diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,43387 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "global_step": 7224,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999584717607973e-07,
+      "loss": 0.2741,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9991694352159466e-07,
+      "loss": 0.298,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99875415282392e-07,
+      "loss": 0.3082,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9983388704318933e-07,
+      "loss": 0.2721,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.997923588039867e-07,
+      "loss": 0.2863,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99750830564784e-07,
+      "loss": 0.317,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.997093023255814e-07,
+      "loss": 0.2503,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9966777408637873e-07,
+      "loss": 0.2891,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9962624584717607e-07,
+      "loss": 0.2888,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.995847176079734e-07,
+      "loss": 0.2852,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9954318936877074e-07,
+      "loss": 0.2454,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.995016611295681e-07,
+      "loss": 0.2783,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.994601328903654e-07,
+      "loss": 0.2842,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9941860465116275e-07,
+      "loss": 0.3016,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9937707641196014e-07,
+      "loss": 0.349,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9933554817275743e-07,
+      "loss": 0.2816,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.992940199335548e-07,
+      "loss": 0.2859,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9925249169435215e-07,
+      "loss": 0.3002,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.992109634551495e-07,
+      "loss": 0.3092,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9916943521594683e-07,
+      "loss": 0.3506,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9912790697674416e-07,
+      "loss": 0.2695,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.990863787375415e-07,
+      "loss": 0.2432,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9904485049833884e-07,
+      "loss": 0.2915,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.990033222591362e-07,
+      "loss": 0.2355,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9896179401993356e-07,
+      "loss": 0.256,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.989202657807309e-07,
+      "loss": 0.3233,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9887873754152824e-07,
+      "loss": 0.2532,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.988372093023256e-07,
+      "loss": 0.2992,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.987956810631229e-07,
+      "loss": 0.2946,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9875415282392025e-07,
+      "loss": 0.2447,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.987126245847176e-07,
+      "loss": 0.3025,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.986710963455149e-07,
+      "loss": 0.2576,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.986295681063123e-07,
+      "loss": 0.2625,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.985880398671096e-07,
+      "loss": 0.2981,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98546511627907e-07,
+      "loss": 0.2541,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.985049833887043e-07,
+      "loss": 0.3089,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9846345514950166e-07,
+      "loss": 0.2398,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98421926910299e-07,
+      "loss": 0.313,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9838039867109633e-07,
+      "loss": 0.3359,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9833887043189367e-07,
+      "loss": 0.2941,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.98297342192691e-07,
+      "loss": 0.2578,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9825581395348834e-07,
+      "loss": 0.3146,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9821428571428573e-07,
+      "loss": 0.3113,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9817275747508307e-07,
+      "loss": 0.269,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9813122923588035e-07,
+      "loss": 0.2985,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9808970099667774e-07,
+      "loss": 0.2797,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.980481727574751e-07,
+      "loss": 0.3325,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.980066445182724e-07,
+      "loss": 0.2696,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9796511627906975e-07,
+      "loss": 0.2875,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.979235880398671e-07,
+      "loss": 0.2701,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9788205980066443e-07,
+      "loss": 0.3241,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9784053156146176e-07,
+      "loss": 0.2942,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.977990033222591e-07,
+      "loss": 0.2576,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.977574750830565e-07,
+      "loss": 0.3741,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.977159468438538e-07,
+      "loss": 0.304,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9767441860465116e-07,
+      "loss": 0.3331,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.976328903654485e-07,
+      "loss": 0.274,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9759136212624584e-07,
+      "loss": 0.3276,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.975498338870432e-07,
+      "loss": 0.3786,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.975083056478405e-07,
+      "loss": 0.293,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9746677740863785e-07,
+      "loss": 0.2897,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.974252491694352e-07,
+      "loss": 0.2906,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.973837209302325e-07,
+      "loss": 0.241,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.973421926910299e-07,
+      "loss": 0.3048,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9730066445182725e-07,
+      "loss": 0.2733,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.972591362126246e-07,
+      "loss": 0.3397,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.972176079734219e-07,
+      "loss": 0.2915,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9717607973421926e-07,
+      "loss": 0.315,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.971345514950166e-07,
+      "loss": 0.3171,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9709302325581393e-07,
+      "loss": 0.3146,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9705149501661127e-07,
+      "loss": 0.2948,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9700996677740866e-07,
+      "loss": 0.3454,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9696843853820594e-07,
+      "loss": 0.328,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9692691029900333e-07,
+      "loss": 0.2541,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9688538205980067e-07,
+      "loss": 0.2976,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9684385382059795e-07,
+      "loss": 0.245,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9680232558139534e-07,
+      "loss": 0.3014,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.967607973421927e-07,
+      "loss": 0.3208,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9671926910299e-07,
+      "loss": 0.3238,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9667774086378735e-07,
+      "loss": 0.2578,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.966362126245847e-07,
+      "loss": 0.3301,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.965946843853821e-07,
+      "loss": 0.3101,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9655315614617936e-07,
+      "loss": 0.3343,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.965116279069767e-07,
+      "loss": 0.3157,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.964700996677741e-07,
+      "loss": 0.2906,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9642857142857143e-07,
+      "loss": 0.2642,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9638704318936876e-07,
+      "loss": 0.2445,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.963455149501661e-07,
+      "loss": 0.3,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9630398671096344e-07,
+      "loss": 0.2857,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.962624584717608e-07,
+      "loss": 0.2647,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.962209302325581e-07,
+      "loss": 0.2667,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9617940199335545e-07,
+      "loss": 0.2486,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9613787375415284e-07,
+      "loss": 0.2789,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.960963455149501e-07,
+      "loss": 0.3256,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.960548172757475e-07,
+      "loss": 0.2463,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9601328903654485e-07,
+      "loss": 0.2892,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.959717607973422e-07,
+      "loss": 0.3467,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.959302325581395e-07,
+      "loss": 0.3141,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9588870431893686e-07,
+      "loss": 0.3108,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.958471760797342e-07,
+      "loss": 0.2623,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9580564784053153e-07,
+      "loss": 0.2824,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9576411960132887e-07,
+      "loss": 0.2866,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9572259136212626e-07,
+      "loss": 0.3452,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.956810631229236e-07,
+      "loss": 0.3129,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9563953488372093e-07,
+      "loss": 0.2922,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9559800664451827e-07,
+      "loss": 0.2777,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.955564784053156e-07,
+      "loss": 0.2439,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9551495016611294e-07,
+      "loss": 0.2286,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.954734219269103e-07,
+      "loss": 0.3625,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.954318936877076e-07,
+      "loss": 0.3402,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.95390365448505e-07,
+      "loss": 0.3505,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.953488372093023e-07,
+      "loss": 0.2611,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.953073089700997e-07,
+      "loss": 0.2819,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.95265780730897e-07,
+      "loss": 0.2339,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.952242524916943e-07,
+      "loss": 0.385,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.951827242524917e-07,
+      "loss": 0.3715,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.95141196013289e-07,
+      "loss": 0.2619,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9509966777408636e-07,
+      "loss": 0.3402,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.950581395348837e-07,
+      "loss": 0.3203,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9501661129568104e-07,
+      "loss": 0.2989,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.949750830564784e-07,
+      "loss": 0.2542,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.949335548172757e-07,
+      "loss": 0.2348,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9489202657807305e-07,
+      "loss": 0.2668,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9485049833887044e-07,
+      "loss": 0.2682,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9480897009966777e-07,
+      "loss": 0.2857,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.947674418604651e-07,
+      "loss": 0.3198,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9472591362126245e-07,
+      "loss": 0.2465,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.946843853820598e-07,
+      "loss": 0.304,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.946428571428571e-07,
+      "loss": 0.2837,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9460132890365446e-07,
+      "loss": 0.2806,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.945598006644518e-07,
+      "loss": 0.2755,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.945182724252492e-07,
+      "loss": 0.273,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9447674418604647e-07,
+      "loss": 0.2413,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9443521594684386e-07,
+      "loss": 0.3055,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.943936877076412e-07,
+      "loss": 0.2753,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9435215946843853e-07,
+      "loss": 0.2857,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9431063122923587e-07,
+      "loss": 0.2718,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.942691029900332e-07,
+      "loss": 0.2652,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9422757475083054e-07,
+      "loss": 0.2194,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.941860465116279e-07,
+      "loss": 0.2602,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.941445182724252e-07,
+      "loss": 0.2604,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.941029900332226e-07,
+      "loss": 0.3043,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.940614617940199e-07,
+      "loss": 0.2558,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.940199335548173e-07,
+      "loss": 0.2332,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.939784053156146e-07,
+      "loss": 0.3264,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9393687707641195e-07,
+      "loss": 0.2493,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.938953488372093e-07,
+      "loss": 0.2643,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.938538205980066e-07,
+      "loss": 0.2607,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9381229235880396e-07,
+      "loss": 0.3177,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.937707641196013e-07,
+      "loss": 0.2819,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9372923588039864e-07,
+      "loss": 0.2581,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.93687707641196e-07,
+      "loss": 0.2821,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9364617940199336e-07,
+      "loss": 0.2901,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.936046511627907e-07,
+      "loss": 0.2833,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9356312292358804e-07,
+      "loss": 0.3169,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9352159468438537e-07,
+      "loss": 0.3144,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.934800664451827e-07,
+      "loss": 0.2796,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9343853820598005e-07,
+      "loss": 0.3108,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.933970099667774e-07,
+      "loss": 0.297,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9335548172757477e-07,
+      "loss": 0.3041,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9331395348837206e-07,
+      "loss": 0.3225,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9327242524916945e-07,
+      "loss": 0.27,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.932308970099668e-07,
+      "loss": 0.2324,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.931893687707641e-07,
+      "loss": 0.2898,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9314784053156146e-07,
+      "loss": 0.3077,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.931063122923588e-07,
+      "loss": 0.2773,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9306478405315613e-07,
+      "loss": 0.358,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9302325581395347e-07,
+      "loss": 0.3093,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.929817275747508e-07,
+      "loss": 0.2787,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9294019933554814e-07,
+      "loss": 0.3356,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9289867109634553e-07,
+      "loss": 0.2829,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.928571428571428e-07,
+      "loss": 0.3034,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.928156146179402e-07,
+      "loss": 0.3346,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9277408637873754e-07,
+      "loss": 0.3622,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.927325581395349e-07,
+      "loss": 0.35,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.926910299003322e-07,
+      "loss": 0.2886,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9264950166112955e-07,
+      "loss": 0.308,
+      "step": 177
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.926079734219269e-07,
+      "loss": 0.3198,
+      "step": 178
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.925664451827242e-07,
+      "loss": 0.3252,
+      "step": 179
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9252491694352156e-07,
+      "loss": 0.3458,
+      "step": 180
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9248338870431895e-07,
+      "loss": 0.2528,
+      "step": 181
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9244186046511623e-07,
+      "loss": 0.2478,
+      "step": 182
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.924003322259136e-07,
+      "loss": 0.2454,
+      "step": 183
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9235880398671096e-07,
+      "loss": 0.2987,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.923172757475083e-07,
+      "loss": 0.276,
+      "step": 185
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9227574750830563e-07,
+      "loss": 0.319,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9223421926910297e-07,
+      "loss": 0.2899,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.921926910299003e-07,
+      "loss": 0.3155,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9215116279069764e-07,
+      "loss": 0.275,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.92109634551495e-07,
+      "loss": 0.3242,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9206810631229237e-07,
+      "loss": 0.2885,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.920265780730897e-07,
+      "loss": 0.2743,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9198504983388704e-07,
+      "loss": 0.3149,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.919435215946844e-07,
+      "loss": 0.3019,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.919019933554817e-07,
+      "loss": 0.2882,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9186046511627906e-07,
+      "loss": 0.2858,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.918189368770764e-07,
+      "loss": 0.2699,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9177740863787373e-07,
+      "loss": 0.3015,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.917358803986711e-07,
+      "loss": 0.2433,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.916943521594684e-07,
+      "loss": 0.3095,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.916528239202658e-07,
+      "loss": 0.2829,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9161129568106313e-07,
+      "loss": 0.2664,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.915697674418604e-07,
+      "loss": 0.2957,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.915282392026578e-07,
+      "loss": 0.397,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9148671096345514e-07,
+      "loss": 0.3311,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.914451827242525e-07,
+      "loss": 0.2397,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.914036544850498e-07,
+      "loss": 0.3699,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9136212624584715e-07,
+      "loss": 0.3062,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9132059800664454e-07,
+      "loss": 0.2851,
+      "step": 209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.912790697674418e-07,
+      "loss": 0.3223,
+      "step": 210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9123754152823916e-07,
+      "loss": 0.264,
+      "step": 211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9119601328903655e-07,
+      "loss": 0.3219,
+      "step": 212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.911544850498339e-07,
+      "loss": 0.3043,
+      "step": 213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.911129568106312e-07,
+      "loss": 0.2985,
+      "step": 214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9107142857142856e-07,
+      "loss": 0.2877,
+      "step": 215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.910299003322259e-07,
+      "loss": 0.298,
+      "step": 216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9098837209302323e-07,
+      "loss": 0.2641,
+      "step": 217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9094684385382057e-07,
+      "loss": 0.2741,
+      "step": 218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.909053156146179e-07,
+      "loss": 0.2992,
+      "step": 219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.908637873754153e-07,
+      "loss": 0.2838,
+      "step": 220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.908222591362126e-07,
+      "loss": 0.2861,
+      "step": 221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9078073089700997e-07,
+      "loss": 0.2641,
+      "step": 222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.907392026578073e-07,
+      "loss": 0.267,
+      "step": 223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9069767441860464e-07,
+      "loss": 0.2482,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.90656146179402e-07,
+      "loss": 0.2809,
+      "step": 225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.906146179401993e-07,
+      "loss": 0.2585,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9057308970099665e-07,
+      "loss": 0.2967,
+      "step": 227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.90531561461794e-07,
+      "loss": 0.3149,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9049003322259133e-07,
+      "loss": 0.3015,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.904485049833887e-07,
+      "loss": 0.3158,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9040697674418605e-07,
+      "loss": 0.3409,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.903654485049834e-07,
+      "loss": 0.3315,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9032392026578073e-07,
+      "loss": 0.3379,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9028239202657806e-07,
+      "loss": 0.242,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.902408637873754e-07,
+      "loss": 0.26,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9019933554817274e-07,
+      "loss": 0.2971,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.901578073089701e-07,
+      "loss": 0.4209,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9011627906976747e-07,
+      "loss": 0.3775,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9007475083056475e-07,
+      "loss": 0.2726,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9003322259136214e-07,
+      "loss": 0.2972,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.899916943521595e-07,
+      "loss": 0.2824,
+      "step": 241
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8995016611295676e-07,
+      "loss": 0.2669,
+      "step": 242
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8990863787375415e-07,
+      "loss": 0.289,
+      "step": 243
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.898671096345515e-07,
+      "loss": 0.3432,
+      "step": 244
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.898255813953488e-07,
+      "loss": 0.3387,
+      "step": 245
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8978405315614616e-07,
+      "loss": 0.2627,
+      "step": 246
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.897425249169435e-07,
+      "loss": 0.2743,
+      "step": 247
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.897009966777409e-07,
+      "loss": 0.2827,
+      "step": 248
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8965946843853817e-07,
+      "loss": 0.2614,
+      "step": 249
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.896179401993355e-07,
+      "loss": 0.2486,
+      "step": 250
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.895764119601329e-07,
+      "loss": 0.3232,
+      "step": 251
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8953488372093023e-07,
+      "loss": 0.4173,
+      "step": 252
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8949335548172757e-07,
+      "loss": 0.3825,
+      "step": 253
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.894518272425249e-07,
+      "loss": 0.3054,
+      "step": 254
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8941029900332224e-07,
+      "loss": 0.2975,
+      "step": 255
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.893687707641196e-07,
+      "loss": 0.2936,
+      "step": 256
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.893272425249169e-07,
+      "loss": 0.2916,
+      "step": 257
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8928571428571425e-07,
+      "loss": 0.2678,
+      "step": 258
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8924418604651164e-07,
+      "loss": 0.3074,
+      "step": 259
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8920265780730893e-07,
+      "loss": 0.2892,
+      "step": 260
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.891611295681063e-07,
+      "loss": 0.3046,
+      "step": 261
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8911960132890365e-07,
+      "loss": 0.3055,
+      "step": 262
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.89078073089701e-07,
+      "loss": 0.3146,
+      "step": 263
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8903654485049833e-07,
+      "loss": 0.277,
+      "step": 264
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8899501661129566e-07,
+      "loss": 0.3129,
+      "step": 265
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.88953488372093e-07,
+      "loss": 0.2836,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8891196013289034e-07,
+      "loss": 0.2637,
+      "step": 267
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.888704318936877e-07,
+      "loss": 0.3238,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8882890365448506e-07,
+      "loss": 0.2981,
+      "step": 269
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8878737541528235e-07,
+      "loss": 0.2864,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8874584717607974e-07,
+      "loss": 0.2685,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.887043189368771e-07,
+      "loss": 0.3212,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.886627906976744e-07,
+      "loss": 0.2843,
+      "step": 273
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8862126245847175e-07,
+      "loss": 0.3263,
+      "step": 274
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.885797342192691e-07,
+      "loss": 0.3284,
+      "step": 275
