2548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6496836496836503e-06,
+      "loss": 2.0364,
+      "step": 2549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6430236430236432e-06,
+      "loss": 1.7824,
+      "step": 2550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.7068,
+      "step": 2551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.62970362970363e-06,
+      "loss": 2.1315,
+      "step": 2552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6230436230436233e-06,
+      "loss": 1.7739,
+      "step": 2553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.616383616383617e-06,
+      "loss": 2.0664,
+      "step": 2554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.60972360972361e-06,
+      "loss": 1.7463,
+      "step": 2555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6030636030636034e-06,
+      "loss": 1.6195,
+      "step": 2556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5964035964035963e-06,
+      "loss": 2.0109,
+      "step": 2557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 2.2592,
+      "step": 2558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5830835830835834e-06,
+      "loss": 2.1352,
+      "step": 2559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5764235764235768e-06,
+      "loss": 1.7939,
+      "step": 2560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.56976356976357e-06,
+      "loss": 1.8208,
+      "step": 2561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.563103563103563e-06,
+      "loss": 2.4233,
+      "step": 2562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.556443556443557e-06,
+      "loss": 1.8261,
+      "step": 2563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.54978354978355e-06,
+      "loss": 1.8469,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5431235431235435e-06,
+      "loss": 2.0721,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.536463536463537e-06,
+      "loss": 1.8521,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.52980352980353e-06,
+      "loss": 1.7292,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5231435231435236e-06,
+      "loss": 2.2353,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.516483516483517e-06,
+      "loss": 2.1036,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.50982350982351e-06,
+      "loss": 1.8954,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5031635031635037e-06,
+      "loss": 2.0393,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4965034965034966e-06,
+      "loss": 1.8863,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4898434898434904e-06,
+      "loss": 2.0592,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4831834831834833e-06,
+      "loss": 1.9426,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4765234765234766e-06,
+      "loss": 2.2739,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4698634698634704e-06,
+      "loss": 1.9724,
+      "step": 2576
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4632034632034634e-06,
+      "loss": 2.0577,
+      "step": 2577
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4565434565434567e-06,
+      "loss": 1.7979,
+      "step": 2578
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.44988344988345e-06,
+      "loss": 1.6873,
+      "step": 2579
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4432234432234434e-06,
+      "loss": 2.309,
+      "step": 2580
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.436563436563437e-06,
+      "loss": 1.9245,
+      "step": 2581
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.42990342990343e-06,
+      "loss": 2.1611,
+      "step": 2582
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4232434232434235e-06,
+      "loss": 2.0066,
+      "step": 2583
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.416583416583417e-06,
+      "loss": 2.1198,
+      "step": 2584
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.40992340992341e-06,
+      "loss": 2.1826,
+      "step": 2585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.403263403263404e-06,
+      "loss": 2.2786,
+      "step": 2586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.396603396603397e-06,
+      "loss": 2.1113,
+      "step": 2587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3899433899433902e-06,
+      "loss": 1.9467,
+      "step": 2588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.383283383283383e-06,
+      "loss": 1.6909,
+      "step": 2589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.376623376623377e-06,
+      "loss": 2.0089,
+      "step": 2590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3699633699633703e-06,
+      "loss": 1.9891,
+      "step": 2591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3633033633033637e-06,
+      "loss": 1.9586,
+      "step": 2592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.356643356643357e-06,
+      "loss": 2.0667,
+      "step": 2593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.34998334998335e-06,
+      "loss": 1.9525,
+      "step": 2594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3433233433233437e-06,
+      "loss": 2.3157,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.336663336663337e-06,
+      "loss": 1.6493,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.33000333000333e-06,
+      "loss": 2.0162,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3233433233433238e-06,
+      "loss": 2.1108,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3166833166833167e-06,
+      "loss": 2.0537,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.3100233100233105e-06,
+      "loss": 2.3352,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.303363303363304e-06,
+      "loss": 1.9304,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2967032967032968e-06,
+      "loss": 2.1931,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2900432900432905e-06,
+      "loss": 2.0329,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2833832833832835e-06,
+      "loss": 2.1341,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2767232767232773e-06,
+      "loss": 2.1186,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.27006327006327e-06,
+      "loss": 1.8713,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2634032634032635e-06,
+      "loss": 2.0786,
+      "step": 2607
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2567432567432573e-06,
+      "loss": 1.8797,
+      "step": 2608
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2500832500832502e-06,
+      "loss": 1.6101,
+      "step": 2609
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2434232434232436e-06,
+      "loss": 1.7378,
+      "step": 2610
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.236763236763237e-06,
+      "loss": 1.9784,
+      "step": 2611
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2301032301032303e-06,
+      "loss": 2.2617,
+      "step": 2612
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.223443223443224e-06,
+      "loss": 1.8483,
+      "step": 2613
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.216783216783217e-06,
+      "loss": 1.726,
+      "step": 2614
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2101232101232104e-06,
+      "loss": 1.6993,
+      "step": 2615
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2034632034632033e-06,
+      "loss": 1.8843,
+      "step": 2616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.196803196803197e-06,
+      "loss": 1.866,