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.885382059800664e-07,
+      "loss": 0.3223,
+      "step": 276
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8849667774086376e-07,
+      "loss": 0.3633,
+      "step": 277
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.884551495016611e-07,
+      "loss": 0.2605,
+      "step": 278
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.884136212624585e-07,
+      "loss": 0.2646,
+      "step": 279
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.883720930232558e-07,
+      "loss": 0.2616,
+      "step": 280
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.883305647840531e-07,
+      "loss": 0.2985,
+      "step": 281
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.882890365448505e-07,
+      "loss": 0.2997,
+      "step": 282
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8824750830564783e-07,
+      "loss": 0.2349,
+      "step": 283
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8820598006644517e-07,
+      "loss": 0.3191,
+      "step": 284
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.881644518272425e-07,
+      "loss": 0.3256,
+      "step": 285
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8812292358803984e-07,
+      "loss": 0.3169,
+      "step": 286
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8808139534883723e-07,
+      "loss": 0.2823,
+      "step": 287
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.880398671096345e-07,
+      "loss": 0.3323,
+      "step": 288
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8799833887043185e-07,
+      "loss": 0.2873,
+      "step": 289
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8795681063122924e-07,
+      "loss": 0.3583,
+      "step": 290
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.879152823920266e-07,
+      "loss": 0.2889,
+      "step": 291
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.878737541528239e-07,
+      "loss": 0.3346,
+      "step": 292
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8783222591362125e-07,
+      "loss": 0.3143,
+      "step": 293
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.877906976744186e-07,
+      "loss": 0.2772,
+      "step": 294
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8774916943521593e-07,
+      "loss": 0.2741,
+      "step": 295
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8770764119601326e-07,
+      "loss": 0.2524,
+      "step": 296
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.876661129568106e-07,
+      "loss": 0.3699,
+      "step": 297
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.87624584717608e-07,
+      "loss": 0.2611,
+      "step": 298
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.875830564784053e-07,
+      "loss": 0.3144,
+      "step": 299
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8754152823920266e-07,
+      "loss": 0.3147,
+      "step": 300
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.875e-07,
+      "loss": 0.2896,
+      "step": 301
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8745847176079734e-07,
+      "loss": 0.2456,
+      "step": 302
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.874169435215947e-07,
+      "loss": 0.3431,
+      "step": 303
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.87375415282392e-07,
+      "loss": 0.2754,
+      "step": 304
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8733388704318935e-07,
+      "loss": 0.314,
+      "step": 305
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.872923588039867e-07,
+      "loss": 0.2934,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.87250830564784e-07,
+      "loss": 0.3195,
+      "step": 307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.872093023255814e-07,
+      "loss": 0.2498,
+      "step": 308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.871677740863787e-07,
+      "loss": 0.3968,
+      "step": 309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.871262458471761e-07,
+      "loss": 0.2944,
+      "step": 310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.870847176079734e-07,
+      "loss": 0.2453,
+      "step": 311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8704318936877076e-07,
+      "loss": 0.2853,
+      "step": 312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.870016611295681e-07,
+      "loss": 0.2557,
+      "step": 313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8696013289036543e-07,
+      "loss": 0.2968,
+      "step": 314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8691860465116277e-07,
+      "loss": 0.3351,
+      "step": 315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.868770764119601e-07,
+      "loss": 0.3196,
+      "step": 316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8683554817275744e-07,
+      "loss": 0.3229,
+      "step": 317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8679401993355483e-07,
+      "loss": 0.3317,
+      "step": 318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8675249169435217e-07,
+      "loss": 0.3008,
+      "step": 319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.867109634551495e-07,
+      "loss": 0.28,
+      "step": 320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8666943521594684e-07,
+      "loss": 0.3425,
+      "step": 321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.866279069767442e-07,
+      "loss": 0.2306,
+      "step": 322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.865863787375415e-07,
+      "loss": 0.2892,
+      "step": 323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8654485049833885e-07,
+      "loss": 0.2939,
+      "step": 324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.865033222591362e-07,
+      "loss": 0.324,
+      "step": 325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.864617940199336e-07,
+      "loss": 0.3196,
+      "step": 326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8642026578073086e-07,
+      "loss": 0.2932,
+      "step": 327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8637873754152825e-07,
+      "loss": 0.2522,
+      "step": 328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.863372093023256e-07,
+      "loss": 0.2872,
+      "step": 329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8629568106312287e-07,
+      "loss": 0.3635,
+      "step": 330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8625415282392026e-07,
+      "loss": 0.3668,
+      "step": 331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.862126245847176e-07,
+      "loss": 0.2328,
+      "step": 332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8617109634551494e-07,
+      "loss": 0.2919,
+      "step": 333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8612956810631227e-07,
+      "loss": 0.3037,
+      "step": 334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.860880398671096e-07,
+      "loss": 0.2902,
+      "step": 335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8604651162790695e-07,
+      "loss": 0.2852,
+      "step": 336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.860049833887043e-07,
+      "loss": 0.3301,
+      "step": 337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.859634551495016e-07,
+      "loss": 0.2737,
+      "step": 338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.85921926910299e-07,
+      "loss": 0.2577,
+      "step": 339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8588039867109635e-07,
+      "loss": 0.2861,
+      "step": 340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.858388704318937e-07,
+      "loss": 0.3079,
+      "step": 341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.85797342192691e-07,
+      "loss": 0.3228,
+      "step": 342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.8575581395348836e-07,
+      "loss": 0.2506,
+      "step": 343
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.857142857142857e-07,
+      "loss": 0.3304,
+      "step": 344
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8567275747508303e-07,
+      "loss": 0.2992,
+      "step": 345
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8563122923588037e-07,
+      "loss": 0.3069,
+      "step": 346
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8558970099667776e-07,
+      "loss": 0.2831,
+      "step": 347
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8554817275747504e-07,
+      "loss": 0.3429,
+      "step": 348
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8550664451827243e-07,
+      "loss": 0.3035,
+      "step": 349
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8546511627906977e-07,
+      "loss": 0.3043,
+      "step": 350
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.854235880398671e-07,
+      "loss": 0.3536,
+      "step": 351
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8538205980066444e-07,
+      "loss": 0.2743,
+      "step": 352
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.853405315614618e-07,
+      "loss": 0.2833,
+      "step": 353
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.852990033222591e-07,
+      "loss": 0.285,
+      "step": 354
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8525747508305645e-07,
+      "loss": 0.285,
+      "step": 355
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.852159468438538e-07,
+      "loss": 0.3431,
+      "step": 356
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.851744186046512e-07,
+      "loss": 0.2596,
+      "step": 357
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.851328903654485e-07,
+      "loss": 0.2947,
+      "step": 358
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8509136212624585e-07,
+      "loss": 0.3008,
+      "step": 359
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.850498338870432e-07,
+      "loss": 0.2879,
+      "step": 360
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.850083056478405e-07,
+      "loss": 0.2391,
+      "step": 361
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8496677740863786e-07,
+      "loss": 0.3631,
+      "step": 362
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.849252491694352e-07,
+      "loss": 0.2556,
+      "step": 363
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8488372093023254e-07,
+      "loss": 0.2616,
+      "step": 364
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.848421926910299e-07,
+      "loss": 0.2736,
+      "step": 365
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.848006644518272e-07,
+      "loss": 0.2666,
+      "step": 366
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.847591362126246e-07,
+      "loss": 0.2965,
+      "step": 367
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8471760797342194e-07,
+      "loss": 0.302,
+      "step": 368
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.846760797342192e-07,
+      "loss": 0.2818,
+      "step": 369
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.846345514950166e-07,
+      "loss": 0.2558,
+      "step": 370
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8459302325581395e-07,
+      "loss": 0.2858,
+      "step": 371
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.845514950166113e-07,
+      "loss": 0.3063,
+      "step": 372
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.845099667774086e-07,
+      "loss": 0.345,
+      "step": 373
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8446843853820596e-07,
+      "loss": 0.2404,
+      "step": 374
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8442691029900335e-07,
+      "loss": 0.3308,
+      "step": 375
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8438538205980063e-07,
+      "loss": 0.3153,
+      "step": 376
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8434385382059797e-07,
+      "loss": 0.3179,
+      "step": 377
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.8430232558139536e-07,
+      "loss": 0.2133,
+      "step": 378
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.842607973421927e-07,
+      "loss": 0.3011,
+      "step": 379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8421926910299003e-07,
+      "loss": 0.2666,
+      "step": 380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8417774086378737e-07,
+      "loss": 0.293,
+      "step": 381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.841362126245847e-07,
+      "loss": 0.2732,
+      "step": 382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8409468438538204e-07,
+      "loss": 0.3547,
+      "step": 383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.840531561461794e-07,
+      "loss": 0.2965,
+      "step": 384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.840116279069767e-07,
+      "loss": 0.2735,
+      "step": 385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.839700996677741e-07,
+      "loss": 0.263,
+      "step": 386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.839285714285714e-07,
+      "loss": 0.2416,
+      "step": 387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.838870431893688e-07,
+      "loss": 0.2697,
+      "step": 388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.838455149501661e-07,
+      "loss": 0.2912,
+      "step": 389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8380398671096345e-07,
+      "loss": 0.2551,
+      "step": 390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.837624584717608e-07,
+      "loss": 0.2752,
+      "step": 391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.837209302325581e-07,
+      "loss": 0.2787,
+      "step": 392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8367940199335546e-07,
+      "loss": 0.2942,
+      "step": 393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.836378737541528e-07,
+      "loss": 0.2392,
+      "step": 394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8359634551495013e-07,
+      "loss": 0.3273,
+      "step": 395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.835548172757475e-07,
+      "loss": 0.295,
+      "step": 396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.835132890365448e-07,
+      "loss": 0.2702,
+      "step": 397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.834717607973422e-07,
+      "loss": 0.3701,
+      "step": 398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8343023255813953e-07,
+      "loss": 0.2603,
+      "step": 399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8338870431893687e-07,
+      "loss": 0.3321,
+      "step": 400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.833471760797342e-07,
+      "loss": 0.253,
+      "step": 401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8330564784053155e-07,
+      "loss": 0.3558,
+      "step": 402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.832641196013289e-07,
+      "loss": 0.2921,
+      "step": 403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.832225913621262e-07,
+      "loss": 0.3759,
+      "step": 404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8318106312292356e-07,
+      "loss": 0.3135,
+      "step": 405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8313953488372095e-07,
+      "loss": 0.2717,
+      "step": 406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.830980066445183e-07,
+      "loss": 0.2444,
+      "step": 407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8305647840531557e-07,
+      "loss": 0.2493,
+      "step": 408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8301495016611296e-07,
+      "loss": 0.3026,
+      "step": 409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.829734219269103e-07,
+      "loss": 0.2835,
+      "step": 410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8293189368770763e-07,
+      "loss": 0.2663,
+      "step": 411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.8289036544850497e-07,
+      "loss": 0.3268,
+      "step": 412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.828488372093023e-07,
+      "loss": 0.2481,
+      "step": 413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.828073089700997e-07,
+      "loss": 0.2836,
+      "step": 414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.82765780730897e-07,
+      "loss": 0.3391,
+      "step": 415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.827242524916943e-07,
+      "loss": 0.3363,
+      "step": 416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.826827242524917e-07,
+      "loss": 0.2654,
+      "step": 417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.82641196013289e-07,
+      "loss": 0.3038,
+      "step": 418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.825996677740864e-07,
+      "loss": 0.2791,
+      "step": 419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.825581395348837e-07,
+      "loss": 0.2514,
+      "step": 420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8251661129568105e-07,
+      "loss": 0.3063,
+      "step": 421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.824750830564784e-07,
+      "loss": 0.2987,
+      "step": 422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.824335548172757e-07,
+      "loss": 0.2453,
+      "step": 423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8239202657807306e-07,
+      "loss": 0.2933,
+      "step": 424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8235049833887045e-07,
+      "loss": 0.3291,
+      "step": 425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8230897009966773e-07,
+      "loss": 0.287,
+      "step": 426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.822674418604651e-07,
+      "loss": 0.2774,
+      "step": 427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8222591362126246e-07,
+      "loss": 0.2478,
+      "step": 428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.821843853820598e-07,
+      "loss": 0.3022,
+      "step": 429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8214285714285713e-07,
+      "loss": 0.27,
+      "step": 430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8210132890365447e-07,
+      "loss": 0.2515,
+      "step": 431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.820598006644518e-07,
+      "loss": 0.2752,
+      "step": 432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8201827242524914e-07,
+      "loss": 0.289,
+      "step": 433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.819767441860465e-07,
+      "loss": 0.3714,
+      "step": 434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8193521594684387e-07,
+      "loss": 0.2671,
+      "step": 435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8189368770764115e-07,
+      "loss": 0.2835,
+      "step": 436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8185215946843854e-07,
+      "loss": 0.2607,
+      "step": 437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.818106312292359e-07,
+      "loss": 0.301,
+      "step": 438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.817691029900332e-07,
+      "loss": 0.2606,
+      "step": 439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8172757475083055e-07,
+      "loss": 0.2355,
+      "step": 440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.816860465116279e-07,
+      "loss": 0.2845,
+      "step": 441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8164451827242523e-07,
+      "loss": 0.3075,
+      "step": 442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8160299003322257e-07,
+      "loss": 0.3205,
+      "step": 443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.815614617940199e-07,
+      "loss": 0.3471,
+      "step": 444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.815199335548173e-07,
+      "loss": 0.2646,
+      "step": 445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8147840531561463e-07,
+      "loss": 0.367,
+      "step": 446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.814368770764119e-07,
+      "loss": 0.2666,
+      "step": 447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.813953488372093e-07,
+      "loss": 0.2859,
+      "step": 448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.8135382059800664e-07,
+      "loss": 0.3173,
+      "step": 449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.81312292358804e-07,
+      "loss": 0.3417,
+      "step": 450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.812707641196013e-07,
+      "loss": 0.3355,
+      "step": 451
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8122923588039865e-07,
+      "loss": 0.3129,
+      "step": 452
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8118770764119604e-07,
+      "loss": 0.3618,
+      "step": 453
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.811461794019933e-07,
+      "loss": 0.3719,
+      "step": 454
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8110465116279066e-07,
+      "loss": 0.2469,
+      "step": 455
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8106312292358805e-07,
+      "loss": 0.3209,
+      "step": 456
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8102159468438533e-07,
+      "loss": 0.3173,
+      "step": 457
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.809800664451827e-07,
+      "loss": 0.2879,
+      "step": 458
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8093853820598006e-07,
+      "loss": 0.2552,
+      "step": 459
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.808970099667774e-07,
+      "loss": 0.3406,
+      "step": 460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8085548172757473e-07,
+      "loss": 0.3165,
+      "step": 461
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8081395348837207e-07,
+      "loss": 0.3136,
+      "step": 462
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.807724252491694e-07,
+      "loss": 0.3344,
+      "step": 463
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8073089700996674e-07,
+      "loss": 0.3162,
+      "step": 464
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.806893687707641e-07,
+      "loss": 0.2898,
+      "step": 465
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8064784053156147e-07,
+      "loss": 0.3082,
+      "step": 466
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.806063122923588e-07,
+      "loss": 0.3616,
+      "step": 467
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8056478405315614e-07,
+      "loss": 0.3188,
+      "step": 468
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.805232558139535e-07,
+      "loss": 0.2798,
+      "step": 469
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.804817275747508e-07,
+      "loss": 0.3203,
+      "step": 470
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8044019933554815e-07,
+      "loss": 0.3163,
+      "step": 471
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.803986710963455e-07,
+      "loss": 0.2991,
+      "step": 472
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8035714285714283e-07,
+      "loss": 0.3339,
+      "step": 473
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.803156146179402e-07,
+      "loss": 0.2809,
+      "step": 474
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.802740863787375e-07,
+      "loss": 0.2846,
+      "step": 475
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.802325581395349e-07,
+      "loss": 0.2721,
+      "step": 476
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8019102990033223e-07,
+      "loss": 0.3204,
+      "step": 477
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8014950166112956e-07,
+      "loss": 0.294,
+      "step": 478
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.801079734219269e-07,
+      "loss": 0.3509,
+      "step": 479
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.8006644518272424e-07,
+      "loss": 0.3514,
+      "step": 480
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.800249169435216e-07,
+      "loss": 0.2639,
+      "step": 481
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.799833887043189e-07,
+      "loss": 0.2475,
+      "step": 482
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.7994186046511625e-07,
+      "loss": 0.2704,
+      "step": 483
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.7990033222591364e-07,
+      "loss": 0.2494,
+      "step": 484
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.79858803986711e-07,
+      "loss": 0.2967,
+      "step": 485
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.798172757475083e-07,
+      "loss": 0.2507,
+      "step": 486
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.7977574750830565e-07,
+      "loss": 0.3355,
+      "step": 487
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.79734219269103e-07,
+      "loss": 0.2561,
+      "step": 488
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.796926910299003e-07,
+      "loss": 0.3196,
+      "step": 489
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7965116279069766e-07,
+      "loss": 0.29,
+      "step": 490
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.79609634551495e-07,
+      "loss": 0.2361,
+      "step": 491
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.795681063122924e-07,
+      "loss": 0.291,
+      "step": 492
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7952657807308967e-07,
+      "loss": 0.2937,
+      "step": 493
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.79485049833887e-07,
+      "loss": 0.2931,
+      "step": 494
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.794435215946844e-07,
+      "loss": 0.3026,
+      "step": 495
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.794019933554817e-07,
+      "loss": 0.3126,
+      "step": 496
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7936046511627907e-07,
+      "loss": 0.3916,
+      "step": 497
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.793189368770764e-07,
+      "loss": 0.3044,
+      "step": 498
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7927740863787374e-07,
+      "loss": 0.2841,
+      "step": 499
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.792358803986711e-07,
+      "loss": 0.2372,
+      "step": 500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.791943521594684e-07,
+      "loss": 0.2951,
+      "step": 501
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7915282392026575e-07,
+      "loss": 0.2666,
+      "step": 502
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.791112956810631e-07,
+      "loss": 0.2722,
+      "step": 503
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7906976744186043e-07,
+      "loss": 0.2714,
+      "step": 504
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.790282392026578e-07,
+      "loss": 0.346,
+      "step": 505
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7898671096345515e-07,
+      "loss": 0.2756,
+      "step": 506
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.789451827242525e-07,
+      "loss": 0.2861,
+      "step": 507
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7890365448504983e-07,
+      "loss": 0.334,
+      "step": 508
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7886212624584716e-07,
+      "loss": 0.2501,
+      "step": 509
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.788205980066445e-07,
+      "loss": 0.2749,
+      "step": 510
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7877906976744184e-07,
+      "loss": 0.3093,
+      "step": 511
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.787375415282392e-07,
+      "loss": 0.3085,
+      "step": 512
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7869601328903656e-07,
+      "loss": 0.2408,
+      "step": 513
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7865448504983385e-07,
+      "loss": 0.29,
+      "step": 514
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7861295681063124e-07,
+      "loss": 0.2864,
+      "step": 515
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.785714285714286e-07,
+      "loss": 0.3071,
+      "step": 516
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.785299003322259e-07,
+      "loss": 0.3224,
+      "step": 517
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7848837209302325e-07,
+      "loss": 0.2701,
+      "step": 518
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.784468438538206e-07,
+      "loss": 0.3207,
+      "step": 519
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.784053156146179e-07,
+      "loss": 0.3472,
+      "step": 520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7836378737541526e-07,
+      "loss": 0.4023,
+      "step": 521
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.783222591362126e-07,
+      "loss": 0.2681,
+      "step": 522
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.7828073089701e-07,
+      "loss": 0.2849,
+      "step": 523
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7823920265780727e-07,
+      "loss": 0.3084,
+      "step": 524
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7819767441860466e-07,
+      "loss": 0.2395,
+      "step": 525
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.78156146179402e-07,
+      "loss": 0.2684,
+      "step": 526
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7811461794019933e-07,
+      "loss": 0.2844,
+      "step": 527
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7807308970099667e-07,
+      "loss": 0.3228,
+      "step": 528
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.78031561461794e-07,
+      "loss": 0.2238,
+      "step": 529
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7799003322259134e-07,
+      "loss": 0.3693,
+      "step": 530
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.779485049833887e-07,
+      "loss": 0.2408,
+      "step": 531
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.77906976744186e-07,
+      "loss": 0.2652,
+      "step": 532
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.778654485049834e-07,
+      "loss": 0.3027,
+      "step": 533
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7782392026578074e-07,
+      "loss": 0.2903,
+      "step": 534
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.77782392026578e-07,
+      "loss": 0.3015,
+      "step": 535
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.777408637873754e-07,
+      "loss": 0.3016,
+      "step": 536
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7769933554817275e-07,
+      "loss": 0.3067,
+      "step": 537
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.776578073089701e-07,
+      "loss": 0.347,
+      "step": 538
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.776162790697674e-07,
+      "loss": 0.2775,
+      "step": 539
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7757475083056476e-07,
+      "loss": 0.2749,
+      "step": 540
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7753322259136215e-07,
+      "loss": 0.237,
+      "step": 541
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7749169435215944e-07,
+      "loss": 0.3316,
+      "step": 542
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7745016611295677e-07,
+      "loss": 0.2767,
+      "step": 543
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7740863787375416e-07,
+      "loss": 0.3234,
+      "step": 544
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7736710963455145e-07,
+      "loss": 0.3118,
+      "step": 545
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7732558139534884e-07,
+      "loss": 0.3431,
+      "step": 546
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7728405315614617e-07,
+      "loss": 0.2663,
+      "step": 547
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.772425249169435e-07,
+      "loss": 0.2907,
+      "step": 548
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7720099667774085e-07,
+      "loss": 0.2853,
+      "step": 549
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.771594684385382e-07,
+      "loss": 0.3208,
+      "step": 550
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.771179401993355e-07,
+      "loss": 0.3586,
+      "step": 551
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.770764119601329e-07,
+      "loss": 0.2759,
+      "step": 552
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.770348837209302e-07,
+      "loss": 0.2526,
+      "step": 553
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.769933554817276e-07,
+      "loss": 0.3609,
+      "step": 554
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.769518272425249e-07,
+      "loss": 0.2744,
+      "step": 555
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7691029900332226e-07,
+      "loss": 0.2894,
+      "step": 556
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.768687707641196e-07,
+      "loss": 0.2931,
+      "step": 557
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7682724252491693e-07,
+      "loss": 0.3077,
+      "step": 558
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.7678571428571427e-07,
+      "loss": 0.2897,
+      "step": 559
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.767441860465116e-07,
+      "loss": 0.2971,
+      "step": 560
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7670265780730894e-07,
+      "loss": 0.2988,
+      "step": 561
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7666112956810633e-07,
+      "loss": 0.2727,
+      "step": 562
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.766196013289036e-07,
+      "loss": 0.2651,
+      "step": 563
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.76578073089701e-07,
+      "loss": 0.2885,
+      "step": 564
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7653654485049834e-07,
+      "loss": 0.3017,
+      "step": 565
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.764950166112957e-07,
+      "loss": 0.2765,
+      "step": 566
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.76453488372093e-07,
+      "loss": 0.2795,
+      "step": 567
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7641196013289035e-07,
+      "loss": 0.3121,
+      "step": 568
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.763704318936877e-07,
+      "loss": 0.2766,
+      "step": 569
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.76328903654485e-07,
+      "loss": 0.2611,
+      "step": 570
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7628737541528236e-07,
+      "loss": 0.3516,
+      "step": 571
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7624584717607975e-07,
+      "loss": 0.2978,
+      "step": 572
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.762043189368771e-07,
+      "loss": 0.3516,
+      "step": 573
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7616279069767437e-07,
+      "loss": 0.2673,
+      "step": 574
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7612126245847176e-07,
+      "loss": 0.371,
+      "step": 575
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.760797342192691e-07,
+      "loss": 0.2794,
+      "step": 576
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7603820598006644e-07,
+      "loss": 0.3036,
+      "step": 577
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7599667774086377e-07,
+      "loss": 0.2782,
+      "step": 578
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.759551495016611e-07,
+      "loss": 0.2764,
+      "step": 579
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.759136212624585e-07,
+      "loss": 0.353,
+      "step": 580
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.758720930232558e-07,
+      "loss": 0.2385,
+      "step": 581
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.758305647840531e-07,
+      "loss": 0.2605,
+      "step": 582
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.757890365448505e-07,
+      "loss": 0.3093,
+      "step": 583
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.757475083056478e-07,
+      "loss": 0.2556,
+      "step": 584
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.757059800664452e-07,
+      "loss": 0.3009,
+      "step": 585
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.756644518272425e-07,
+      "loss": 0.2383,
+      "step": 586
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7562292358803986e-07,
+      "loss": 0.3432,
+      "step": 587
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.755813953488372e-07,
+      "loss": 0.3489,
+      "step": 588
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7553986710963453e-07,
+      "loss": 0.345,
+      "step": 589
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7549833887043187e-07,
+      "loss": 0.3057,
+      "step": 590
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.754568106312292e-07,
+      "loss": 0.2966,
+      "step": 591
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7541528239202654e-07,
+      "loss": 0.3502,
+      "step": 592
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7537375415282393e-07,
+      "loss": 0.3058,
+      "step": 593
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.7533222591362127e-07,
+      "loss": 0.3124,
+      "step": 594
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.752906976744186e-07,
+      "loss": 0.305,
+      "step": 595
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7524916943521594e-07,
+      "loss": 0.2824,
+      "step": 596
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.752076411960133e-07,
+      "loss": 0.244,
+      "step": 597
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.751661129568106e-07,
+      "loss": 0.2889,
+      "step": 598
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7512458471760795e-07,
+      "loss": 0.294,
+      "step": 599
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.750830564784053e-07,
+      "loss": 0.2946,
+      "step": 600
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.750415282392027e-07,
+      "loss": 0.2702,
+      "step": 601
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7499999999999996e-07,
+      "loss": 0.2239,
+      "step": 602
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7495847176079735e-07,
+      "loss": 0.2324,
+      "step": 603
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.749169435215947e-07,
+      "loss": 0.2838,
+      "step": 604
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.74875415282392e-07,
+      "loss": 0.2898,
+      "step": 605
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7483388704318936e-07,
+      "loss": 0.1951,
+      "step": 606
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.747923588039867e-07,
+      "loss": 0.305,
+      "step": 607
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7475083056478403e-07,
+      "loss": 0.2373,
+      "step": 608
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7470930232558137e-07,
+      "loss": 0.3052,
+      "step": 609
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.746677740863787e-07,
+      "loss": 0.3189,
+      "step": 610
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.746262458471761e-07,
+      "loss": 0.2993,
+      "step": 611
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.745847176079734e-07,
+      "loss": 0.3741,
+      "step": 612
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.745431893687707e-07,
+      "loss": 0.2416,
+      "step": 613
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.745016611295681e-07,
+      "loss": 0.3263,
+      "step": 614
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7446013289036545e-07,
+      "loss": 0.2666,
+      "step": 615
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.744186046511628e-07,
+      "loss": 0.2942,
+      "step": 616
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.743770764119601e-07,
+      "loss": 0.3007,
+      "step": 617
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7433554817275746e-07,
+      "loss": 0.3045,
+      "step": 618
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7429401993355485e-07,
+      "loss": 0.3367,
+      "step": 619
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7425249169435213e-07,
+      "loss": 0.2945,
+      "step": 620
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7421096345514947e-07,
+      "loss": 0.2842,
+      "step": 621
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7416943521594686e-07,
+      "loss": 0.2673,
+      "step": 622
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7412790697674414e-07,
+      "loss": 0.3066,
+      "step": 623
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7408637873754153e-07,
+      "loss": 0.2766,
+      "step": 624
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7404485049833887e-07,
+      "loss": 0.2845,
+      "step": 625
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.740033222591362e-07,
+      "loss": 0.2424,
+      "step": 626
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7396179401993354e-07,
+      "loss": 0.2696,
+      "step": 627
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.739202657807309e-07,
+      "loss": 0.3219,
+      "step": 628
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.738787375415282e-07,
+      "loss": 0.3387,
+      "step": 629
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.7383720930232555e-07,
+      "loss": 0.3149,
+      "step": 630
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.737956810631229e-07,
+      "loss": 0.3312,
+      "step": 631
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.737541528239203e-07,
+      "loss": 0.2553,
+      "step": 632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.737126245847176e-07,
+      "loss": 0.3544,
+      "step": 633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7367109634551495e-07,
+      "loss": 0.2868,
+      "step": 634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.736295681063123e-07,
+      "loss": 0.2896,
+      "step": 635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.735880398671096e-07,
+      "loss": 0.2884,
+      "step": 636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7354651162790696e-07,
+      "loss": 0.3274,
+      "step": 637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.735049833887043e-07,
+      "loss": 0.3502,
+      "step": 638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7346345514950163e-07,
+      "loss": 0.2653,
+      "step": 639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.73421926910299e-07,
+      "loss": 0.3066,
+      "step": 640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.733803986710963e-07,
+      "loss": 0.2865,
+      "step": 641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.733388704318937e-07,
+      "loss": 0.289,
+      "step": 642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7329734219269103e-07,
+      "loss": 0.283,
+      "step": 643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7325581395348837e-07,
+      "loss": 0.2823,
+      "step": 644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.732142857142857e-07,
+      "loss": 0.2892,
+      "step": 645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7317275747508304e-07,
+      "loss": 0.3023,
+      "step": 646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.731312292358804e-07,
+      "loss": 0.3247,
+      "step": 647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.730897009966777e-07,
+      "loss": 0.2581,
+      "step": 648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7304817275747505e-07,
+      "loss": 0.2743,
+      "step": 649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7300664451827244e-07,
+      "loss": 0.2972,
+      "step": 650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7296511627906973e-07,
+      "loss": 0.2958,
+      "step": 651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.729235880398671e-07,
+      "loss": 0.3656,
+      "step": 652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7288205980066445e-07,
+      "loss": 0.3513,
+      "step": 653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.728405315614618e-07,
+      "loss": 0.3257,
+      "step": 654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7279900332225913e-07,
+      "loss": 0.3084,
+      "step": 655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7275747508305647e-07,
+      "loss": 0.3009,
+      "step": 656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.727159468438538e-07,
+      "loss": 0.3201,
+      "step": 657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7267441860465114e-07,
+      "loss": 0.3307,
+      "step": 658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.726328903654485e-07,
+      "loss": 0.2887,
+      "step": 659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.725913621262458e-07,
+      "loss": 0.3088,
+      "step": 660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.725498338870432e-07,
+      "loss": 0.3081,
+      "step": 661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.725083056478405e-07,
+      "loss": 0.2924,
+      "step": 662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.724667774086379e-07,
+      "loss": 0.288,
+      "step": 663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.724252491694352e-07,
+      "loss": 0.3406,
+      "step": 664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7238372093023255e-07,
+      "loss": 0.353,
+      "step": 665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.723421926910299e-07,
+      "loss": 0.2973,
+      "step": 666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.723006644518272e-07,
+      "loss": 0.303,
+      "step": 667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.7225913621262456e-07,
+      "loss": 0.2381,
+      "step": 668
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.722176079734219e-07,
+      "loss": 0.3055,
+      "step": 669
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7217607973421923e-07,
+      "loss": 0.2819,
+      "step": 670
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.721345514950166e-07,
+      "loss": 0.2645,
+      "step": 671
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.720930232558139e-07,
+      "loss": 0.2641,
+      "step": 672
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.720514950166113e-07,
+      "loss": 0.3061,
+      "step": 673
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7200996677740863e-07,
+      "loss": 0.3289,
+      "step": 674
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7196843853820597e-07,
+      "loss": 0.3333,
+      "step": 675
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.719269102990033e-07,
+      "loss": 0.3469,
+      "step": 676
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7188538205980064e-07,
+      "loss": 0.264,
+      "step": 677
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.71843853820598e-07,
+      "loss": 0.2802,
+      "step": 678
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7180232558139537e-07,
+      "loss": 0.297,
+      "step": 679
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7176079734219265e-07,
+      "loss": 0.2518,
+      "step": 680
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7171926910299004e-07,
+      "loss": 0.3003,
+      "step": 681
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.716777408637874e-07,
+      "loss": 0.273,
+      "step": 682
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.716362126245847e-07,
+      "loss": 0.272,
+      "step": 683
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7159468438538205e-07,
+      "loss": 0.2946,
+      "step": 684
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.715531561461794e-07,
+      "loss": 0.2441,
+      "step": 685
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7151162790697673e-07,
+      "loss": 0.2823,
+      "step": 686
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7147009966777406e-07,
+      "loss": 0.3348,
+      "step": 687
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.714285714285714e-07,
+      "loss": 0.2779,
+      "step": 688
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.713870431893688e-07,
+      "loss": 0.3423,
+      "step": 689
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.713455149501661e-07,
+      "loss": 0.2392,
+      "step": 690
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7130398671096346e-07,
+      "loss": 0.2718,
+      "step": 691
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.712624584717608e-07,
+      "loss": 0.3739,
+      "step": 692
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7122093023255814e-07,
+      "loss": 0.2892,
+      "step": 693
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.711794019933555e-07,
+      "loss": 0.309,
+      "step": 694
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.711378737541528e-07,
+      "loss": 0.2984,
+      "step": 695
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7109634551495015e-07,
+      "loss": 0.3047,
+      "step": 696
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.710548172757475e-07,
+      "loss": 0.2775,
+      "step": 697
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.710132890365448e-07,
+      "loss": 0.2534,
+      "step": 698
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.709717607973422e-07,
+      "loss": 0.2662,
+      "step": 699
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7093023255813955e-07,
+      "loss": 0.2782,
+      "step": 700
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7088870431893683e-07,
+      "loss": 0.3836,
+      "step": 701
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.708471760797342e-07,
+      "loss": 0.322,
+      "step": 702
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.7080564784053156e-07,
+      "loss": 0.2741,
+      "step": 703
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.707641196013289e-07,
+      "loss": 0.313,
+      "step": 704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7072259136212623e-07,
+      "loss": 0.2522,
+      "step": 705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7068106312292357e-07,
+      "loss": 0.2779,
+      "step": 706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.706395348837209e-07,
+      "loss": 0.2571,
+      "step": 707
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7059800664451824e-07,
+      "loss": 0.3014,
+      "step": 708
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.705564784053156e-07,
+      "loss": 0.3326,
+      "step": 709
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7051495016611297e-07,
+      "loss": 0.29,
+      "step": 710
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7047342192691025e-07,
+      "loss": 0.323,
+      "step": 711
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7043189368770764e-07,
+      "loss": 0.3352,
+      "step": 712
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.70390365448505e-07,
+      "loss": 0.3197,
+      "step": 713
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.703488372093023e-07,
+      "loss": 0.3301,
+      "step": 714
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7030730897009965e-07,
+      "loss": 0.3109,
+      "step": 715
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.70265780730897e-07,
+      "loss": 0.2747,
+      "step": 716
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7022425249169433e-07,
+      "loss": 0.322,
+      "step": 717
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7018272425249166e-07,
+      "loss": 0.3059,
+      "step": 718
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.70141196013289e-07,
+      "loss": 0.3245,
+      "step": 719
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.700996677740864e-07,
+      "loss": 0.3012,
+      "step": 720
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7005813953488373e-07,
+      "loss": 0.3121,
+      "step": 721
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.7001661129568106e-07,
+      "loss": 0.302,
+      "step": 722
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.699750830564784e-07,
+      "loss": 0.299,
+      "step": 723
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6993355481727574e-07,
+      "loss": 0.302,
+      "step": 724
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.698920265780731e-07,
+      "loss": 0.3327,
+      "step": 725
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.698504983388704e-07,
+      "loss": 0.2812,
+      "step": 726
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6980897009966775e-07,
+      "loss": 0.3201,
+      "step": 727
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6976744186046514e-07,
+      "loss": 0.3074,
+      "step": 728
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.697259136212624e-07,
+      "loss": 0.2736,
+      "step": 729
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.696843853820598e-07,
+      "loss": 0.2674,
+      "step": 730
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6964285714285715e-07,
+      "loss": 0.2648,
+      "step": 731
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6960132890365443e-07,
+      "loss": 0.3295,
+      "step": 732
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.695598006644518e-07,
+      "loss": 0.2927,
+      "step": 733
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6951827242524916e-07,
+      "loss": 0.2453,
+      "step": 734
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.694767441860465e-07,
+      "loss": 0.2941,
+      "step": 735
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6943521594684383e-07,
+      "loss": 0.276,
+      "step": 736
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6939368770764117e-07,
+      "loss": 0.3112,
+      "step": 737
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6935215946843856e-07,
+      "loss": 0.2614,
+      "step": 738
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.6931063122923584e-07,
+      "loss": 0.3445,
+      "step": 739
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.692691029900332e-07,
+      "loss": 0.2995,
+      "step": 740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6922757475083057e-07,
+      "loss": 0.2573,
+      "step": 741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.691860465116279e-07,
+      "loss": 0.3272,
+      "step": 742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6914451827242524e-07,
+      "loss": 0.2674,
+      "step": 743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.691029900332226e-07,
+      "loss": 0.3652,
+      "step": 744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.690614617940199e-07,
+      "loss": 0.2915,
+      "step": 745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.690199335548173e-07,
+      "loss": 0.3159,
+      "step": 746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.689784053156146e-07,
+      "loss": 0.3354,
+      "step": 747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.689368770764119e-07,
+      "loss": 0.2968,
+      "step": 748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.688953488372093e-07,
+      "loss": 0.2529,
+      "step": 749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.688538205980066e-07,
+      "loss": 0.2726,
+      "step": 750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.68812292358804e-07,
+      "loss": 0.295,
+      "step": 751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.687707641196013e-07,
+      "loss": 0.3391,
+      "step": 752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6872923588039866e-07,
+      "loss": 0.3831,
+      "step": 753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.68687707641196e-07,