+      "step": 2617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1901431901431904e-06,
+      "loss": 2.2558,
+      "step": 2618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1834831834831838e-06,
+      "loss": 2.1352,
+      "step": 2619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.176823176823177e-06,
+      "loss": 1.8598,
+      "step": 2620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.17016317016317e-06,
+      "loss": 1.968,
+      "step": 2621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.163503163503164e-06,
+      "loss": 1.7732,
+      "step": 2622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.156843156843157e-06,
+      "loss": 2.143,
+      "step": 2623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1501831501831505e-06,
+      "loss": 1.7932,
+      "step": 2624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.143523143523144e-06,
+      "loss": 2.0069,
+      "step": 2625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.136863136863137e-06,
+      "loss": 1.9317,
+      "step": 2626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1302031302031306e-06,
+      "loss": 1.7652,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.123543123543124e-06,
+      "loss": 1.5746,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.116883116883117e-06,
+      "loss": 1.5153,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1102231102231107e-06,
+      "loss": 2.1893,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.1035631035631036e-06,
+      "loss": 2.2149,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0969030969030974e-06,
+      "loss": 1.8631,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0902430902430907e-06,
+      "loss": 2.1265,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0835830835830837e-06,
+      "loss": 2.0795,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 2.2795,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0702630702630704e-06,
+      "loss": 1.7978,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0636030636030637e-06,
+      "loss": 2.0706,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.056943056943057e-06,
+      "loss": 1.8167,
+      "step": 2638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0502830502830504e-06,
+      "loss": 1.6607,
+      "step": 2639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.043623043623044e-06,
+      "loss": 1.8576,
+      "step": 2640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.036963036963037e-06,
+      "loss": 2.118,
+      "step": 2641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 1.8902,
+      "step": 2642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.023643023643024e-06,
+      "loss": 1.7275,
+      "step": 2643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.016983016983017e-06,
+      "loss": 2.0944,
+      "step": 2644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.010323010323011e-06,
+      "loss": 2.2222,
+      "step": 2645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.003663003663004e-06,
+      "loss": 2.0911,
+      "step": 2646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.9970029970029973e-06,
+      "loss": 2.3857,
+      "step": 2647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.99034299034299e-06,
+      "loss": 1.9891,
+      "step": 2648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.983682983682984e-06,
+      "loss": 1.9829,
+      "step": 2649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9770229770229773e-06,
+      "loss": 2.0259,
+      "step": 2650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9703629703629707e-06,
+      "loss": 1.759,
+      "step": 2651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.963702963702964e-06,
+      "loss": 1.9512,
+      "step": 2652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.957042957042957e-06,
+      "loss": 2.0152,
+      "step": 2653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9503829503829507e-06,
+      "loss": 1.9364,
+      "step": 2654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.943722943722944e-06,
+      "loss": 1.8597,
+      "step": 2655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.937062937062937e-06,
+      "loss": 1.7587,
+      "step": 2656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.930402930402931e-06,
+      "loss": 2.0831,
+      "step": 2657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9237429237429237e-06,
+      "loss": 2.319,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9170829170829175e-06,
+      "loss": 2.4253,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.910422910422911e-06,
+      "loss": 2.2275,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.9037629037629038e-06,
+      "loss": 1.6849,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8971028971028976e-06,
+      "loss": 2.0697,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8904428904428905e-06,
+      "loss": 1.922,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8837828837828843e-06,
+      "loss": 1.8066,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8771228771228776e-06,
+      "loss": 1.7257,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8704628704628705e-06,
+      "loss": 2.1196,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8638028638028643e-06,
+      "loss": 2.1048,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.939,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8504828504828506e-06,
+      "loss": 2.2973,
+      "step": 2669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.843822843822844e-06,
+      "loss": 1.9336,
+      "step": 2670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8371628371628373e-06,
+      "loss": 1.6022,
+      "step": 2671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.830502830502831e-06,
+      "loss": 1.9851,
+      "step": 2672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.823842823842824e-06,
+      "loss": 2.2141,
+      "step": 2673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8171828171828174e-06,
+      "loss": 2.1887,
+      "step": 2674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8105228105228103e-06,
+      "loss": 2.1972,
+      "step": 2675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.803862803862804e-06,
+      "loss": 1.9988,
+      "step": 2676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.7972027972027974e-06,
+      "loss": 1.8736,
+      "step": 2677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.790542790542791e-06,
+      "loss": 2.0561,
+      "step": 2678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.783882783882784e-06,
+      "loss": 1.8979,
+      "step": 2679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.777222777222777e-06,
+      "loss": 1.9893,
+      "step": 2680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.770562770562771e-06,
+      "loss": 2.0568,
+      "step": 2681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.763902763902764e-06,
+      "loss": 1.8259,
+      "step": 2682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7572427572427576e-06,
+      "loss": 1.8468,
+      "step": 2683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.750582750582751e-06,
+      "loss": 1.8255,
+      "step": 2684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.743922743922744e-06,
+      "loss": 1.7395,
+      "step": 2685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7372627372627376e-06,
+      "loss": 1.7919,
+      "step": 2686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.730602730602731e-06,