+      "loss": 0.2189,
+      "step": 754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6864617940199334e-07,
+      "loss": 0.3344,
+      "step": 755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6860465116279067e-07,
+      "loss": 0.343,
+      "step": 756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.68563122923588e-07,
+      "loss": 0.2863,
+      "step": 757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6852159468438535e-07,
+      "loss": 0.3516,
+      "step": 758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6848006644518274e-07,
+      "loss": 0.2867,
+      "step": 759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6843853820598007e-07,
+      "loss": 0.2883,
+      "step": 760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.683970099667774e-07,
+      "loss": 0.2668,
+      "step": 761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6835548172757475e-07,
+      "loss": 0.3757,
+      "step": 762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.683139534883721e-07,
+      "loss": 0.3155,
+      "step": 763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.682724252491694e-07,
+      "loss": 0.2636,
+      "step": 764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6823089700996676e-07,
+      "loss": 0.2933,
+      "step": 765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.681893687707641e-07,
+      "loss": 0.2886,
+      "step": 766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.681478405315615e-07,
+      "loss": 0.2406,
+      "step": 767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6810631229235877e-07,
+      "loss": 0.2765,
+      "step": 768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6806478405315616e-07,
+      "loss": 0.3197,
+      "step": 769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.680232558139535e-07,
+      "loss": 0.2229,
+      "step": 770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.679817275747508e-07,
+      "loss": 0.3021,
+      "step": 771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6794019933554817e-07,
+      "loss": 0.2938,
+      "step": 772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.678986710963455e-07,
+      "loss": 0.2851,
+      "step": 773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.6785714285714284e-07,
+      "loss": 0.3419,
+      "step": 774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.678156146179402e-07,
+      "loss": 0.3047,
+      "step": 775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.677740863787375e-07,
+      "loss": 0.2673,
+      "step": 776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.677325581395349e-07,
+      "loss": 0.2694,
+      "step": 777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.676910299003322e-07,
+      "loss": 0.3042,
+      "step": 778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.676495016611295e-07,
+      "loss": 0.2956,
+      "step": 779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.676079734219269e-07,
+      "loss": 0.2301,
+      "step": 780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6756644518272425e-07,
+      "loss": 0.2749,
+      "step": 781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.675249169435216e-07,
+      "loss": 0.3044,
+      "step": 782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.674833887043189e-07,
+      "loss": 0.2836,
+      "step": 783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6744186046511626e-07,
+      "loss": 0.2509,
+      "step": 784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.674003322259136e-07,
+      "loss": 0.2468,
+      "step": 785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6735880398671094e-07,
+      "loss": 0.302,
+      "step": 786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6731727574750827e-07,
+      "loss": 0.2612,
+      "step": 787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6727574750830566e-07,
+      "loss": 0.3254,
+      "step": 788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6723421926910295e-07,
+      "loss": 0.3068,
+      "step": 789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6719269102990034e-07,
+      "loss": 0.2975,
+      "step": 790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6715116279069767e-07,
+      "loss": 0.2768,
+      "step": 791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.67109634551495e-07,
+      "loss": 0.218,
+      "step": 792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6706810631229235e-07,
+      "loss": 0.276,
+      "step": 793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.670265780730897e-07,
+      "loss": 0.2701,
+      "step": 794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.66985049833887e-07,
+      "loss": 0.2679,
+      "step": 795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6694352159468436e-07,
+      "loss": 0.2667,
+      "step": 796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.669019933554817e-07,
+      "loss": 0.3148,
+      "step": 797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.668604651162791e-07,
+      "loss": 0.3024,
+      "step": 798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6681893687707637e-07,
+      "loss": 0.289,
+      "step": 799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6677740863787376e-07,
+      "loss": 0.345,
+      "step": 800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.667358803986711e-07,
+      "loss": 0.3322,
+      "step": 801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6669435215946843e-07,
+      "loss": 0.2722,
+      "step": 802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6665282392026577e-07,
+      "loss": 0.2717,
+      "step": 803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.666112956810631e-07,
+      "loss": 0.3101,
+      "step": 804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6656976744186044e-07,
+      "loss": 0.2969,
+      "step": 805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.665282392026578e-07,
+      "loss": 0.3199,
+      "step": 806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.664867109634551e-07,
+      "loss": 0.2603,
+      "step": 807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.664451827242525e-07,
+      "loss": 0.3166,
+      "step": 808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6640365448504984e-07,
+      "loss": 0.2962,
+      "step": 809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.663621262458472e-07,
+      "loss": 0.281,
+      "step": 810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.663205980066445e-07,
+      "loss": 0.3122,
+      "step": 811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.6627906976744185e-07,
+      "loss": 0.2687,
+      "step": 812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.662375415282392e-07,
+      "loss": 0.3122,
+      "step": 813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.661960132890365e-07,
+      "loss": 0.2751,
+      "step": 814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6615448504983386e-07,
+      "loss": 0.2836,
+      "step": 815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6611295681063125e-07,
+      "loss": 0.2433,
+      "step": 816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6607142857142853e-07,
+      "loss": 0.2889,
+      "step": 817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.660299003322259e-07,
+      "loss": 0.2806,
+      "step": 818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6598837209302326e-07,
+      "loss": 0.3223,
+      "step": 819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.659468438538206e-07,
+      "loss": 0.2797,
+      "step": 820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6590531561461793e-07,
+      "loss": 0.2781,
+      "step": 821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6586378737541527e-07,
+      "loss": 0.2556,
+      "step": 822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.658222591362126e-07,
+      "loss": 0.3169,
+      "step": 823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6578073089700995e-07,
+      "loss": 0.4656,
+      "step": 824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.657392026578073e-07,
+      "loss": 0.2827,
+      "step": 825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.656976744186046e-07,
+      "loss": 0.3217,
+      "step": 826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.65656146179402e-07,
+      "loss": 0.2715,
+      "step": 827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.656146179401993e-07,
+      "loss": 0.2809,
+      "step": 828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.655730897009967e-07,
+      "loss": 0.2935,
+      "step": 829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.65531561461794e-07,
+      "loss": 0.2385,
+      "step": 830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6549003322259136e-07,
+      "loss": 0.3507,
+      "step": 831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.654485049833887e-07,
+      "loss": 0.339,
+      "step": 832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6540697674418603e-07,
+      "loss": 0.2897,
+      "step": 833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6536544850498337e-07,
+      "loss": 0.2765,
+      "step": 834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.653239202657807e-07,
+      "loss": 0.3059,
+      "step": 835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6528239202657804e-07,
+      "loss": 0.2925,
+      "step": 836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6524086378737543e-07,
+      "loss": 0.3981,
+      "step": 837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.651993355481727e-07,
+      "loss": 0.3214,
+      "step": 838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.651578073089701e-07,
+      "loss": 0.2861,
+      "step": 839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6511627906976744e-07,
+      "loss": 0.2583,
+      "step": 840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.650747508305648e-07,
+      "loss": 0.3089,
+      "step": 841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.650332225913621e-07,
+      "loss": 0.361,
+      "step": 842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6499169435215945e-07,
+      "loss": 0.2993,
+      "step": 843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.649501661129568e-07,
+      "loss": 0.3471,
+      "step": 844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.649086378737541e-07,
+      "loss": 0.3335,
+      "step": 845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6486710963455146e-07,
+      "loss": 0.2051,
+      "step": 846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6482558139534885e-07,
+      "loss": 0.2847,
+      "step": 847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.647840531561462e-07,
+      "loss": 0.2727,
+      "step": 848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.647425249169435e-07,
+      "loss": 0.2904,
+      "step": 849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6470099667774086e-07,
+      "loss": 0.3157,
+      "step": 850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.646594684385382e-07,
+      "loss": 0.3641,
+      "step": 851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6461794019933553e-07,
+      "loss": 0.257,
+      "step": 852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6457641196013287e-07,
+      "loss": 0.276,
+      "step": 853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.645348837209302e-07,
+      "loss": 0.2517,
+      "step": 854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.644933554817276e-07,
+      "loss": 0.2959,
+      "step": 855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.644518272425249e-07,
+      "loss": 0.313,
+      "step": 856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6441029900332227e-07,
+      "loss": 0.2562,
+      "step": 857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.643687707641196e-07,
+      "loss": 0.266,
+      "step": 858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.643272425249169e-07,
+      "loss": 0.2739,
+      "step": 859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.642857142857143e-07,
+      "loss": 0.3114,
+      "step": 860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.642441860465116e-07,
+      "loss": 0.3433,
+      "step": 861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6420265780730895e-07,
+      "loss": 0.311,
+      "step": 862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.641611295681063e-07,
+      "loss": 0.2863,
+      "step": 863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6411960132890363e-07,
+      "loss": 0.3156,
+      "step": 864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.64078073089701e-07,
+      "loss": 0.2558,
+      "step": 865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.640365448504983e-07,
+      "loss": 0.3221,
+      "step": 866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6399501661129564e-07,
+      "loss": 0.3782,
+      "step": 867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6395348837209303e-07,
+      "loss": 0.3126,
+      "step": 868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6391196013289037e-07,
+      "loss": 0.3066,
+      "step": 869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.638704318936877e-07,
+      "loss": 0.3397,
+      "step": 870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6382890365448504e-07,
+      "loss": 0.3503,
+      "step": 871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.637873754152824e-07,
+      "loss": 0.2651,
+      "step": 872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.637458471760797e-07,
+      "loss": 0.317,
+      "step": 873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6370431893687705e-07,
+      "loss": 0.2749,
+      "step": 874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.636627906976744e-07,
+      "loss": 0.2487,
+      "step": 875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.636212624584718e-07,
+      "loss": 0.3324,
+      "step": 876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6357973421926906e-07,
+      "loss": 0.311,
+      "step": 877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6353820598006645e-07,
+      "loss": 0.3005,
+      "step": 878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.634966777408638e-07,
+      "loss": 0.255,
+      "step": 879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.634551495016611e-07,
+      "loss": 0.298,
+      "step": 880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6341362126245846e-07,
+      "loss": 0.308,
+      "step": 881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.633720930232558e-07,
+      "loss": 0.2534,
+      "step": 882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6333056478405313e-07,
+      "loss": 0.2585,
+      "step": 883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6328903654485047e-07,
+      "loss": 0.237,
+      "step": 884
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.632475083056478e-07,
+      "loss": 0.2859,
+      "step": 885
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.632059800664452e-07,
+      "loss": 0.3085,
+      "step": 886
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6316445182724253e-07,
+      "loss": 0.3599,
+      "step": 887
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6312292358803987e-07,
+      "loss": 0.2859,
+      "step": 888
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.630813953488372e-07,
+      "loss": 0.2291,
+      "step": 889
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6303986710963454e-07,
+      "loss": 0.2818,
+      "step": 890
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.629983388704319e-07,
+      "loss": 0.2709,
+      "step": 891
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.629568106312292e-07,
+      "loss": 0.3149,
+      "step": 892
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6291528239202655e-07,
+      "loss": 0.321,
+      "step": 893
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6287375415282394e-07,
+      "loss": 0.3129,
+      "step": 894
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6283222591362123e-07,
+      "loss": 0.3292,
+      "step": 895
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.627906976744186e-07,
+      "loss": 0.3037,
+      "step": 896
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6274916943521595e-07,
+      "loss": 0.297,
+      "step": 897
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6270764119601324e-07,
+      "loss": 0.324,
+      "step": 898
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6266611295681063e-07,
+      "loss": 0.3077,
+      "step": 899
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6262458471760796e-07,
+      "loss": 0.2617,
+      "step": 900
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.625830564784053e-07,
+      "loss": 0.3418,
+      "step": 901
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6254152823920264e-07,
+      "loss": 0.3558,
+      "step": 902
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.625e-07,
+      "loss": 0.3214,
+      "step": 903
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6245847176079736e-07,
+      "loss": 0.2891,
+      "step": 904
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6241694352159465e-07,
+      "loss": 0.3061,
+      "step": 905
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.62375415282392e-07,
+      "loss": 0.2592,
+      "step": 906
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.623338870431894e-07,
+      "loss": 0.299,
+      "step": 907
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.622923588039867e-07,
+      "loss": 0.2546,
+      "step": 908
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6225083056478405e-07,
+      "loss": 0.282,
+      "step": 909
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.622093023255814e-07,
+      "loss": 0.2924,
+      "step": 910
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.621677740863787e-07,
+      "loss": 0.2627,
+      "step": 911
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6212624584717606e-07,
+      "loss": 0.2611,
+      "step": 912
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.620847176079734e-07,
+      "loss": 0.2826,
+      "step": 913
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6204318936877073e-07,
+      "loss": 0.34,
+      "step": 914
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.620016611295681e-07,
+      "loss": 0.3204,
+      "step": 915
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.619601328903654e-07,
+      "loss": 0.2594,
+      "step": 916
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.619186046511628e-07,
+      "loss": 0.2704,
+      "step": 917
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6187707641196013e-07,
+      "loss": 0.3013,
+      "step": 918
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6183554817275747e-07,
+      "loss": 0.3253,
+      "step": 919
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.617940199335548e-07,
+      "loss": 0.2649,
+      "step": 920
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.6175249169435214e-07,
+      "loss": 0.2876,
+      "step": 921
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.617109634551495e-07,
+      "loss": 0.2998,
+      "step": 922
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.616694352159468e-07,
+      "loss": 0.3104,
+      "step": 923
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6162790697674415e-07,
+      "loss": 0.2669,
+      "step": 924
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6158637873754154e-07,
+      "loss": 0.2518,
+      "step": 925
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6154485049833883e-07,
+      "loss": 0.2749,
+      "step": 926
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.615033222591362e-07,
+      "loss": 0.2584,
+      "step": 927
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6146179401993355e-07,
+      "loss": 0.256,
+      "step": 928
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.614202657807309e-07,
+      "loss": 0.2671,
+      "step": 929
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6137873754152823e-07,
+      "loss": 0.2824,
+      "step": 930
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6133720930232556e-07,
+      "loss": 0.2595,
+      "step": 931
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.612956810631229e-07,
+      "loss": 0.3015,
+      "step": 932
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6125415282392024e-07,
+      "loss": 0.3004,
+      "step": 933
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.612126245847176e-07,
+      "loss": 0.3097,
+      "step": 934
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6117109634551496e-07,
+      "loss": 0.3048,
+      "step": 935
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.611295681063123e-07,
+      "loss": 0.3155,
+      "step": 936
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.610880398671096e-07,
+      "loss": 0.2799,
+      "step": 937
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.61046511627907e-07,
+      "loss": 0.3879,
+      "step": 938
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.610049833887043e-07,
+      "loss": 0.3723,
+      "step": 939
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6096345514950165e-07,
+      "loss": 0.3684,
+      "step": 940
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.60921926910299e-07,
+      "loss": 0.299,
+      "step": 941
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.608803986710963e-07,
+      "loss": 0.2836,
+      "step": 942
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.608388704318937e-07,
+      "loss": 0.2501,
+      "step": 943
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.60797342192691e-07,
+      "loss": 0.2635,
+      "step": 944
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6075581395348833e-07,
+      "loss": 0.2815,
+      "step": 945
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.607142857142857e-07,
+      "loss": 0.2834,
+      "step": 946
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6067275747508306e-07,
+      "loss": 0.2659,
+      "step": 947
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.606312292358804e-07,
+      "loss": 0.3475,
+      "step": 948
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6058970099667773e-07,
+      "loss": 0.2814,
+      "step": 949
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6054817275747507e-07,
+      "loss": 0.239,
+      "step": 950
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.605066445182724e-07,
+      "loss": 0.3475,
+      "step": 951
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6046511627906974e-07,
+      "loss": 0.2782,
+      "step": 952
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.604235880398671e-07,
+      "loss": 0.263,
+      "step": 953
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6038205980066447e-07,
+      "loss": 0.3763,
+      "step": 954
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6034053156146175e-07,
+      "loss": 0.2401,
+      "step": 955
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.6029900332225914e-07,
+      "loss": 0.3137,
+      "step": 956
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.602574750830565e-07,
+      "loss": 0.2137,
+      "step": 957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.602159468438538e-07,
+      "loss": 0.3475,
+      "step": 958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6017441860465115e-07,
+      "loss": 0.2926,
+      "step": 959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.601328903654485e-07,
+      "loss": 0.296,
+      "step": 960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.600913621262458e-07,
+      "loss": 0.3196,
+      "step": 961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.6004983388704316e-07,
+      "loss": 0.3034,
+      "step": 962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.600083056478405e-07,
+      "loss": 0.2664,
+      "step": 963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.599667774086379e-07,
+      "loss": 0.2243,
+      "step": 964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5992524916943517e-07,
+      "loss": 0.318,
+      "step": 965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5988372093023256e-07,
+      "loss": 0.3328,
+      "step": 966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.598421926910299e-07,
+      "loss": 0.3255,
+      "step": 967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5980066445182724e-07,
+      "loss": 0.275,
+      "step": 968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5975913621262457e-07,
+      "loss": 0.308,
+      "step": 969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.597176079734219e-07,
+      "loss": 0.2782,
+      "step": 970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5967607973421925e-07,
+      "loss": 0.3151,
+      "step": 971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.596345514950166e-07,
+      "loss": 0.3121,
+      "step": 972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.595930232558139e-07,
+      "loss": 0.3002,
+      "step": 973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.595514950166113e-07,
+      "loss": 0.2794,
+      "step": 974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5950996677740865e-07,
+      "loss": 0.2466,
+      "step": 975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.59468438538206e-07,
+      "loss": 0.3215,
+      "step": 976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.594269102990033e-07,
+      "loss": 0.2809,
+      "step": 977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5938538205980066e-07,
+      "loss": 0.2839,
+      "step": 978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.59343853820598e-07,
+      "loss": 0.2544,
+      "step": 979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5930232558139533e-07,
+      "loss": 0.2749,
+      "step": 980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5926079734219267e-07,
+      "loss": 0.3174,
+      "step": 981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5921926910299006e-07,
+      "loss": 0.2953,
+      "step": 982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5917774086378734e-07,
+      "loss": 0.2505,
+      "step": 983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5913621262458473e-07,
+      "loss": 0.2647,
+      "step": 984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5909468438538207e-07,
+      "loss": 0.2829,
+      "step": 985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5905315614617935e-07,
+      "loss": 0.3104,
+      "step": 986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5901162790697674e-07,
+      "loss": 0.2875,
+      "step": 987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.589700996677741e-07,
+      "loss": 0.2517,
+      "step": 988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.589285714285714e-07,
+      "loss": 0.2985,
+      "step": 989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5888704318936875e-07,
+      "loss": 0.2992,
+      "step": 990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.588455149501661e-07,
+      "loss": 0.3305,
+      "step": 991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.588039867109634e-07,
+      "loss": 0.2357,
+      "step": 992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.5876245847176076e-07,
+      "loss": 0.2546,
+      "step": 993
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.587209302325581e-07,
+      "loss": 0.3297,
+      "step": 994
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.586794019933555e-07,
+      "loss": 0.305,
+      "step": 995
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.586378737541528e-07,
+      "loss": 0.3116,
+      "step": 996
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5859634551495016e-07,
+      "loss": 0.2588,
+      "step": 997
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.585548172757475e-07,
+      "loss": 0.311,
+      "step": 998
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5851328903654484e-07,
+      "loss": 0.3413,
+      "step": 999
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5847176079734217e-07,
+      "loss": 0.2722,
+      "step": 1000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.584302325581395e-07,
+      "loss": 0.288,
+      "step": 1001
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5838870431893685e-07,
+      "loss": 0.2884,
+      "step": 1002
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5834717607973424e-07,
+      "loss": 0.2985,
+      "step": 1003
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.583056478405315e-07,
+      "loss": 0.3051,
+      "step": 1004
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.582641196013289e-07,
+      "loss": 0.2462,
+      "step": 1005
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5822259136212625e-07,
+      "loss": 0.2892,
+      "step": 1006
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.581810631229236e-07,
+      "loss": 0.2999,
+      "step": 1007
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.581395348837209e-07,
+      "loss": 0.3072,
+      "step": 1008
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5809800664451826e-07,
+      "loss": 0.3165,
+      "step": 1009
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.580564784053156e-07,
+      "loss": 0.2598,
+      "step": 1010
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5801495016611293e-07,
+      "loss": 0.3303,
+      "step": 1011
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5797342192691027e-07,
+      "loss": 0.3127,
+      "step": 1012
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5793189368770766e-07,
+      "loss": 0.3285,
+      "step": 1013
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.57890365448505e-07,
+      "loss": 0.2548,
+      "step": 1014
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5784883720930233e-07,
+      "loss": 0.2931,
+      "step": 1015
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5780730897009967e-07,
+      "loss": 0.2972,
+      "step": 1016
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.57765780730897e-07,
+      "loss": 0.2677,
+      "step": 1017
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5772425249169434e-07,
+      "loss": 0.2532,
+      "step": 1018
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.576827242524917e-07,
+      "loss": 0.3105,
+      "step": 1019
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.57641196013289e-07,
+      "loss": 0.2771,
+      "step": 1020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.575996677740864e-07,
+      "loss": 0.3134,
+      "step": 1021
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.575581395348837e-07,
+      "loss": 0.2438,
+      "step": 1022
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.575166112956811e-07,
+      "loss": 0.323,
+      "step": 1023
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.574750830564784e-07,
+      "loss": 0.2769,
+      "step": 1024
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.574335548172757e-07,
+      "loss": 0.2485,
+      "step": 1025
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.573920265780731e-07,
+      "loss": 0.3056,
+      "step": 1026
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.573504983388704e-07,
+      "loss": 0.309,
+      "step": 1027
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.5730897009966776e-07,
+      "loss": 0.2722,
+      "step": 1028
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.572674418604651e-07,
+      "loss": 0.3019,
+      "step": 1029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5722591362126243e-07,
+      "loss": 0.2163,
+      "step": 1030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.571843853820598e-07,
+      "loss": 0.2769,
+      "step": 1031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.571428571428571e-07,
+      "loss": 0.2934,
+      "step": 1032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5710132890365445e-07,
+      "loss": 0.3003,
+      "step": 1033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5705980066445184e-07,
+      "loss": 0.2923,
+      "step": 1034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5701827242524917e-07,
+      "loss": 0.3103,
+      "step": 1035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.569767441860465e-07,
+      "loss": 0.2765,
+      "step": 1036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5693521594684385e-07,
+      "loss": 0.3061,
+      "step": 1037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.568936877076412e-07,
+      "loss": 0.2505,
+      "step": 1038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.568521594684385e-07,
+      "loss": 0.3031,
+      "step": 1039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5681063122923586e-07,
+      "loss": 0.2744,
+      "step": 1040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.567691029900332e-07,
+      "loss": 0.2871,
+      "step": 1041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.567275747508306e-07,
+      "loss": 0.2599,
+      "step": 1042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5668604651162787e-07,
+      "loss": 0.279,
+      "step": 1043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5664451827242526e-07,
+      "loss": 0.2733,
+      "step": 1044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.566029900332226e-07,
+      "loss": 0.3247,
+      "step": 1045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5656146179401993e-07,
+      "loss": 0.3199,
+      "step": 1046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5651993355481727e-07,
+      "loss": 0.372,
+      "step": 1047
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.564784053156146e-07,
+      "loss": 0.2782,
+      "step": 1048
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5643687707641194e-07,
+      "loss": 0.2933,
+      "step": 1049
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.563953488372093e-07,
+      "loss": 0.3,
+      "step": 1050
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.563538205980066e-07,
+      "loss": 0.2944,
+      "step": 1051
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.56312292358804e-07,
+      "loss": 0.3034,
+      "step": 1052
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.562707641196013e-07,
+      "loss": 0.333,
+      "step": 1053
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.562292358803987e-07,
+      "loss": 0.276,
+      "step": 1054
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.56187707641196e-07,
+      "loss": 0.2499,
+      "step": 1055
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5614617940199335e-07,
+      "loss": 0.22,
+      "step": 1056
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.561046511627907e-07,
+      "loss": 0.2921,
+      "step": 1057
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.56063122923588e-07,
+      "loss": 0.3392,
+      "step": 1058
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5602159468438536e-07,
+      "loss": 0.3134,
+      "step": 1059
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.559800664451827e-07,
+      "loss": 0.2549,
+      "step": 1060
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5593853820598003e-07,
+      "loss": 0.2752,
+      "step": 1061
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.558970099667774e-07,
+      "loss": 0.2639,
+      "step": 1062
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5585548172757476e-07,
+      "loss": 0.2928,
+      "step": 1063
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5581395348837204e-07,
+      "loss": 0.2989,
+      "step": 1064
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.5577242524916943e-07,
+      "loss": 0.3651,
+      "step": 1065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5573089700996677e-07,
+      "loss": 0.2395,
+      "step": 1066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.556893687707641e-07,
+      "loss": 0.2534,
+      "step": 1067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5564784053156144e-07,
+      "loss": 0.3441,
+      "step": 1068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.556063122923588e-07,
+      "loss": 0.3293,
+      "step": 1069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5556478405315617e-07,
+      "loss": 0.2605,
+      "step": 1070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5552325581395345e-07,
+      "loss": 0.2841,
+      "step": 1071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.554817275747508e-07,
+      "loss": 0.2714,
+      "step": 1072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.554401993355482e-07,
+      "loss": 0.2577,
+      "step": 1073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.553986710963455e-07,
+      "loss": 0.2584,
+      "step": 1074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5535714285714286e-07,
+      "loss": 0.2729,
+      "step": 1075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.553156146179402e-07,
+      "loss": 0.2362,
+      "step": 1076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5527408637873753e-07,
+      "loss": 0.2531,
+      "step": 1077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5523255813953487e-07,
+      "loss": 0.2623,
+      "step": 1078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.551910299003322e-07,
+      "loss": 0.289,
+      "step": 1079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5514950166112954e-07,
+      "loss": 0.2663,
+      "step": 1080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5510797342192693e-07,
+      "loss": 0.2824,
+      "step": 1081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.550664451827242e-07,
+      "loss": 0.3409,
+      "step": 1082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.550249169435216e-07,
+      "loss": 0.308,
+      "step": 1083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5498338870431894e-07,
+      "loss": 0.2904,
+      "step": 1084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.549418604651163e-07,
+      "loss": 0.285,
+      "step": 1085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.549003322259136e-07,
+      "loss": 0.2737,
+      "step": 1086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5485880398671095e-07,
+      "loss": 0.3568,
+      "step": 1087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.548172757475083e-07,
+      "loss": 0.3164,
+      "step": 1088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.547757475083056e-07,
+      "loss": 0.3347,
+      "step": 1089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5473421926910296e-07,
+      "loss": 0.2702,
+      "step": 1090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5469269102990035e-07,
+      "loss": 0.2679,
+      "step": 1091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5465116279069763e-07,
+      "loss": 0.2866,
+      "step": 1092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.54609634551495e-07,
+      "loss": 0.2686,
+      "step": 1093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5456810631229236e-07,
+      "loss": 0.2681,
+      "step": 1094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.545265780730897e-07,
+      "loss": 0.3091,
+      "step": 1095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5448504983388703e-07,
+      "loss": 0.3158,
+      "step": 1096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5444352159468437e-07,
+      "loss": 0.2073,
+      "step": 1097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.544019933554817e-07,
+      "loss": 0.2735,
+      "step": 1098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5436046511627904e-07,
+      "loss": 0.3427,
+      "step": 1099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.543189368770764e-07,
+      "loss": 0.2981,
+      "step": 1100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.5427740863787377e-07,
+      "loss": 0.2954,
+      "step": 1101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.542358803986711e-07,
+      "loss": 0.3507,
+      "step": 1102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.541943521594684e-07,
+      "loss": 0.2549,
+      "step": 1103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.541528239202658e-07,
+      "loss": 0.3202,
+      "step": 1104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.541112956810631e-07,
+      "loss": 0.2811,
+      "step": 1105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5406976744186045e-07,
+      "loss": 0.3542,
+      "step": 1106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.540282392026578e-07,
+      "loss": 0.2803,
+      "step": 1107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5398671096345513e-07,
+      "loss": 0.3091,
+      "step": 1108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.539451827242525e-07,
+      "loss": 0.3261,
+      "step": 1109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.539036544850498e-07,
+      "loss": 0.2977,
+      "step": 1110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5386212624584714e-07,
+      "loss": 0.2724,
+      "step": 1111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5382059800664453e-07,
+      "loss": 0.2434,
+      "step": 1112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.537790697674418e-07,
+      "loss": 0.3399,
+      "step": 1113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.537375415282392e-07,
+      "loss": 0.2886,
+      "step": 1114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5369601328903654e-07,
+      "loss": 0.2832,
+      "step": 1115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.536544850498339e-07,
+      "loss": 0.2838,
+      "step": 1116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.536129568106312e-07,
+      "loss": 0.2946,
+      "step": 1117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5357142857142855e-07,
+      "loss": 0.2514,
+      "step": 1118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.535299003322259e-07,
+      "loss": 0.3254,
+      "step": 1119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.534883720930232e-07,
+      "loss": 0.266,
+      "step": 1120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5344684385382056e-07,
+      "loss": 0.2203,
+      "step": 1121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5340531561461795e-07,
+      "loss": 0.2712,
+      "step": 1122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.533637873754153e-07,
+      "loss": 0.3119,
+      "step": 1123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.533222591362126e-07,
+      "loss": 0.2799,
+      "step": 1124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5328073089700996e-07,
+      "loss": 0.3136,
+      "step": 1125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.532392026578073e-07,
+      "loss": 0.2914,
+      "step": 1126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5319767441860463e-07,
+      "loss": 0.3409,
+      "step": 1127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5315614617940197e-07,
+      "loss": 0.3606,
+      "step": 1128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.531146179401993e-07,
+      "loss": 0.2846,
+      "step": 1129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.530730897009967e-07,
+      "loss": 0.2669,
+      "step": 1130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.53031561461794e-07,
+      "loss": 0.326,
+      "step": 1131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5299003322259137e-07,
+      "loss": 0.2902,
+      "step": 1132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.529485049833887e-07,
+      "loss": 0.2797,
+      "step": 1133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5290697674418604e-07,
+      "loss": 0.3348,
+      "step": 1134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.528654485049834e-07,
+      "loss": 0.2942,
+      "step": 1135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.528239202657807e-07,
+      "loss": 0.2993,
+      "step": 1136
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.5278239202657805e-07,
+      "loss": 0.31,
+      "step": 1137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.527408637873754e-07,
+      "loss": 0.2584,
+      "step": 1138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5269933554817273e-07,
+      "loss": 0.2603,
+      "step": 1139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.526578073089701e-07,
+      "loss": 0.3091,
+      "step": 1140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5261627906976745e-07,
+      "loss": 0.2991,
+      "step": 1141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.525747508305648e-07,
+      "loss": 0.3241,
+      "step": 1142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5253322259136213e-07,
+      "loss": 0.3232,
+      "step": 1143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5249169435215946e-07,
+      "loss": 0.3029,
+      "step": 1144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.524501661129568e-07,
+      "loss": 0.3288,
+      "step": 1145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5240863787375414e-07,
+      "loss": 0.2743,
+      "step": 1146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.523671096345515e-07,
+      "loss": 0.2648,
+      "step": 1147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5232558139534886e-07,
+      "loss": 0.2968,
+      "step": 1148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5228405315614615e-07,
+      "loss": 0.2663,
+      "step": 1149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.522425249169435e-07,
+      "loss": 0.191,
+      "step": 1150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.522009966777409e-07,
+      "loss": 0.3312,
+      "step": 1151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5215946843853816e-07,
+      "loss": 0.2559,
+      "step": 1152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5211794019933555e-07,
+      "loss": 0.2772,
+      "step": 1153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.520764119601329e-07,
+      "loss": 0.2945,
+      "step": 1154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.520348837209302e-07,
+      "loss": 0.323,
+      "step": 1155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5199335548172756e-07,
+      "loss": 0.281,
+      "step": 1156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.519518272425249e-07,
+      "loss": 0.264,
+      "step": 1157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5191029900332223e-07,
+      "loss": 0.244,
+      "step": 1158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5186877076411957e-07,
+      "loss": 0.2785,
+      "step": 1159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.518272425249169e-07,
+      "loss": 0.2634,
+      "step": 1160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.517857142857143e-07,
+      "loss": 0.3149,
+      "step": 1161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5174418604651163e-07,
+      "loss": 0.2877,
+      "step": 1162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5170265780730897e-07,
+      "loss": 0.3067,
+      "step": 1163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.516611295681063e-07,
+      "loss": 0.3325,
+      "step": 1164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5161960132890364e-07,
+      "loss": 0.2394,
+      "step": 1165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.51578073089701e-07,
+      "loss": 0.2976,
+      "step": 1166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.515365448504983e-07,
+      "loss": 0.2625,
+      "step": 1167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5149501661129565e-07,
+      "loss": 0.2671,
+      "step": 1168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5145348837209304e-07,
+      "loss": 0.2803,
+      "step": 1169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.514119601328903e-07,
+      "loss": 0.2769,
+      "step": 1170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.513704318936877e-07,
+      "loss": 0.3488,
+      "step": 1171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.5132890365448505e-07,
+      "loss": 0.3135,
+      "step": 1172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.512873754152824e-07,
+      "loss": 0.3114,
+      "step": 1173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.512458471760797e-07,
+      "loss": 0.3084,
+      "step": 1174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5120431893687706e-07,
+      "loss": 0.2711,
+      "step": 1175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.511627906976744e-07,
+      "loss": 0.3195,
+      "step": 1176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5112126245847174e-07,
+      "loss": 0.2855,
+      "step": 1177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5107973421926907e-07,
+      "loss": 0.2467,
+      "step": 1178
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5103820598006646e-07,
+      "loss": 0.29,
+      "step": 1179
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5099667774086375e-07,
+      "loss": 0.2806,
+      "step": 1180
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5095514950166114e-07,
+      "loss": 0.238,
+      "step": 1181
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5091362126245847e-07,
+      "loss": 0.2789,
+      "step": 1182
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.508720930232558e-07,
+      "loss": 0.2023,
+      "step": 1183
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5083056478405315e-07,
+      "loss": 0.2733,
+      "step": 1184
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507890365448505e-07,
+      "loss": 0.2936,
+      "step": 1185
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.507475083056478e-07,
+      "loss": 0.2844,
+      "step": 1186
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5070598006644516e-07,
+      "loss": 0.3081,
+      "step": 1187
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506644518272425e-07,
+      "loss": 0.2953,
+      "step": 1188
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.506229235880399e-07,
+      "loss": 0.3059,
+      "step": 1189
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.505813953488372e-07,
+      "loss": 0.2726,
+      "step": 1190
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.505398671096345e-07,
+      "loss": 0.3006,
+      "step": 1191
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.504983388704319e-07,
+      "loss": 0.2709,
+      "step": 1192
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5045681063122923e-07,
+      "loss": 0.3006,
+      "step": 1193
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5041528239202657e-07,
+      "loss": 0.2698,
+      "step": 1194
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.503737541528239e-07,
+      "loss": 0.2235,
+      "step": 1195
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5033222591362124e-07,
+      "loss": 0.2886,
+      "step": 1196
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5029069767441863e-07,
+      "loss": 0.2925,
+      "step": 1197
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.502491694352159e-07,
+      "loss": 0.3728,
+      "step": 1198
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5020764119601325e-07,
+      "loss": 0.3405,
+      "step": 1199
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5016611295681064e-07,
+      "loss": 0.2527,
+      "step": 1200
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.50124584717608e-07,
+      "loss": 0.2268,
+      "step": 1201
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.500830564784053e-07,
+      "loss": 0.2585,
+      "step": 1202
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5004152823920265e-07,
+      "loss": 0.3507,
+      "step": 1203
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.5e-07,
+      "loss": 0.3054,
+      "step": 1204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.499584717607973e-07,
+      "loss": 0.3004,
+      "step": 1205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.4991694352159466e-07,
+      "loss": 0.2662,
+      "step": 1206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.49875415282392e-07,
+      "loss": 0.2892,
+      "step": 1207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.498338870431894e-07,
+      "loss": 0.2904,
+      "step": 1208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.4979235880398667e-07,
+      "loss": 0.2631,
+      "step": 1209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.4975083056478406e-07,
+      "loss": 0.2575,
+      "step": 1210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.497093023255814e-07,
+      "loss": 0.3115,
+      "step": 1211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4966777408637874e-07,
+      "loss": 0.2967,
+      "step": 1212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4962624584717607e-07,
+      "loss": 0.2445,
+      "step": 1213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495847176079734e-07,
+      "loss": 0.3266,
+      "step": 1214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4954318936877075e-07,
+      "loss": 0.3318,
+      "step": 1215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.495016611295681e-07,
+      "loss": 0.3441,
+      "step": 1216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.494601328903654e-07,
+      "loss": 0.2832,
+      "step": 1217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.494186046511628e-07,
+      "loss": 0.3196,
+      "step": 1218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.493770764119601e-07,
+      "loss": 0.3091,
+      "step": 1219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.493355481727575e-07,
+      "loss": 0.2444,
+      "step": 1220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.492940199335548e-07,
+      "loss": 0.334,
+      "step": 1221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4925249169435216e-07,
+      "loss": 0.3313,
+      "step": 1222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.492109634551495e-07,
+      "loss": 0.2264,
+      "step": 1223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4916943521594683e-07,
+      "loss": 0.2202,
+      "step": 1224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4912790697674417e-07,
+      "loss": 0.2839,
+      "step": 1225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.490863787375415e-07,
+      "loss": 0.3465,
+      "step": 1226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4904485049833884e-07,
+      "loss": 0.2491,
+      "step": 1227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4900332225913623e-07,
+      "loss": 0.3251,
+      "step": 1228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4896179401993357e-07,
+      "loss": 0.3195,
+      "step": 1229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4892026578073085e-07,
+      "loss": 0.3272,
+      "step": 1230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4887873754152824e-07,
+      "loss": 0.2881,
+      "step": 1231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.488372093023256e-07,