+      "loss": 1.9528,
+      "step": 2687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.723942723942724e-06,
+      "loss": 1.8757,
+      "step": 2688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7172827172827177e-06,
+      "loss": 1.8934,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7106227106227106e-06,
+      "loss": 2.2122,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.7039627039627044e-06,
+      "loss": 1.9897,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6973026973026977e-06,
+      "loss": 2.0166,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6906426906426907e-06,
+      "loss": 1.8786,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6839826839826844e-06,
+      "loss": 1.9952,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6773226773226774e-06,
+      "loss": 2.1598,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6706626706626707e-06,
+      "loss": 2.5528,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6640026640026645e-06,
+      "loss": 1.8533,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6573426573426574e-06,
+      "loss": 1.9001,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6506826506826512e-06,
+      "loss": 2.2528,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.644022644022644e-06,
+      "loss": 2.1046,
+      "step": 2700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6373626373626375e-06,
+      "loss": 1.7071,
+      "step": 2701
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.630702630702631e-06,
+      "loss": 2.1185,
+      "step": 2702
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.624042624042624e-06,
+      "loss": 2.1909,
+      "step": 2703
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.617382617382618e-06,
+      "loss": 1.9968,
+      "step": 2704
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.610722610722611e-06,
+      "loss": 2.1661,
+      "step": 2705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.6040626040626043e-06,
+      "loss": 1.9262,
+      "step": 2706
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.597402597402597e-06,
+      "loss": 2.0115,
+      "step": 2707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.590742590742591e-06,
+      "loss": 1.9445,
+      "step": 2708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.5840825840825843e-06,
+      "loss": 2.2417,
+      "step": 2709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5774225774225777e-06,
+      "loss": 1.8247,
+      "step": 2710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.570762570762571e-06,
+      "loss": 1.9021,
+      "step": 2711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.564102564102564e-06,
+      "loss": 2.1089,
+      "step": 2712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5574425574425577e-06,
+      "loss": 1.6914,
+      "step": 2713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.550782550782551e-06,
+      "loss": 2.2085,
+      "step": 2714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.544122544122544e-06,
+      "loss": 2.0876,
+      "step": 2715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.537462537462538e-06,
+      "loss": 1.9454,
+      "step": 2716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5308025308025307e-06,
+      "loss": 1.7823,
+      "step": 2717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5241425241425245e-06,
+      "loss": 1.8929,
+      "step": 2718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.517482517482518e-06,
+      "loss": 1.6444,
+      "step": 2719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.510822510822511e-06,
+      "loss": 1.6621,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.5041625041625046e-06,
+      "loss": 2.3329,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.497502497502498e-06,
+      "loss": 1.7677,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4908424908424913e-06,
+      "loss": 2.5295,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.484182484182484e-06,
+      "loss": 2.3237,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4775224775224776e-06,
+      "loss": 1.9852,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.470862470862471e-06,
+      "loss": 1.7146,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4642024642024647e-06,
+      "loss": 2.3486,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4575424575424576e-06,
+      "loss": 1.9361,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.450882450882451e-06,
+      "loss": 1.8448,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4442224442224443e-06,
+      "loss": 2.2813,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4375624375624377e-06,
+      "loss": 1.7235,
+      "step": 2731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.430902430902431e-06,
+      "loss": 1.9363,
+      "step": 2732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 2.152,
+      "step": 2733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4175824175824177e-06,
+      "loss": 2.0231,
+      "step": 2734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.410922410922411e-06,
+      "loss": 1.9941,
+      "step": 2735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.4042624042624045e-06,
+      "loss": 2.2737,
+      "step": 2736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.397602397602398e-06,
+      "loss": 1.9058,
+      "step": 2737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.390942390942391e-06,
+      "loss": 2.3321,
+      "step": 2738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.3842823842823845e-06,
+      "loss": 1.9409,
+      "step": 2739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.377622377622378e-06,
+      "loss": 1.7822,
+      "step": 2740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3709623709623712e-06,
+      "loss": 1.8209,
+      "step": 2741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3643023643023646e-06,
+      "loss": 2.1039,
+      "step": 2742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.357642357642358e-06,
+      "loss": 1.885,
+      "step": 2743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3509823509823513e-06,
+      "loss": 1.9629,
+      "step": 2744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3443223443223446e-06,
+      "loss": 2.1546,
+      "step": 2745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.337662337662338e-06,
+      "loss": 1.9472,
+      "step": 2746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.331002331002331e-06,
+      "loss": 2.0688,
+      "step": 2747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3243423243423247e-06,
+      "loss": 2.3951,
+      "step": 2748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.317682317682318e-06,
+      "loss": 1.978,
+      "step": 2749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3110223110223114e-06,
+      "loss": 1.9699,
+      "step": 2750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.3043623043623043e-06,
+      "loss": 1.8789,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2977022977022977e-06,
+      "loss": 1.677,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2910422910422915e-06,
+      "loss": 1.8919,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.284382284382285e-06,
+      "loss": 2.3004,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2777222777222777e-06,
+      "loss": 2.2049,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.271062271062271e-06,