+      "loss": 0.318,
+      "step": 1232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.487956810631229e-07,
+      "loss": 0.3225,
+      "step": 1233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4875415282392025e-07,
+      "loss": 0.2637,
+      "step": 1234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.487126245847176e-07,
+      "loss": 0.3199,
+      "step": 1235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48671096345515e-07,
+      "loss": 0.296,
+      "step": 1236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4862956810631226e-07,
+      "loss": 0.2816,
+      "step": 1237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.485880398671096e-07,
+      "loss": 0.2578,
+      "step": 1238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48546511627907e-07,
+      "loss": 0.3055,
+      "step": 1239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4850498338870427e-07,
+      "loss": 0.2927,
+      "step": 1240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4846345514950166e-07,
+      "loss": 0.2962,
+      "step": 1241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48421926910299e-07,
+      "loss": 0.3079,
+      "step": 1242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4838039867109634e-07,
+      "loss": 0.2232,
+      "step": 1243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4833887043189367e-07,
+      "loss": 0.2712,
+      "step": 1244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.48297342192691e-07,
+      "loss": 0.2957,
+      "step": 1245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.4825581395348835e-07,
+      "loss": 0.2877,
+      "step": 1246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.482142857142857e-07,
+      "loss": 0.2375,
+      "step": 1247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.48172757475083e-07,
+      "loss": 0.2951,
+      "step": 1248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.481312292358804e-07,
+      "loss": 0.2525,
+      "step": 1249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4808970099667775e-07,
+      "loss": 0.2586,
+      "step": 1250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.480481727574751e-07,
+      "loss": 0.325,
+      "step": 1251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.480066445182724e-07,
+      "loss": 0.2783,
+      "step": 1252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4796511627906976e-07,
+      "loss": 0.2566,
+      "step": 1253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.479235880398671e-07,
+      "loss": 0.3161,
+      "step": 1254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4788205980066443e-07,
+      "loss": 0.3111,
+      "step": 1255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4784053156146177e-07,
+      "loss": 0.2807,
+      "step": 1256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4779900332225916e-07,
+      "loss": 0.2745,
+      "step": 1257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4775747508305644e-07,
+      "loss": 0.2955,
+      "step": 1258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4771594684385383e-07,
+      "loss": 0.3092,
+      "step": 1259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4767441860465117e-07,
+      "loss": 0.3042,
+      "step": 1260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.476328903654485e-07,
+      "loss": 0.2985,
+      "step": 1261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4759136212624584e-07,
+      "loss": 0.2956,
+      "step": 1262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.475498338870432e-07,
+      "loss": 0.2846,
+      "step": 1263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.475083056478405e-07,
+      "loss": 0.2873,
+      "step": 1264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4746677740863785e-07,
+      "loss": 0.258,
+      "step": 1265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.474252491694352e-07,
+      "loss": 0.2921,
+      "step": 1266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.473837209302326e-07,
+      "loss": 0.2658,
+      "step": 1267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.473421926910299e-07,
+      "loss": 0.3656,
+      "step": 1268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.473006644518272e-07,
+      "loss": 0.2458,
+      "step": 1269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.472591362126246e-07,
+      "loss": 0.3142,
+      "step": 1270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.472176079734219e-07,
+      "loss": 0.2474,
+      "step": 1271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4717607973421926e-07,
+      "loss": 0.266,
+      "step": 1272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.471345514950166e-07,
+      "loss": 0.2947,
+      "step": 1273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4709302325581393e-07,
+      "loss": 0.305,
+      "step": 1274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470514950166113e-07,
+      "loss": 0.3018,
+      "step": 1275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.470099667774086e-07,
+      "loss": 0.3001,
+      "step": 1276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4696843853820594e-07,
+      "loss": 0.3961,
+      "step": 1277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4692691029900333e-07,
+      "loss": 0.2894,
+      "step": 1278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.468853820598006e-07,
+      "loss": 0.2679,
+      "step": 1279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.46843853820598e-07,
+      "loss": 0.387,
+      "step": 1280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.4680232558139534e-07,
+      "loss": 0.2827,
+      "step": 1281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.467607973421927e-07,
+      "loss": 0.2682,
+      "step": 1282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4671926910299e-07,
+      "loss": 0.3116,
+      "step": 1283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4667774086378736e-07,
+      "loss": 0.2453,
+      "step": 1284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.466362126245847e-07,
+      "loss": 0.3006,
+      "step": 1285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4659468438538203e-07,
+      "loss": 0.278,
+      "step": 1286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4655315614617937e-07,
+      "loss": 0.2285,
+      "step": 1287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4651162790697676e-07,
+      "loss": 0.2786,
+      "step": 1288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.464700996677741e-07,
+      "loss": 0.2905,
+      "step": 1289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4642857142857143e-07,
+      "loss": 0.2948,
+      "step": 1290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4638704318936877e-07,
+      "loss": 0.2725,
+      "step": 1291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.463455149501661e-07,
+      "loss": 0.3123,
+      "step": 1292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4630398671096344e-07,
+      "loss": 0.2673,
+      "step": 1293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.462624584717608e-07,
+      "loss": 0.286,
+      "step": 1294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.462209302325581e-07,
+      "loss": 0.2923,
+      "step": 1295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.461794019933555e-07,
+      "loss": 0.3244,
+      "step": 1296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.461378737541528e-07,
+      "loss": 0.2963,
+      "step": 1297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.460963455149502e-07,
+      "loss": 0.32,
+      "step": 1298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.460548172757475e-07,
+      "loss": 0.2894,
+      "step": 1299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4601328903654485e-07,
+      "loss": 0.2855,
+      "step": 1300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.459717607973422e-07,
+      "loss": 0.2831,
+      "step": 1301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.459302325581395e-07,
+      "loss": 0.2892,
+      "step": 1302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4588870431893686e-07,
+      "loss": 0.2749,
+      "step": 1303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.458471760797342e-07,
+      "loss": 0.3052,
+      "step": 1304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4580564784053153e-07,
+      "loss": 0.2999,
+      "step": 1305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.457641196013289e-07,
+      "loss": 0.2689,
+      "step": 1306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.457225913621262e-07,
+      "loss": 0.3017,
+      "step": 1307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.456810631229236e-07,
+      "loss": 0.2696,
+      "step": 1308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4563953488372093e-07,
+      "loss": 0.2273,
+      "step": 1309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4559800664451827e-07,
+      "loss": 0.265,
+      "step": 1310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.455564784053156e-07,
+      "loss": 0.2966,
+      "step": 1311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4551495016611294e-07,
+      "loss": 0.3057,
+      "step": 1312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.454734219269103e-07,
+      "loss": 0.2878,
+      "step": 1313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.454318936877076e-07,
+      "loss": 0.2899,
+      "step": 1314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4539036544850495e-07,
+      "loss": 0.323,
+      "step": 1315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.453488372093023e-07,
+      "loss": 0.2631,
+      "step": 1316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.453073089700997e-07,
+      "loss": 0.3221,
+      "step": 1317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.4526578073089696e-07,
+      "loss": 0.2818,
+      "step": 1318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4522425249169435e-07,
+      "loss": 0.3185,
+      "step": 1319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.451827242524917e-07,
+      "loss": 0.3151,
+      "step": 1320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4514119601328903e-07,
+      "loss": 0.342,
+      "step": 1321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4509966777408636e-07,
+      "loss": 0.2807,
+      "step": 1322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.450581395348837e-07,
+      "loss": 0.2992,
+      "step": 1323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4501661129568104e-07,
+      "loss": 0.285,
+      "step": 1324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.449750830564784e-07,
+      "loss": 0.2896,
+      "step": 1325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.449335548172757e-07,
+      "loss": 0.3231,
+      "step": 1326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.448920265780731e-07,
+      "loss": 0.3569,
+      "step": 1327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4485049833887044e-07,
+      "loss": 0.2866,
+      "step": 1328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.448089700996678e-07,
+      "loss": 0.2457,
+      "step": 1329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.447674418604651e-07,
+      "loss": 0.2771,
+      "step": 1330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4472591362126245e-07,
+      "loss": 0.3362,
+      "step": 1331
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.446843853820598e-07,
+      "loss": 0.2896,
+      "step": 1332
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.446428571428571e-07,
+      "loss": 0.2725,
+      "step": 1333
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4460132890365446e-07,
+      "loss": 0.2818,
+      "step": 1334
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4455980066445185e-07,
+      "loss": 0.346,
+      "step": 1335
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4451827242524913e-07,
+      "loss": 0.3359,
+      "step": 1336
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.444767441860465e-07,
+      "loss": 0.3064,
+      "step": 1337
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4443521594684386e-07,
+      "loss": 0.2793,
+      "step": 1338
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.443936877076412e-07,
+      "loss": 0.3347,
+      "step": 1339
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4435215946843853e-07,
+      "loss": 0.2985,
+      "step": 1340
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4431063122923587e-07,
+      "loss": 0.2973,
+      "step": 1341
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.442691029900332e-07,
+      "loss": 0.2759,
+      "step": 1342
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4422757475083054e-07,
+      "loss": 0.265,
+      "step": 1343
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.441860465116279e-07,
+      "loss": 0.3147,
+      "step": 1344
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4414451827242527e-07,
+      "loss": 0.37,
+      "step": 1345
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4410299003322255e-07,
+      "loss": 0.2604,
+      "step": 1346
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4406146179401994e-07,
+      "loss": 0.3185,
+      "step": 1347
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.440199335548173e-07,
+      "loss": 0.2606,
+      "step": 1348
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.439784053156146e-07,
+      "loss": 0.2757,
+      "step": 1349
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4393687707641195e-07,
+      "loss": 0.3254,
+      "step": 1350
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.438953488372093e-07,
+      "loss": 0.2412,
+      "step": 1351
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4385382059800663e-07,
+      "loss": 0.2623,
+      "step": 1352
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.4381229235880396e-07,
+      "loss": 0.2789,
+      "step": 1353
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.437707641196013e-07,
+      "loss": 0.3164,
+      "step": 1354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.437292358803987e-07,
+      "loss": 0.2586,
+      "step": 1355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4368770764119603e-07,
+      "loss": 0.2655,
+      "step": 1356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.436461794019933e-07,
+      "loss": 0.2485,
+      "step": 1357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.436046511627907e-07,
+      "loss": 0.3117,
+      "step": 1358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4356312292358804e-07,
+      "loss": 0.3066,
+      "step": 1359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.435215946843854e-07,
+      "loss": 0.387,
+      "step": 1360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.434800664451827e-07,
+      "loss": 0.2753,
+      "step": 1361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4343853820598005e-07,
+      "loss": 0.2358,
+      "step": 1362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4339700996677744e-07,
+      "loss": 0.3062,
+      "step": 1363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.433554817275747e-07,
+      "loss": 0.2801,
+      "step": 1364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4331395348837206e-07,
+      "loss": 0.2698,
+      "step": 1365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4327242524916945e-07,
+      "loss": 0.2889,
+      "step": 1366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4323089700996673e-07,
+      "loss": 0.2264,
+      "step": 1367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.431893687707641e-07,
+      "loss": 0.2934,
+      "step": 1368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4314784053156146e-07,
+      "loss": 0.2953,
+      "step": 1369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.431063122923588e-07,
+      "loss": 0.3077,
+      "step": 1370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4306478405315613e-07,
+      "loss": 0.3402,
+      "step": 1371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4302325581395347e-07,
+      "loss": 0.3039,
+      "step": 1372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.429817275747508e-07,
+      "loss": 0.2908,
+      "step": 1373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4294019933554814e-07,
+      "loss": 0.2969,
+      "step": 1374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.428986710963455e-07,
+      "loss": 0.2841,
+      "step": 1375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4285714285714287e-07,
+      "loss": 0.3282,
+      "step": 1376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.428156146179402e-07,
+      "loss": 0.2876,
+      "step": 1377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4277408637873754e-07,
+      "loss": 0.2666,
+      "step": 1378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.427325581395349e-07,
+      "loss": 0.2704,
+      "step": 1379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.426910299003322e-07,
+      "loss": 0.301,
+      "step": 1380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4264950166112955e-07,
+      "loss": 0.2955,
+      "step": 1381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.426079734219269e-07,
+      "loss": 0.2689,
+      "step": 1382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.425664451827242e-07,
+      "loss": 0.3157,
+      "step": 1383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.425249169435216e-07,
+      "loss": 0.2899,
+      "step": 1384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.424833887043189e-07,
+      "loss": 0.2408,
+      "step": 1385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.424418604651163e-07,
+      "loss": 0.3146,
+      "step": 1386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.424003322259136e-07,
+      "loss": 0.3234,
+      "step": 1387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4235880398671096e-07,
+      "loss": 0.2832,
+      "step": 1388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.423172757475083e-07,
+      "loss": 0.2966,
+      "step": 1389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.4227574750830564e-07,
+      "loss": 0.2854,
+      "step": 1390
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.42234219269103e-07,
+      "loss": 0.2546,
+      "step": 1391
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.421926910299003e-07,
+      "loss": 0.2873,
+      "step": 1392
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4215116279069765e-07,
+      "loss": 0.2604,
+      "step": 1393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4210963455149504e-07,
+      "loss": 0.3499,
+      "step": 1394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.420681063122924e-07,
+      "loss": 0.2618,
+      "step": 1395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4202657807308966e-07,
+      "loss": 0.3016,
+      "step": 1396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4198504983388705e-07,
+      "loss": 0.2771,
+      "step": 1397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.419435215946844e-07,
+      "loss": 0.3098,
+      "step": 1398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.419019933554817e-07,
+      "loss": 0.2958,
+      "step": 1399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4186046511627906e-07,
+      "loss": 0.3493,
+      "step": 1400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.418189368770764e-07,
+      "loss": 0.2745,
+      "step": 1401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.417774086378738e-07,
+      "loss": 0.3165,
+      "step": 1402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4173588039867107e-07,
+      "loss": 0.2543,
+      "step": 1403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.416943521594684e-07,
+      "loss": 0.2613,
+      "step": 1404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.416528239202658e-07,
+      "loss": 0.2642,
+      "step": 1405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.416112956810631e-07,
+      "loss": 0.298,
+      "step": 1406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4156976744186047e-07,
+      "loss": 0.3324,
+      "step": 1407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.415282392026578e-07,
+      "loss": 0.2906,
+      "step": 1408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4148671096345514e-07,
+      "loss": 0.3896,
+      "step": 1409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.414451827242525e-07,
+      "loss": 0.3093,
+      "step": 1410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.414036544850498e-07,
+      "loss": 0.2204,
+      "step": 1411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4136212624584715e-07,
+      "loss": 0.3511,
+      "step": 1412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.413205980066445e-07,
+      "loss": 0.3034,
+      "step": 1413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.412790697674418e-07,
+      "loss": 0.3743,
+      "step": 1414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.412375415282392e-07,
+      "loss": 0.2785,
+      "step": 1415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4119601328903655e-07,
+      "loss": 0.2616,
+      "step": 1416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.411544850498339e-07,
+      "loss": 0.2833,
+      "step": 1417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.411129568106312e-07,
+      "loss": 0.3262,
+      "step": 1418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4107142857142856e-07,
+      "loss": 0.3349,
+      "step": 1419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.410299003322259e-07,
+      "loss": 0.345,
+      "step": 1420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4098837209302324e-07,
+      "loss": 0.253,
+      "step": 1421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4094684385382057e-07,
+      "loss": 0.2369,
+      "step": 1422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4090531561461796e-07,
+      "loss": 0.2738,
+      "step": 1423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4086378737541525e-07,
+      "loss": 0.3887,
+      "step": 1424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4082225913621264e-07,
+      "loss": 0.2949,
+      "step": 1425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.4078073089700997e-07,
+      "loss": 0.2599,
+      "step": 1426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.407392026578073e-07,
+      "loss": 0.2977,
+      "step": 1427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4069767441860465e-07,
+      "loss": 0.2926,
+      "step": 1428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.40656146179402e-07,
+      "loss": 0.3143,
+      "step": 1429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.406146179401993e-07,
+      "loss": 0.2436,
+      "step": 1430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4057308970099666e-07,
+      "loss": 0.22,
+      "step": 1431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.40531561461794e-07,
+      "loss": 0.2765,
+      "step": 1432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.404900332225914e-07,
+      "loss": 0.2626,
+      "step": 1433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4044850498338867e-07,
+      "loss": 0.3466,
+      "step": 1434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.40406976744186e-07,
+      "loss": 0.2984,
+      "step": 1435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.403654485049834e-07,
+      "loss": 0.3057,
+      "step": 1436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4032392026578073e-07,
+      "loss": 0.279,
+      "step": 1437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4028239202657807e-07,
+      "loss": 0.2494,
+      "step": 1438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.402408637873754e-07,
+      "loss": 0.2674,
+      "step": 1439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4019933554817274e-07,
+      "loss": 0.28,
+      "step": 1440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.401578073089701e-07,
+      "loss": 0.2455,
+      "step": 1441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.401162790697674e-07,
+      "loss": 0.2895,
+      "step": 1442
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4007475083056475e-07,
+      "loss": 0.3614,
+      "step": 1443
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.4003322259136214e-07,
+      "loss": 0.3549,
+      "step": 1444
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.399916943521594e-07,
+      "loss": 0.2993,
+      "step": 1445
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.399501661129568e-07,
+      "loss": 0.3116,
+      "step": 1446
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3990863787375415e-07,
+      "loss": 0.2985,
+      "step": 1447
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.398671096345515e-07,
+      "loss": 0.2895,
+      "step": 1448
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.398255813953488e-07,
+      "loss": 0.2808,
+      "step": 1449
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3978405315614616e-07,
+      "loss": 0.2931,
+      "step": 1450
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.397425249169435e-07,
+      "loss": 0.2997,
+      "step": 1451
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3970099667774084e-07,
+      "loss": 0.2421,
+      "step": 1452
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3965946843853817e-07,
+      "loss": 0.3085,
+      "step": 1453
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3961794019933556e-07,
+      "loss": 0.28,
+      "step": 1454
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.395764119601329e-07,
+      "loss": 0.373,
+      "step": 1455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3953488372093024e-07,
+      "loss": 0.2415,
+      "step": 1456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3949335548172757e-07,
+      "loss": 0.2791,
+      "step": 1457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.394518272425249e-07,
+      "loss": 0.407,
+      "step": 1458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.3941029900332225e-07,
+      "loss": 0.3423,
+      "step": 1459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.393687707641196e-07,
+      "loss": 0.3327,
+      "step": 1460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.393272425249169e-07,
+      "loss": 0.2805,
+      "step": 1461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.392857142857143e-07,
+      "loss": 0.2979,
+      "step": 1462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.392441860465116e-07,
+      "loss": 0.3198,
+      "step": 1463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.39202657807309e-07,
+      "loss": 0.3088,
+      "step": 1464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.391611295681063e-07,
+      "loss": 0.301,
+      "step": 1465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3911960132890366e-07,
+      "loss": 0.3182,
+      "step": 1466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.39078073089701e-07,
+      "loss": 0.2902,
+      "step": 1467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3903654485049833e-07,
+      "loss": 0.3282,
+      "step": 1468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3899501661129567e-07,
+      "loss": 0.2472,
+      "step": 1469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.38953488372093e-07,
+      "loss": 0.2567,
+      "step": 1470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3891196013289034e-07,
+      "loss": 0.2456,
+      "step": 1471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3887043189368773e-07,
+      "loss": 0.3413,
+      "step": 1472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.38828903654485e-07,
+      "loss": 0.2786,
+      "step": 1473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.387873754152824e-07,
+      "loss": 0.2685,
+      "step": 1474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3874584717607974e-07,
+      "loss": 0.3334,
+      "step": 1475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.387043189368771e-07,
+      "loss": 0.3184,
+      "step": 1476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.386627906976744e-07,
+      "loss": 0.2718,
+      "step": 1477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3862126245847175e-07,
+      "loss": 0.3168,
+      "step": 1478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.385797342192691e-07,
+      "loss": 0.2776,
+      "step": 1479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.385382059800664e-07,
+      "loss": 0.2732,
+      "step": 1480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3849667774086376e-07,
+      "loss": 0.3058,
+      "step": 1481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.384551495016611e-07,
+      "loss": 0.2856,
+      "step": 1482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.384136212624585e-07,
+      "loss": 0.2848,
+      "step": 1483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.383720930232558e-07,
+      "loss": 0.2492,
+      "step": 1484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3833056478405313e-07,
+      "loss": 0.3077,
+      "step": 1485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.382890365448505e-07,
+      "loss": 0.2971,
+      "step": 1486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.382475083056478e-07,
+      "loss": 0.2794,
+      "step": 1487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3820598006644517e-07,
+      "loss": 0.2379,
+      "step": 1488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.381644518272425e-07,
+      "loss": 0.3477,
+      "step": 1489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3812292358803987e-07,
+      "loss": 0.3019,
+      "step": 1490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3808139534883718e-07,
+      "loss": 0.3347,
+      "step": 1491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3803986710963454e-07,
+      "loss": 0.255,
+      "step": 1492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3799833887043188e-07,
+      "loss": 0.2911,
+      "step": 1493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3795681063122922e-07,
+      "loss": 0.3145,
+      "step": 1494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3791528239202656e-07,
+      "loss": 0.3318,
+      "step": 1495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3787375415282392e-07,
+      "loss": 0.2643,
+      "step": 1496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.3783222591362126e-07,
+      "loss": 0.3095,
+      "step": 1497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.377906976744186e-07,
+      "loss": 0.2938,
+      "step": 1498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3774916943521593e-07,
+      "loss": 0.3673,
+      "step": 1499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.377076411960133e-07,
+      "loss": 0.3525,
+      "step": 1500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.376661129568106e-07,
+      "loss": 0.2363,
+      "step": 1501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3762458471760797e-07,
+      "loss": 0.2921,
+      "step": 1502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.375830564784053e-07,
+      "loss": 0.226,
+      "step": 1503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3754152823920267e-07,
+      "loss": 0.3376,
+      "step": 1504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3749999999999998e-07,
+      "loss": 0.2775,
+      "step": 1505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3745847176079734e-07,
+      "loss": 0.3244,
+      "step": 1506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3741694352159468e-07,
+      "loss": 0.2948,
+      "step": 1507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.37375415282392e-07,
+      "loss": 0.3361,
+      "step": 1508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3733388704318935e-07,
+      "loss": 0.2914,
+      "step": 1509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.372923588039867e-07,
+      "loss": 0.2962,
+      "step": 1510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3725083056478405e-07,
+      "loss": 0.2568,
+      "step": 1511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3720930232558136e-07,
+      "loss": 0.2821,
+      "step": 1512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3716777408637872e-07,
+      "loss": 0.242,
+      "step": 1513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3712624584717609e-07,
+      "loss": 0.3246,
+      "step": 1514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.370847176079734e-07,
+      "loss": 0.296,
+      "step": 1515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3704318936877073e-07,
+      "loss": 0.263,
+      "step": 1516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.370016611295681e-07,
+      "loss": 0.2521,
+      "step": 1517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3696013289036546e-07,
+      "loss": 0.3142,
+      "step": 1518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3691860465116277e-07,
+      "loss": 0.2792,
+      "step": 1519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.368770764119601e-07,
+      "loss": 0.2561,
+      "step": 1520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3683554817275747e-07,
+      "loss": 0.2537,
+      "step": 1521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3679401993355483e-07,
+      "loss": 0.2936,
+      "step": 1522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3675249169435214e-07,
+      "loss": 0.2961,
+      "step": 1523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3671096345514948e-07,
+      "loss": 0.3079,
+      "step": 1524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3666943521594684e-07,
+      "loss": 0.3294,
+      "step": 1525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3662790697674415e-07,
+      "loss": 0.311,
+      "step": 1526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3658637873754152e-07,
+      "loss": 0.2787,
+      "step": 1527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3654485049833885e-07,
+      "loss": 0.2756,
+      "step": 1528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3650332225913622e-07,
+      "loss": 0.2828,
+      "step": 1529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3646179401993353e-07,
+      "loss": 0.3058,
+      "step": 1530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.364202657807309e-07,
+      "loss": 0.2704,
+      "step": 1531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3637873754152823e-07,
+      "loss": 0.2548,
+      "step": 1532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3633720930232556e-07,
+      "loss": 0.3233,
+      "step": 1533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.362956810631229e-07,
+      "loss": 0.3049,
+      "step": 1534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.3625415282392026e-07,
+      "loss": 0.3114,
+      "step": 1535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.362126245847176e-07,
+      "loss": 0.2858,
+      "step": 1536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3617109634551494e-07,
+      "loss": 0.3149,
+      "step": 1537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3612956810631228e-07,
+      "loss": 0.2825,
+      "step": 1538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3608803986710964e-07,
+      "loss": 0.3194,
+      "step": 1539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3604651162790695e-07,
+      "loss": 0.3322,
+      "step": 1540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.360049833887043e-07,
+      "loss": 0.3014,
+      "step": 1541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3596345514950165e-07,
+      "loss": 0.3464,
+      "step": 1542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.35921926910299e-07,
+      "loss": 0.3441,
+      "step": 1543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3588039867109632e-07,
+      "loss": 0.2786,
+      "step": 1544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3583887043189369e-07,
+      "loss": 0.2957,
+      "step": 1545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3579734219269102e-07,
+      "loss": 0.339,
+      "step": 1546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3575581395348836e-07,
+      "loss": 0.2902,
+      "step": 1547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.357142857142857e-07,
+      "loss": 0.2839,
+      "step": 1548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3567275747508306e-07,
+      "loss": 0.2902,
+      "step": 1549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.356312292358804e-07,
+      "loss": 0.3247,
+      "step": 1550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3558970099667773e-07,
+      "loss": 0.3734,
+      "step": 1551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3554817275747507e-07,
+      "loss": 0.2967,
+      "step": 1552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3550664451827243e-07,
+      "loss": 0.2708,
+      "step": 1553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3546511627906974e-07,
+      "loss": 0.265,
+      "step": 1554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.354235880398671e-07,
+      "loss": 0.2566,
+      "step": 1555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3538205980066444e-07,
+      "loss": 0.3155,
+      "step": 1556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.353405315614618e-07,
+      "loss": 0.3176,
+      "step": 1557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3529900332225912e-07,
+      "loss": 0.2865,
+      "step": 1558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3525747508305648e-07,
+      "loss": 0.2467,
+      "step": 1559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3521594684385382e-07,
+      "loss": 0.249,
+      "step": 1560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3517441860465113e-07,
+      "loss": 0.2897,
+      "step": 1561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.351328903654485e-07,
+      "loss": 0.2874,
+      "step": 1562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3509136212624583e-07,
+      "loss": 0.2844,
+      "step": 1563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.350498338870432e-07,
+      "loss": 0.337,
+      "step": 1564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.350083056478405e-07,
+      "loss": 0.3049,
+      "step": 1565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3496677740863786e-07,
+      "loss": 0.297,
+      "step": 1566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.349252491694352e-07,
+      "loss": 0.3264,
+      "step": 1567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3488372093023254e-07,
+      "loss": 0.2925,
+      "step": 1568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3484219269102987e-07,
+      "loss": 0.2581,
+      "step": 1569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3480066445182724e-07,
+      "loss": 0.3111,
+      "step": 1570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.3475913621262457e-07,
+      "loss": 0.2438,
+      "step": 1571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.347176079734219e-07,
+      "loss": 0.3128,
+      "step": 1572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3467607973421925e-07,
+      "loss": 0.276,
+      "step": 1573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.346345514950166e-07,
+      "loss": 0.3272,
+      "step": 1574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3459302325581392e-07,
+      "loss": 0.2694,
+      "step": 1575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3455149501661128e-07,
+      "loss": 0.2556,
+      "step": 1576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3450996677740862e-07,
+      "loss": 0.2878,
+      "step": 1577
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3446843853820598e-07,
+      "loss": 0.3336,
+      "step": 1578
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.344269102990033e-07,
+      "loss": 0.3405,
+      "step": 1579
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3438538205980066e-07,
+      "loss": 0.3455,
+      "step": 1580
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.34343853820598e-07,
+      "loss": 0.2449,
+      "step": 1581
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3430232558139536e-07,
+      "loss": 0.351,
+      "step": 1582
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3426079734219267e-07,
+      "loss": 0.3023,
+      "step": 1583
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3421926910299003e-07,
+      "loss": 0.3269,
+      "step": 1584
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3417774086378737e-07,
+      "loss": 0.2507,
+      "step": 1585
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.341362126245847e-07,
+      "loss": 0.316,
+      "step": 1586
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3409468438538204e-07,
+      "loss": 0.2905,
+      "step": 1587
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.340531561461794e-07,
+      "loss": 0.2733,
+      "step": 1588
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3401162790697674e-07,
+      "loss": 0.2854,
+      "step": 1589
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3397009966777408e-07,
+      "loss": 0.289,
+      "step": 1590
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3392857142857142e-07,
+      "loss": 0.3127,
+      "step": 1591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3388704318936878e-07,
+      "loss": 0.2908,
+      "step": 1592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.338455149501661e-07,
+      "loss": 0.292,
+      "step": 1593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3380398671096345e-07,
+      "loss": 0.3103,
+      "step": 1594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.337624584717608e-07,
+      "loss": 0.2299,
+      "step": 1595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3372093023255815e-07,
+      "loss": 0.2575,
+      "step": 1596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3367940199335546e-07,
+      "loss": 0.3274,
+      "step": 1597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3363787375415283e-07,
+      "loss": 0.3019,
+      "step": 1598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3359634551495016e-07,
+      "loss": 0.2291,
+      "step": 1599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3355481727574747e-07,
+      "loss": 0.271,
+      "step": 1600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3351328903654484e-07,
+      "loss": 0.2276,
+      "step": 1601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.334717607973422e-07,
+      "loss": 0.3172,
+      "step": 1602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3343023255813954e-07,
+      "loss": 0.279,
+      "step": 1603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3338870431893685e-07,
+      "loss": 0.2816,
+      "step": 1604
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.333471760797342e-07,
+      "loss": 0.3402,
+      "step": 1605
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3330564784053157e-07,
+      "loss": 0.296,
+      "step": 1606
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.3326411960132888e-07,
+      "loss": 0.3018,
+      "step": 1607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3322259136212622e-07,
+      "loss": 0.2551,
+      "step": 1608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3318106312292358e-07,
+      "loss": 0.231,
+      "step": 1609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3313953488372092e-07,
+      "loss": 0.281,
+      "step": 1610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3309800664451826e-07,
+      "loss": 0.2557,
+      "step": 1611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.330564784053156e-07,
+      "loss": 0.2939,
+      "step": 1612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3301495016611296e-07,
+      "loss": 0.3004,
+      "step": 1613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3297342192691027e-07,
+      "loss": 0.241,
+      "step": 1614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3293189368770763e-07,
+      "loss": 0.2844,
+      "step": 1615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3289036544850497e-07,
+      "loss": 0.2601,
+      "step": 1616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3284883720930233e-07,
+      "loss": 0.3081,
+      "step": 1617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3280730897009964e-07,
+      "loss": 0.2475,
+      "step": 1618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.32765780730897e-07,
+      "loss": 0.2523,
+      "step": 1619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3272425249169434e-07,
+      "loss": 0.3149,
+      "step": 1620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3268272425249168e-07,
+      "loss": 0.2917,
+      "step": 1621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3264119601328902e-07,
+      "loss": 0.3069,
+      "step": 1622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3259966777408638e-07,
+      "loss": 0.2743,
+      "step": 1623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3255813953488372e-07,
+      "loss": 0.3224,
+      "step": 1624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3251661129568105e-07,
+      "loss": 0.2418,
+      "step": 1625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.324750830564784e-07,
+      "loss": 0.2612,
+      "step": 1626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3243355481727575e-07,
+      "loss": 0.2952,
+      "step": 1627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3239202657807306e-07,
+      "loss": 0.3256,
+      "step": 1628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3235049833887043e-07,
+      "loss": 0.3495,
+      "step": 1629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3230897009966776e-07,
+      "loss": 0.2574,
+      "step": 1630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3226744186046513e-07,
+      "loss": 0.2783,
+      "step": 1631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3222591362126244e-07,
+      "loss": 0.3342,
+      "step": 1632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.321843853820598e-07,
+      "loss": 0.2999,
+      "step": 1633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3214285714285714e-07,
+      "loss": 0.3247,
+      "step": 1634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3210132890365445e-07,
+      "loss": 0.2454,
+      "step": 1635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.320598006644518e-07,
+      "loss": 0.2935,
+      "step": 1636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3201827242524917e-07,
+      "loss": 0.2909,
+      "step": 1637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.319767441860465e-07,
+      "loss": 0.3462,
+      "step": 1638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3193521594684382e-07,
+      "loss": 0.2508,
+      "step": 1639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3189368770764118e-07,
+      "loss": 0.3043,
+      "step": 1640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3185215946843855e-07,
+      "loss": 0.3421,
+      "step": 1641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.3181063122923586e-07,
+      "loss": 0.3074,
+      "step": 1642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.317691029900332e-07,
+      "loss": 0.307,
+      "step": 1643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3172757475083056e-07,
+      "loss": 0.3173,
+      "step": 1644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3168604651162792e-07,
+      "loss": 0.3404,
+      "step": 1645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3164451827242523e-07,
+      "loss": 0.2941,
+      "step": 1646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3160299003322257e-07,
+      "loss": 0.2885,
+      "step": 1647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3156146179401993e-07,
+      "loss": 0.235,
+      "step": 1648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.315199335548173e-07,
+      "loss": 0.2197,
+      "step": 1649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.314784053156146e-07,
+      "loss": 0.2571,
+      "step": 1650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3143687707641194e-07,
+      "loss": 0.3181,
+      "step": 1651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.313953488372093e-07,
+      "loss": 0.2386,
+      "step": 1652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3135382059800661e-07,
+      "loss": 0.3133,
+      "step": 1653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3131229235880398e-07,
+      "loss": 0.3065,
+      "step": 1654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3127076411960131e-07,
+      "loss": 0.3413,
+      "step": 1655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3122923588039868e-07,
+      "loss": 0.2766,
+      "step": 1656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.31187707641196e-07,
+      "loss": 0.3679,
+      "step": 1657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3114617940199335e-07,
+      "loss": 0.3337,
+      "step": 1658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.311046511627907e-07,
+      "loss": 0.2558,
+      "step": 1659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3106312292358802e-07,
+      "loss": 0.282,
+      "step": 1660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3102159468438536e-07,
+      "loss": 0.2944,
+      "step": 1661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3098006644518272e-07,
+      "loss": 0.2754,
+      "step": 1662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3093853820598006e-07,
+      "loss": 0.2164,
+      "step": 1663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.308970099667774e-07,
+      "loss": 0.356,
+      "step": 1664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3085548172757474e-07,
+      "loss": 0.3181,
+      "step": 1665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.308139534883721e-07,
+      "loss": 0.2896,
+      "step": 1666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.307724252491694e-07,
+      "loss": 0.2821,
+      "step": 1667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3073089700996677e-07,
+      "loss": 0.2301,
+      "step": 1668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.306893687707641e-07,
+      "loss": 0.3277,
+      "step": 1669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3064784053156147e-07,
+      "loss": 0.3133,
+      "step": 1670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3060631229235878e-07,
+      "loss": 0.3501,
+      "step": 1671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3056478405315615e-07,
+      "loss": 0.2207,
+      "step": 1672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3052325581395348e-07,
+      "loss": 0.3196,
+      "step": 1673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.304817275747508e-07,
+      "loss": 0.2951,
+      "step": 1674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3044019933554816e-07,
+      "loss": 0.2393,
+      "step": 1675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3039867109634552e-07,
+      "loss": 0.292,
+      "step": 1676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3035714285714286e-07,
+      "loss": 0.2986,
+      "step": 1677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3031561461794017e-07,
+      "loss": 0.2552,
+      "step": 1678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.3027408637873753e-07,
+      "loss": 0.2872,
+      "step": 1679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.302325581395349e-07,
+      "loss": 0.263,
+      "step": 1680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.301910299003322e-07,
+      "loss": 0.2797,
+      "step": 1681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3014950166112954e-07,
+      "loss": 0.3164,
+      "step": 1682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.301079734219269e-07,
+      "loss": 0.3,
+      "step": 1683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3006644518272427e-07,
+      "loss": 0.2644,
+      "step": 1684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3002491694352158e-07,
+      "loss": 0.3183,
+      "step": 1685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2998338870431891e-07,
+      "loss": 0.274,
+      "step": 1686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2994186046511628e-07,
+      "loss": 0.344,
+      "step": 1687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.299003322259136e-07,
+      "loss": 0.3242,
+      "step": 1688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2985880398671095e-07,
+      "loss": 0.3111,
+      "step": 1689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.298172757475083e-07,
+      "loss": 0.2788,
+      "step": 1690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2977574750830565e-07,
+      "loss": 0.3164,
+      "step": 1691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2973421926910296e-07,
+      "loss": 0.3272,
+      "step": 1692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2969269102990032e-07,
+      "loss": 0.2753,
+      "step": 1693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2965116279069766e-07,
+      "loss": 0.3381,
+      "step": 1694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.29609634551495e-07,
+      "loss": 0.3376,
+      "step": 1695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2956810631229233e-07,
+      "loss": 0.2662,
+      "step": 1696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.295265780730897e-07,
+      "loss": 0.3538,
+      "step": 1697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2948504983388703e-07,
+      "loss": 0.2412,
+      "step": 1698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2944352159468437e-07,
+      "loss": 0.3026,
+      "step": 1699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.294019933554817e-07,
+      "loss": 0.2732,
+      "step": 1700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2936046511627907e-07,
+      "loss": 0.3011,
+      "step": 1701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2931893687707638e-07,
+      "loss": 0.2806,
+      "step": 1702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2927740863787374e-07,
+      "loss": 0.3297,
+      "step": 1703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2923588039867108e-07,
+      "loss": 0.3625,
+      "step": 1704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2919435215946845e-07,
+      "loss": 0.2975,
+      "step": 1705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2915282392026576e-07,
+      "loss": 0.2704,
+      "step": 1706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2911129568106312e-07,
+      "loss": 0.2746,
+      "step": 1707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2906976744186046e-07,
+      "loss": 0.3467,