+      "loss": 1.6747,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2644022644022645e-06,
+      "loss": 2.0845,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.257742257742258e-06,
+      "loss": 1.9814,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.251082251082251e-06,
+      "loss": 1.8545,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2444222444222445e-06,
+      "loss": 2.0408,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.237762237762238e-06,
+      "loss": 2.2447,
+      "step": 2761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2311022311022312e-06,
+      "loss": 2.0417,
+      "step": 2762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2244422244422246e-06,
+      "loss": 2.0239,
+      "step": 2763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.217782217782218e-06,
+      "loss": 1.89,
+      "step": 2764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2111222111222113e-06,
+      "loss": 2.0043,
+      "step": 2765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2044622044622046e-06,
+      "loss": 2.2046,
+      "step": 2766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.197802197802198e-06,
+      "loss": 1.7284,
+      "step": 2767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1911421911421913e-06,
+      "loss": 2.0531,
+      "step": 2768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1844821844821847e-06,
+      "loss": 2.0151,
+      "step": 2769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.177822177822178e-06,
+      "loss": 1.8967,
+      "step": 2770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.1711621711621714e-06,
+      "loss": 2.0339,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1645021645021648e-06,
+      "loss": 2.0054,
+      "step": 2772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.157842157842158e-06,
+      "loss": 1.9374,
+      "step": 2773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.151182151182151e-06,
+      "loss": 1.9793,
+      "step": 2774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.144522144522145e-06,
+      "loss": 1.9044,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.137862137862138e-06,
+      "loss": 1.7161,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1312021312021315e-06,
+      "loss": 1.8227,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1245421245421245e-06,
+      "loss": 1.5168,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.117882117882118e-06,
+      "loss": 2.0689,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1112221112221116e-06,
+      "loss": 2.0753,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.104562104562105e-06,
+      "loss": 1.9704,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0979020979020983e-06,
+      "loss": 2.2385,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0912420912420912e-06,
+      "loss": 1.9084,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0845820845820846e-06,
+      "loss": 2.1634,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0779220779220784e-06,
+      "loss": 2.0201,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0712620712620717e-06,
+      "loss": 2.1233,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0646020646020646e-06,
+      "loss": 1.8229,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.057942057942058e-06,
+      "loss": 2.085,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0512820512820513e-06,
+      "loss": 1.9666,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0446220446220447e-06,
+      "loss": 1.8481,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.037962037962038e-06,
+      "loss": 2.0763,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0313020313020314e-06,
+      "loss": 1.6823,
+      "step": 2792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0246420246420248e-06,
+      "loss": 1.7841,
+      "step": 2793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.017982017982018e-06,
+      "loss": 1.9636,
+      "step": 2794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.0113220113220115e-06,
+      "loss": 2.3038,
+      "step": 2795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.004662004662005e-06,
+      "loss": 1.8493,
+      "step": 2796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.998001998001998e-06,
+      "loss": 1.8274,
+      "step": 2797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9913419913419915e-06,
+      "loss": 2.13,
+      "step": 2798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.984681984681985e-06,
+      "loss": 1.943,
+      "step": 2799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9780219780219782e-06,
+      "loss": 1.737,
+      "step": 2800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9713619713619716e-06,
+      "loss": 1.9928,
+      "step": 2801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.964701964701965e-06,
+      "loss": 2.0522,
+      "step": 2802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9580419580419583e-06,
+      "loss": 1.9215,
+      "step": 2803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9513819513819516e-06,
+      "loss": 1.9929,
+      "step": 2804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.944721944721945e-06,
+      "loss": 2.252,
+      "step": 2805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.938061938061938e-06,
+      "loss": 2.0327,
+      "step": 2806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9314019314019317e-06,
+      "loss": 1.9841,
+      "step": 2807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.924741924741925e-06,
+      "loss": 2.2705,
+      "step": 2808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9180819180819184e-06,
+      "loss": 2.0134,
+      "step": 2809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9114219114219113e-06,
+      "loss": 1.6529,
+      "step": 2810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 1.9499,
+      "step": 2811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8981018981018985e-06,
+      "loss": 2.1204,
+      "step": 2812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8914418914418916e-06,
+      "loss": 2.1039,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.884781884781885e-06,
+      "loss": 1.9148,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8781218781218783e-06,
+      "loss": 1.819,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8714618714618715e-06,
+      "loss": 1.6106,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.864801864801865e-06,
+      "loss": 1.8653,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8581418581418584e-06,
+      "loss": 2.0476,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8514818514818517e-06,
+      "loss": 1.7388,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8448218448218449e-06,
+      "loss": 1.9558,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8381618381618382e-06,
+      "loss": 1.8915,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8315018315018316e-06,
+      "loss": 1.8255,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8248418248418252e-06,
+      "loss": 2.2679,
+      "step": 2823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.9413,
+      "step": 2824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8115218115218116e-06,
+      "loss": 1.9877,
+      "step": 2825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.804861804861805e-06,
+      "loss": 1.9301,
+      "step": 2826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7982017982017981e-06,
+      "loss": 1.9515,
+      "step": 2827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7915417915417917e-06,
+      "loss": 1.8557,
+      "step": 2828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.784881784881785e-06,
+      "loss": 1.7094,
+      "step": 2829
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7782217782217784e-06,
+      "loss": 1.9562,
+      "step": 2830
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7715617715617718e-06,
+      "loss": 1.8862,
+      "step": 2831
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.764901764901765e-06,
+      "loss": 1.9862,
+      "step": 2832
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7582417582417585e-06,
+      "loss": 1.7044,
+      "step": 2833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7515817515817518e-06,
+      "loss": 1.875,
+      "step": 2834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7449217449217452e-06,
+      "loss": 2.4221,
+      "step": 2835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7382617382617383e-06,
+      "loss": 1.9546,
+      "step": 2836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7316017316017317e-06,
+      "loss": 1.9833,
+      "step": 2837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.724941724941725e-06,
+      "loss": 2.2547,
+      "step": 2838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7182817182817186e-06,
+      "loss": 1.9999,
+      "step": 2839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7116217116217117e-06,
+      "loss": 1.7458,
+      "step": 2840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.704961704961705e-06,
+      "loss": 1.953,
+      "step": 2841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6983016983016984e-06,
+      "loss": 2.1139,
+      "step": 2842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6916416916416916e-06,
+      "loss": 2.0455,
+      "step": 2843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6849816849816852e-06,
+      "loss": 1.9555,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6783216783216785e-06,
+      "loss": 1.9777,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6716616716616719e-06,
+      "loss": 1.864,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.665001665001665e-06,
+      "loss": 2.3211,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6583416583416584e-06,
+      "loss": 2.0073,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.651681651681652e-06,
+      "loss": 1.9756,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6450216450216453e-06,
+      "loss": 1.9428,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6383616383616386e-06,
+      "loss": 1.542,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6317016317016318e-06,
+      "loss": 1.9811,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250416250416251e-06,
+      "loss": 1.8381,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6183816183816185e-06,
+      "loss": 1.7823,
+      "step": 2854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.611721611721612e-06,
+      "loss": 1.9113,
+      "step": 2855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6050616050616052e-06,
+      "loss": 1.9139,
+      "step": 2856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5984015984015985e-06,
+      "loss": 2.0163,
+      "step": 2857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5917415917415919e-06,
+      "loss": 2.3367,
+      "step": 2858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.585081585081585e-06,
+      "loss": 1.9127,
+      "step": 2859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5784215784215786e-06,
+      "loss": 2.0498,
+      "step": 2860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.571761571761572e-06,
+      "loss": 1.8497,
+      "step": 2861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5651015651015653e-06,
+      "loss": 1.9486,
+      "step": 2862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5584415584415584e-06,
+      "loss": 1.8664,
+      "step": 2863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5517815517815518e-06,
+      "loss": 1.8083,
+      "step": 2864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5451215451215454e-06,
+      "loss": 2.165,
+      "step": 2865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 1.9013,
+      "step": 2866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5318015318015319e-06,
+      "loss": 2.0324,
+      "step": 2867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5251415251415252e-06,
+      "loss": 2.0099,
+      "step": 2868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5184815184815186e-06,
+      "loss": 1.7755,
+      "step": 2869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.511821511821512e-06,
+      "loss": 1.8476,
+      "step": 2870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5051615051615055e-06,
+      "loss": 1.8851,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4985014985014986e-06,
+      "loss": 1.933,
+      "step": 2872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.491841491841492e-06,
+      "loss": 1.7756,
+      "step": 2873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4851814851814853e-06,
+      "loss": 1.8031,
+      "step": 2874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4785214785214785e-06,
+      "loss": 1.8866,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.471861471861472e-06,
+      "loss": 2.0067,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4652014652014654e-06,
+      "loss": 1.8909,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4585414585414587e-06,
+      "loss": 1.7809,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4518814518814519e-06,
+      "loss": 2.0218,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4452214452214452e-06,
+      "loss": 1.7879,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4385614385614388e-06,
+      "loss": 2.2144,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4319014319014322e-06,
+      "loss": 1.9877,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4252414252414253e-06,
+      "loss": 1.6677,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4185814185814187e-06,
+      "loss": 1.8754,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.411921411921412e-06,
+      "loss": 1.6221,
+      "step": 2885
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4052614052614052e-06,
+      "loss": 1.5338,
+      "step": 2886
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3986013986013987e-06,
+      "loss": 1.7822,
+      "step": 2887
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.391941391941392e-06,
+      "loss": 2.2628,
+      "step": 2888
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3852813852813854e-06,
+      "loss": 1.742,
+      "step": 2889
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3786213786213788e-06,
+      "loss": 1.9622,
+      "step": 2890
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.371961371961372e-06,
+      "loss": 2.0785,
+      "step": 2891
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3653013653013655e-06,
+      "loss": 1.7295,
+      "step": 2892
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3586413586413588e-06,
+      "loss": 1.8134,
+      "step": 2893
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3519813519813522e-06,
+      "loss": 2.1421,
+      "step": 2894
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3453213453213453e-06,
+      "loss": 1.8368,
+      "step": 2895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3386613386613387e-06,
+      "loss": 1.7578,
+      "step": 2896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3320013320013323e-06,
+      "loss": 2.1298,
+      "step": 2897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3253413253413256e-06,
+      "loss": 1.8168,
+      "step": 2898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3186813186813187e-06,
+      "loss": 1.984,
+      "step": 2899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.312021312021312e-06,
+      "loss": 1.9197,
+      "step": 2900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.3053613053613055e-06,
+      "loss": 1.9744,
+      "step": 2901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2987012987012986e-06,
+      "loss": 1.7817,
+      "step": 2902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2920412920412922e-06,
+      "loss": 2.069,
+      "step": 2903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2853812853812855e-06,
+      "loss": 1.8945,
+      "step": 2904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2787212787212789e-06,
+      "loss": 2.0727,
+      "step": 2905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.272061272061272e-06,
+      "loss": 2.2375,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2654012654012654e-06,
+      "loss": 2.0984,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.258741258741259e-06,
+      "loss": 1.9086,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2520812520812523e-06,
+      "loss": 1.975,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2454212454212456e-06,
+      "loss": 1.8898,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2387612387612388e-06,
+      "loss": 1.9963,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2321012321012323e-06,
+      "loss": 1.98,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2254412254412255e-06,
+      "loss": 2.2493,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2187812187812188e-06,
+      "loss": 2.0349,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.9104,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.2054612054612055e-06,
+      "loss": 1.8888,
+      "step": 2916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.198801198801199e-06,
+      "loss": 1.9641,
+      "step": 2917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1921411921411923e-06,
+      "loss": 1.7532,
+      "step": 2918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1854811854811856e-06,
+      "loss": 2.096,
+      "step": 2919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.178821178821179e-06,
+      "loss": 1.9939,
+      "step": 2920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1721611721611723e-06,
+      "loss": 1.5927,
+      "step": 2921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1655011655011655e-06,
+      "loss": 1.7146,
+      "step": 2922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.158841158841159e-06,
+      "loss": 2.0961,
+      "step": 2923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1521811521811522e-06,
+      "loss": 1.9332,
+      "step": 2924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1455211455211457e-06,
+      "loss": 1.9317,
+      "step": 2925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1388611388611389e-06,
+      "loss": 2.0989,
+      "step": 2926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1322011322011322e-06,
+      "loss": 1.7574,
+      "step": 2927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1255411255411256e-06,
+      "loss": 1.7542,
+      "step": 2928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.118881118881119e-06,
+      "loss": 2.0161,
+      "step": 2929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1122211122211123e-06,
+      "loss": 1.7854,
+      "step": 2930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1055611055611056e-06,
+      "loss": 1.9447,
+      "step": 2931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.098901098901099e-06,
+      "loss": 1.7204,
+      "step": 2932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0922410922410923e-06,
+      "loss": 1.7748,
+      "step": 2933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0855810855810857e-06,
+      "loss": 1.822,
+      "step": 2934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.078921078921079e-06,
+      "loss": 2.108,
+      "step": 2935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0722610722610724e-06,
+      "loss": 1.8686,
+      "step": 2936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0656010656010658e-06,
+      "loss": 2.0485,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.058941058941059e-06,
+      "loss": 1.9032,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0522810522810525e-06,
+      "loss": 2.0574,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0456210456210456e-06,
+      "loss": 2.1938,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0389610389610392e-06,
+      "loss": 2.1654,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0323010323010323e-06,
+      "loss": 2.0828,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0256410256410257e-06,
+      "loss": 1.6676,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.018981018981019e-06,
+      "loss": 1.9626,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0123210123210124e-06,
+      "loss": 1.7018,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0056610056610057e-06,
+      "loss": 2.0649,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99000999000999e-07,
+      "loss": 1.7595,
+      "step": 2947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.923409923409924e-07,
+      "loss": 1.9718,
+      "step": 2948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.856809856809858e-07,
+      "loss": 2.0403,
+      "step": 2949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.790209790209791e-07,
+      "loss": 2.2424,
+      "step": 2950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.723609723609725e-07,
+      "loss": 1.9791,
+      "step": 2951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.657009657009659e-07,
+      "loss": 1.7644,
+      "step": 2952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.590409590409592e-07,
+      "loss": 1.5541,
+      "step": 2953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 2.1386,
+      "step": 2954
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.457209457209458e-07,
+      "loss": 2.1797,
+      "step": 2955
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.390609390609392e-07,
+      "loss": 2.1256,
+      "step": 2956
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.324009324009325e-07,
+      "loss": 1.9454,
+      "step": 2957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.257409257409259e-07,
+      "loss": 1.9889,
+      "step": 2958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.190809190809191e-07,
+      "loss": 1.9281,
+      "step": 2959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.124209124209126e-07,
+      "loss": 2.1418,
+      "step": 2960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.057609057609058e-07,
+      "loss": 1.8493,
+      "step": 2961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.991008991008991e-07,
+      "loss": 2.161,
+      "step": 2962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.924408924408925e-07,
+      "loss": 1.7987,
+      "step": 2963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.857808857808859e-07,
+      "loss": 1.9855,
+      "step": 2964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.791208791208792e-07,
+      "loss": 1.9788,
+      "step": 2965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.724608724608726e-07,
+      "loss": 1.763,
+      "step": 2966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.658008658008658e-07,
+      "loss": 2.1942,
+      "step": 2967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.591408591408593e-07,
+      "loss": 1.6358,
+      "step": 2968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.524808524808525e-07,
+      "loss": 1.8377,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.458208458208458e-07,
+      "loss": 2.0005,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.391608391608393e-07,
+      "loss": 2.323,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.325008325008325e-07,
+      "loss": 1.8735,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.25840825840826e-07,
+      "loss": 1.9494,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.191808191808193e-07,
+      "loss": 2.0182,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.125208125208126e-07,
+      "loss": 2.0502,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.05860805860806e-07,
+      "loss": 1.9014,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.992007992007993e-07,
+      "loss": 2.062,
+      "step": 2977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925407925407925e-07,
+      "loss": 1.9031,
+      "step": 2978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.85880785880786e-07,
+      "loss": 1.859,
+      "step": 2979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.792207792207792e-07,
+      "loss": 1.9504,
+      "step": 2980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.725607725607727e-07,
+      "loss": 2.1963,
+      "step": 2981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.659007659007659e-07,
+      "loss": 1.9263,
+      "step": 2982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.592407592407593e-07,
+      "loss": 1.9932,
+      "step": 2983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.525807525807527e-07,
+      "loss": 1.837,
+      "step": 2984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.45920745920746e-07,
+      "loss": 2.0429,
+      "step": 2985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.392607392607392e-07,
+      "loss": 1.824,
+      "step": 2986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.326007326007327e-07,
+      "loss": 1.8654,
+      "step": 2987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.259407259407259e-07,
+      "loss": 2.0405,
+      "step": 2988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.192807192807194e-07,
+      "loss": 1.8741,
+      "step": 2989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.126207126207127e-07,
+      "loss": 2.1816,
+      "step": 2990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.05960705960706e-07,
+      "loss": 2.3181,
+      "step": 2991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.993006993006994e-07,
+      "loss": 1.7361,
+      "step": 2992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.926406926406927e-07,
+      "loss": 1.9033,
+      "step": 2993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.85980685980686e-07,
+      "loss": 1.7238,
+      "step": 2994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.793206793206794e-07,
+      "loss": 1.764,
+      "step": 2995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.726606726606727e-07,
+      "loss": 1.9648,
+      "step": 2996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.660006660006661e-07,
+      "loss": 1.5611,
+      "step": 2997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.593406593406594e-07,
+      "loss": 1.991,
+      "step": 2998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.526806526806527e-07,
+      "loss": 1.953,
+      "step": 2999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.460206460206461e-07,
+      "loss": 2.0093,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.393606393606394e-07,
+      "loss": 2.3074,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.327006327006327e-07,
+      "loss": 1.8217,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.260406260406261e-07,
+      "loss": 2.094,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.193806193806194e-07,
+      "loss": 1.9225,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.127206127206127e-07,
+      "loss": 1.7851,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 1.9553,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.994005994005995e-07,
+      "loss": 1.9821,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.927405927405928e-07,
+      "loss": 2.2589,
+      "step": 3008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.860805860805862e-07,
+      "loss": 2.0012,
+      "step": 3009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.794205794205795e-07,
+      "loss": 1.8151,
+      "step": 3010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.727605727605729e-07,
+      "loss": 2.1071,
+      "step": 3011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.661005661005661e-07,
+      "loss": 2.2224,
+      "step": 3012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.594405594405595e-07,
+      "loss": 1.8227,
+      "step": 3013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.527805527805528e-07,
+      "loss": 2.1129,
+      "step": 3014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.461205461205462e-07,
+      "loss": 1.9054,
+      "step": 3015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.394605394605395e-07,
+      "loss": 1.915,
+      "step": 3016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.328005328005329e-07,
+      "loss": 1.9488,
+      "step": 3017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.261405261405262e-07,
+      "loss": 1.951,
+      "step": 3018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.194805194805196e-07,
+      "loss": 1.6936,
+      "step": 3019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.128205128205128e-07,
+      "loss": 1.7822,
+      "step": 3020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.061605061605062e-07,
+      "loss": 1.9978,
+      "step": 3021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.995004995004995e-07,
+      "loss": 2.0592,
+      "step": 3022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.928404928404929e-07,
+      "loss": 1.8165,
+      "step": 3023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.861804861804862e-07,
+      "loss": 1.6134,
+      "step": 3024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.795204795204796e-07,
+      "loss": 1.935,
+      "step": 3025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.728604728604729e-07,
+      "loss": 1.9563,
+      "step": 3026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.6620046620046626e-07,
+      "loss": 2.2317,
+      "step": 3027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5954045954045956e-07,
+      "loss": 1.9661,
+      "step": 3028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.528804528804529e-07,
+      "loss": 1.7538,
+      "step": 3029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.4622044622044626e-07,
+      "loss": 1.9068,
+      "step": 3030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.395604395604396e-07,
+      "loss": 1.9978,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.329004329004329e-07,
+      "loss": 1.9518,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.2624042624042627e-07,
+      "loss": 2.1924,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.195804195804196e-07,
+      "loss": 2.2352,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.12920412920413e-07,
+      "loss": 1.6723,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.062604062604063e-07,
+      "loss": 1.9851,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9960039960039963e-07,
+      "loss": 1.8504,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.92940392940393e-07,
+      "loss": 1.9305,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8628038628038634e-07,
+      "loss": 2.4712,
+      "step": 3039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7962037962037964e-07,
+      "loss": 1.5061,
+      "step": 3040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.72960372960373e-07,
+      "loss": 2.0422,
+      "step": 3041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6630036630036635e-07,
+      "loss": 2.0918,
+      "step": 3042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.596403596403597e-07,
+      "loss": 2.0016,
+      "step": 3043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.52980352980353e-07,
+      "loss": 2.2038,
+      "step": 3044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4632034632034636e-07,
+      "loss": 1.9557,
+      "step": 3045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.396603396603397e-07,
+      "loss": 2.3211,
+      "step": 3046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3300033300033306e-07,
+      "loss": 1.9764,
+      "step": 3047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2634032634032636e-07,
+      "loss": 2.1558,
+      "step": 3048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.196803196803197e-07,
+      "loss": 1.792,
+      "step": 3049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1302031302031307e-07,
+      "loss": 2.1205,
+      "step": 3050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0636030636030637e-07,
+      "loss": 2.1813,
+      "step": 3051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.997002997002997e-07,
+      "loss": 1.9318,
+      "step": 3052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.930402930402931e-07,
+      "loss": 1.9583,
+      "step": 3053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8638028638028643e-07,
+      "loss": 1.9695,
+      "step": 3054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7972027972027973e-07,
+      "loss": 1.6165,
+      "step": 3055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.730602730602731e-07,
+      "loss": 1.8683,
+      "step": 3056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6640026640026644e-07,
+      "loss": 1.988,
+      "step": 3057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.597402597402598e-07,
+      "loss": 2.0171,
+      "step": 3058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.530802530802531e-07,
+      "loss": 2.1562,
+      "step": 3059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4642024642024645e-07,
+      "loss": 1.6212,
+      "step": 3060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.397602397602398e-07,
+      "loss": 1.887,
+      "step": 3061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3310023310023313e-07,
+      "loss": 1.8205,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2644022644022646e-07,
+      "loss": 2.0649,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.197802197802198e-07,
+      "loss": 1.8527,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1312021312021314e-07,
+      "loss": 1.781,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.064602064602065e-07,
+      "loss": 2.0159,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9980019980019982e-07,
+      "loss": 2.0297,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9314019314019317e-07,
+      "loss": 1.806,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.864801864801865e-07,
+      "loss": 2.0162,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7982017982017985e-07,
+      "loss": 1.8382,
+      "step": 3070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7316017316017318e-07,
+      "loss": 1.7333,
+      "step": 3071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6650016650016653e-07,
+      "loss": 2.0272,
+      "step": 3072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5984015984015986e-07,
+      "loss": 2.4066,
+      "step": 3073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5318015318015319e-07,
+      "loss": 2.076,
+      "step": 3074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4652014652014654e-07,
+      "loss": 1.8545,
+      "step": 3075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3986013986013987e-07,
+      "loss": 2.2329,
+      "step": 3076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3320013320013322e-07,
+      "loss": 1.9209,
+      "step": 3077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2654012654012655e-07,
+      "loss": 1.7284,
+      "step": 3078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.198801198801199e-07,
+      "loss": 2.1665,
+      "step": 3079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1322011322011323e-07,
+      "loss": 1.7916,
+      "step": 3080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0656010656010657e-07,
+      "loss": 1.6961,
+      "step": 3081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.990009990009991e-08,
+      "loss": 1.8765,
+      "step": 3082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.324009324009325e-08,
+      "loss": 2.2132,
+      "step": 3083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.658008658008659e-08,
+      "loss": 1.8369,
+      "step": 3084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.992007992007993e-08,
+      "loss": 1.778,
+      "step": 3085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.326007326007327e-08,
+      "loss": 2.1466,
+      "step": 3086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.660006660006661e-08,
+      "loss": 1.9119,
+      "step": 3087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.994005994005995e-08,
+      "loss": 2.0758,
+      "step": 3088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.3280053280053284e-08,
+      "loss": 1.8512,
+      "step": 3089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.6620046620046624e-08,
+      "loss": 1.9395,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9960039960039965e-08,
+      "loss": 2.0431,
+      "step": 3091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3300033300033305e-08,
+      "loss": 1.8644,
+      "step": 3092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6640026640026642e-08,
+      "loss": 1.8284,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9980019980019982e-08,
+      "loss": 2.0181,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3320013320013321e-08,
+      "loss": 2.1013,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.6600066600066605e-09,
+      "loss": 2.2829,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "step": 3096,
+      "total_flos": 152356197826560.0,
+      "train_loss": 2.122519214254941,
+      "train_runtime": 34517.9305,
+      "train_samples_per_second": 5.741,
+      "train_steps_per_second": 0.09
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3096,
+  "num_train_epochs": 1,
+  "save_steps": 2000,
+  "total_flos": 152356197826560.0,
+  "trial_name": null,
+  "trial_params": null
+}