+      "step": 1708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.290282392026578e-07,
+      "loss": 0.2347,
+      "step": 1709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2898671096345513e-07,
+      "loss": 0.2832,
+      "step": 1710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.289451827242525e-07,
+      "loss": 0.3056,
+      "step": 1711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2890365448504983e-07,
+      "loss": 0.2885,
+      "step": 1712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2886212624584717e-07,
+      "loss": 0.3135,
+      "step": 1713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.288205980066445e-07,
+      "loss": 0.3359,
+      "step": 1714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.2877906976744187e-07,
+      "loss": 0.2544,
+      "step": 1715
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.287375415282392e-07,
+      "loss": 0.3213,
+      "step": 1716
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2869601328903654e-07,
+      "loss": 0.2473,
+      "step": 1717
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2865448504983388e-07,
+      "loss": 0.2329,
+      "step": 1718
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2861295681063124e-07,
+      "loss": 0.2719,
+      "step": 1719
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2857142857142855e-07,
+      "loss": 0.2332,
+      "step": 1720
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.285299003322259e-07,
+      "loss": 0.3419,
+      "step": 1721
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2848837209302325e-07,
+      "loss": 0.2963,
+      "step": 1722
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.284468438538206e-07,
+      "loss": 0.285,
+      "step": 1723
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2840531561461792e-07,
+      "loss": 0.3242,
+      "step": 1724
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2836378737541526e-07,
+      "loss": 0.2649,
+      "step": 1725
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2832225913621262e-07,
+      "loss": 0.3553,
+      "step": 1726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2828073089700993e-07,
+      "loss": 0.2706,
+      "step": 1727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.282392026578073e-07,
+      "loss": 0.2872,
+      "step": 1728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2819767441860463e-07,
+      "loss": 0.2631,
+      "step": 1729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.28156146179402e-07,
+      "loss": 0.2639,
+      "step": 1730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.281146179401993e-07,
+      "loss": 0.3266,
+      "step": 1731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2807308970099667e-07,
+      "loss": 0.2512,
+      "step": 1732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.28031561461794e-07,
+      "loss": 0.33,
+      "step": 1733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2799003322259134e-07,
+      "loss": 0.2626,
+      "step": 1734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2794850498338868e-07,
+      "loss": 0.2381,
+      "step": 1735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2790697674418604e-07,
+      "loss": 0.2964,
+      "step": 1736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2786544850498338e-07,
+      "loss": 0.2598,
+      "step": 1737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2782392026578072e-07,
+      "loss": 0.2922,
+      "step": 1738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2778239202657805e-07,
+      "loss": 0.2811,
+      "step": 1739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2774086378737542e-07,
+      "loss": 0.2958,
+      "step": 1740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2769933554817273e-07,
+      "loss": 0.3677,
+      "step": 1741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.276578073089701e-07,
+      "loss": 0.2861,
+      "step": 1742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2761627906976743e-07,
+      "loss": 0.2973,
+      "step": 1743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.275747508305648e-07,
+      "loss": 0.2722,
+      "step": 1744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.275332225913621e-07,
+      "loss": 0.2914,
+      "step": 1745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2749169435215947e-07,
+      "loss": 0.2322,
+      "step": 1746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.274501661129568e-07,
+      "loss": 0.3081,
+      "step": 1747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2740863787375414e-07,
+      "loss": 0.2823,
+      "step": 1748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2736710963455148e-07,
+      "loss": 0.2801,
+      "step": 1749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2732558139534884e-07,
+      "loss": 0.3536,
+      "step": 1750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.2728405315614618e-07,
+      "loss": 0.2901,
+      "step": 1751
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.272425249169435e-07,
+      "loss": 0.2865,
+      "step": 1752
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2720099667774085e-07,
+      "loss": 0.2393,
+      "step": 1753
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.271594684385382e-07,
+      "loss": 0.2606,
+      "step": 1754
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2711794019933552e-07,
+      "loss": 0.2623,
+      "step": 1755
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2707641196013289e-07,
+      "loss": 0.2953,
+      "step": 1756
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2703488372093022e-07,
+      "loss": 0.2502,
+      "step": 1757
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2699335548172759e-07,
+      "loss": 0.3196,
+      "step": 1758
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.269518272425249e-07,
+      "loss": 0.2897,
+      "step": 1759
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2691029900332226e-07,
+      "loss": 0.2367,
+      "step": 1760
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.268687707641196e-07,
+      "loss": 0.2983,
+      "step": 1761
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.268272425249169e-07,
+      "loss": 0.3357,
+      "step": 1762
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2678571428571427e-07,
+      "loss": 0.2518,
+      "step": 1763
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2674418604651163e-07,
+      "loss": 0.2637,
+      "step": 1764
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2670265780730897e-07,
+      "loss": 0.2832,
+      "step": 1765
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2666112956810628e-07,
+      "loss": 0.329,
+      "step": 1766
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2661960132890364e-07,
+      "loss": 0.2087,
+      "step": 1767
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.26578073089701e-07,
+      "loss": 0.2737,
+      "step": 1768
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2653654485049832e-07,
+      "loss": 0.2942,
+      "step": 1769
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2649501661129565e-07,
+      "loss": 0.2533,
+      "step": 1770
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2645348837209302e-07,
+      "loss": 0.2548,
+      "step": 1771
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2641196013289038e-07,
+      "loss": 0.3015,
+      "step": 1772
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.263704318936877e-07,
+      "loss": 0.3872,
+      "step": 1773
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2632890365448503e-07,
+      "loss": 0.2869,
+      "step": 1774
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.262873754152824e-07,
+      "loss": 0.2511,
+      "step": 1775
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2624584717607973e-07,
+      "loss": 0.3114,
+      "step": 1776
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2620431893687706e-07,
+      "loss": 0.3756,
+      "step": 1777
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.261627906976744e-07,
+      "loss": 0.3306,
+      "step": 1778
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2612126245847176e-07,
+      "loss": 0.2675,
+      "step": 1779
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2607973421926907e-07,
+      "loss": 0.3019,
+      "step": 1780
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2603820598006644e-07,
+      "loss": 0.261,
+      "step": 1781
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2599667774086377e-07,
+      "loss": 0.3793,
+      "step": 1782
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2595514950166114e-07,
+      "loss": 0.3078,
+      "step": 1783
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2591362126245845e-07,
+      "loss": 0.3357,
+      "step": 1784
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.258720930232558e-07,
+      "loss": 0.2845,
+      "step": 1785
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2583056478405315e-07,
+      "loss": 0.2775,
+      "step": 1786
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.2578903654485048e-07,
+      "loss": 0.2967,
+      "step": 1787
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2574750830564782e-07,
+      "loss": 0.3097,
+      "step": 1788
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2570598006644519e-07,
+      "loss": 0.2912,
+      "step": 1789
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2566445182724252e-07,
+      "loss": 0.3186,
+      "step": 1790
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2562292358803986e-07,
+      "loss": 0.3053,
+      "step": 1791
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.255813953488372e-07,
+      "loss": 0.2853,
+      "step": 1792
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2553986710963456e-07,
+      "loss": 0.2751,
+      "step": 1793
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2549833887043187e-07,
+      "loss": 0.2657,
+      "step": 1794
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2545681063122923e-07,
+      "loss": 0.2744,
+      "step": 1795
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2541528239202657e-07,
+      "loss": 0.2635,
+      "step": 1796
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2537375415282393e-07,
+      "loss": 0.2936,
+      "step": 1797
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2533222591362124e-07,
+      "loss": 0.2385,
+      "step": 1798
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.252906976744186e-07,
+      "loss": 0.3052,
+      "step": 1799
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2524916943521594e-07,
+      "loss": 0.2587,
+      "step": 1800
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2520764119601325e-07,
+      "loss": 0.3415,
+      "step": 1801
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2516611295681062e-07,
+      "loss": 0.2475,
+      "step": 1802
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2512458471760798e-07,
+      "loss": 0.2902,
+      "step": 1803
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2508305647840532e-07,
+      "loss": 0.2436,
+      "step": 1804
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2504152823920263e-07,
+      "loss": 0.3296,
+      "step": 1805
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.25e-07,
+      "loss": 0.3158,
+      "step": 1806
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2495847176079735e-07,
+      "loss": 0.2993,
+      "step": 1807
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2491694352159466e-07,
+      "loss": 0.2908,
+      "step": 1808
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.24875415282392e-07,
+      "loss": 0.2551,
+      "step": 1809
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2483388704318936e-07,
+      "loss": 0.2482,
+      "step": 1810
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2479235880398673e-07,
+      "loss": 0.3005,
+      "step": 1811
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2475083056478404e-07,
+      "loss": 0.2805,
+      "step": 1812
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2470930232558137e-07,
+      "loss": 0.3226,
+      "step": 1813
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2466777408637874e-07,
+      "loss": 0.255,
+      "step": 1814
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2462624584717605e-07,
+      "loss": 0.2667,
+      "step": 1815
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.245847176079734e-07,
+      "loss": 0.3266,
+      "step": 1816
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2454318936877075e-07,
+      "loss": 0.2881,
+      "step": 1817
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.245016611295681e-07,
+      "loss": 0.3053,
+      "step": 1818
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2446013289036542e-07,
+      "loss": 0.3633,
+      "step": 1819
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2441860465116278e-07,
+      "loss": 0.2722,
+      "step": 1820
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2437707641196012e-07,
+      "loss": 0.3754,
+      "step": 1821
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2433554817275746e-07,
+      "loss": 0.3392,
+      "step": 1822
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.242940199335548e-07,
+      "loss": 0.2881,
+      "step": 1823
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.2425249169435216e-07,
+      "loss": 0.303,
+      "step": 1824
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.242109634551495e-07,
+      "loss": 0.302,
+      "step": 1825
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2416943521594683e-07,
+      "loss": 0.2608,
+      "step": 1826
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2412790697674417e-07,
+      "loss": 0.2901,
+      "step": 1827
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2408637873754153e-07,
+      "loss": 0.3366,
+      "step": 1828
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2404485049833884e-07,
+      "loss": 0.3109,
+      "step": 1829
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.240033222591362e-07,
+      "loss": 0.2293,
+      "step": 1830
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2396179401993354e-07,
+      "loss": 0.3034,
+      "step": 1831
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.239202657807309e-07,
+      "loss": 0.2868,
+      "step": 1832
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2387873754152822e-07,
+      "loss": 0.2528,
+      "step": 1833
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2383720930232558e-07,
+      "loss": 0.2964,
+      "step": 1834
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2379568106312292e-07,
+      "loss": 0.2355,
+      "step": 1835
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2375415282392025e-07,
+      "loss": 0.2405,
+      "step": 1836
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.237126245847176e-07,
+      "loss": 0.2582,
+      "step": 1837
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2367109634551495e-07,
+      "loss": 0.2825,
+      "step": 1838
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.236295681063123e-07,
+      "loss": 0.2598,
+      "step": 1839
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.235880398671096e-07,
+      "loss": 0.3005,
+      "step": 1840
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2354651162790696e-07,
+      "loss": 0.3796,
+      "step": 1841
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2350498338870433e-07,
+      "loss": 0.333,
+      "step": 1842
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2346345514950166e-07,
+      "loss": 0.2847,
+      "step": 1843
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2342192691029897e-07,
+      "loss": 0.2749,
+      "step": 1844
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2338039867109634e-07,
+      "loss": 0.2432,
+      "step": 1845
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.233388704318937e-07,
+      "loss": 0.2822,
+      "step": 1846
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.23297342192691e-07,
+      "loss": 0.2817,
+      "step": 1847
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2325581395348835e-07,
+      "loss": 0.2922,
+      "step": 1848
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.232142857142857e-07,
+      "loss": 0.3047,
+      "step": 1849
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2317275747508307e-07,
+      "loss": 0.2683,
+      "step": 1850
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2313122923588038e-07,
+      "loss": 0.2892,
+      "step": 1851
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2308970099667772e-07,
+      "loss": 0.2408,
+      "step": 1852
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2304817275747508e-07,
+      "loss": 0.2897,
+      "step": 1853
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.230066445182724e-07,
+      "loss": 0.2685,
+      "step": 1854
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2296511627906976e-07,
+      "loss": 0.2738,
+      "step": 1855
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.229235880398671e-07,
+      "loss": 0.344,
+      "step": 1856
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2288205980066446e-07,
+      "loss": 0.31,
+      "step": 1857
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2284053156146177e-07,
+      "loss": 0.2472,
+      "step": 1858
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2279900332225913e-07,
+      "loss": 0.2693,
+      "step": 1859
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.2275747508305647e-07,
+      "loss": 0.301,
+      "step": 1860
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.227159468438538e-07,
+      "loss": 0.3037,
+      "step": 1861
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2267441860465114e-07,
+      "loss": 0.3042,
+      "step": 1862
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.226328903654485e-07,
+      "loss": 0.3062,
+      "step": 1863
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2259136212624584e-07,
+      "loss": 0.2718,
+      "step": 1864
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2254983388704318e-07,
+      "loss": 0.2478,
+      "step": 1865
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2250830564784051e-07,
+      "loss": 0.295,
+      "step": 1866
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2246677740863788e-07,
+      "loss": 0.2918,
+      "step": 1867
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.224252491694352e-07,
+      "loss": 0.2607,
+      "step": 1868
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2238372093023255e-07,
+      "loss": 0.3498,
+      "step": 1869
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.223421926910299e-07,
+      "loss": 0.3309,
+      "step": 1870
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2230066445182725e-07,
+      "loss": 0.2495,
+      "step": 1871
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2225913621262456e-07,
+      "loss": 0.2823,
+      "step": 1872
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2221760797342193e-07,
+      "loss": 0.2635,
+      "step": 1873
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2217607973421926e-07,
+      "loss": 0.3364,
+      "step": 1874
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.221345514950166e-07,
+      "loss": 0.2609,
+      "step": 1875
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2209302325581394e-07,
+      "loss": 0.3216,
+      "step": 1876
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.220514950166113e-07,
+      "loss": 0.3436,
+      "step": 1877
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2200996677740864e-07,
+      "loss": 0.2801,
+      "step": 1878
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2196843853820597e-07,
+      "loss": 0.3306,
+      "step": 1879
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.219269102990033e-07,
+      "loss": 0.3122,
+      "step": 1880
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2188538205980067e-07,
+      "loss": 0.3097,
+      "step": 1881
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2184385382059798e-07,
+      "loss": 0.2673,
+      "step": 1882
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2180232558139535e-07,
+      "loss": 0.2792,
+      "step": 1883
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2176079734219268e-07,
+      "loss": 0.2728,
+      "step": 1884
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2171926910299005e-07,
+      "loss": 0.2581,
+      "step": 1885
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2167774086378736e-07,
+      "loss": 0.287,
+      "step": 1886
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2163621262458472e-07,
+      "loss": 0.3528,
+      "step": 1887
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2159468438538206e-07,
+      "loss": 0.2927,
+      "step": 1888
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2155315614617937e-07,
+      "loss": 0.2968,
+      "step": 1889
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2151162790697673e-07,
+      "loss": 0.3084,
+      "step": 1890
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2147009966777407e-07,
+      "loss": 0.2852,
+      "step": 1891
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2142857142857143e-07,
+      "loss": 0.2065,
+      "step": 1892
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2138704318936874e-07,
+      "loss": 0.282,
+      "step": 1893
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.213455149501661e-07,
+      "loss": 0.3084,
+      "step": 1894
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2130398671096344e-07,
+      "loss": 0.2588,
+      "step": 1895
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.2126245847176078e-07,
+      "loss": 0.2626,
+      "step": 1896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2122093023255811e-07,
+      "loss": 0.3677,
+      "step": 1897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2117940199335548e-07,
+      "loss": 0.3243,
+      "step": 1898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2113787375415281e-07,
+      "loss": 0.2754,
+      "step": 1899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2109634551495015e-07,
+      "loss": 0.2642,
+      "step": 1900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.210548172757475e-07,
+      "loss": 0.2882,
+      "step": 1901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2101328903654485e-07,
+      "loss": 0.3429,
+      "step": 1902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.209717607973422e-07,
+      "loss": 0.3033,
+      "step": 1903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2093023255813952e-07,
+      "loss": 0.2586,
+      "step": 1904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2088870431893686e-07,
+      "loss": 0.2446,
+      "step": 1905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2084717607973422e-07,
+      "loss": 0.3303,
+      "step": 1906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2080564784053153e-07,
+      "loss": 0.3125,
+      "step": 1907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.207641196013289e-07,
+      "loss": 0.3767,
+      "step": 1908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2072259136212623e-07,
+      "loss": 0.3068,
+      "step": 1909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.206810631229236e-07,
+      "loss": 0.2861,
+      "step": 1910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.206395348837209e-07,
+      "loss": 0.296,
+      "step": 1911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2059800664451827e-07,
+      "loss": 0.2557,
+      "step": 1912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.205564784053156e-07,
+      "loss": 0.3668,
+      "step": 1913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2051495016611295e-07,
+      "loss": 0.231,
+      "step": 1914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2047342192691028e-07,
+      "loss": 0.3118,
+      "step": 1915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2043189368770765e-07,
+      "loss": 0.3641,
+      "step": 1916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2039036544850498e-07,
+      "loss": 0.3295,
+      "step": 1917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2034883720930232e-07,
+      "loss": 0.2882,
+      "step": 1918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2030730897009966e-07,
+      "loss": 0.322,
+      "step": 1919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2026578073089702e-07,
+      "loss": 0.3102,
+      "step": 1920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2022425249169433e-07,
+      "loss": 0.2805,
+      "step": 1921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.201827242524917e-07,
+      "loss": 0.3454,
+      "step": 1922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2014119601328903e-07,
+      "loss": 0.3133,
+      "step": 1923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.200996677740864e-07,
+      "loss": 0.2603,
+      "step": 1924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.200581395348837e-07,
+      "loss": 0.2613,
+      "step": 1925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2001661129568107e-07,
+      "loss": 0.3353,
+      "step": 1926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.199750830564784e-07,
+      "loss": 0.3305,
+      "step": 1927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.199335548172757e-07,
+      "loss": 0.3622,
+      "step": 1928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1989202657807308e-07,
+      "loss": 0.3278,
+      "step": 1929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1985049833887044e-07,
+      "loss": 0.3275,
+      "step": 1930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1980897009966778e-07,
+      "loss": 0.2949,
+      "step": 1931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.1976744186046509e-07,
+      "loss": 0.2968,
+      "step": 1932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1972591362126245e-07,
+      "loss": 0.3148,
+      "step": 1933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.196843853820598e-07,
+      "loss": 0.2873,
+      "step": 1934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1964285714285712e-07,
+      "loss": 0.2803,
+      "step": 1935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1960132890365446e-07,
+      "loss": 0.2378,
+      "step": 1936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1955980066445182e-07,
+      "loss": 0.3015,
+      "step": 1937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1951827242524919e-07,
+      "loss": 0.3238,
+      "step": 1938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.194767441860465e-07,
+      "loss": 0.2767,
+      "step": 1939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1943521594684383e-07,
+      "loss": 0.2712,
+      "step": 1940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.193936877076412e-07,
+      "loss": 0.3165,
+      "step": 1941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.193521594684385e-07,
+      "loss": 0.3228,
+      "step": 1942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1931063122923587e-07,
+      "loss": 0.3312,
+      "step": 1943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.192691029900332e-07,
+      "loss": 0.3049,
+      "step": 1944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1922757475083057e-07,
+      "loss": 0.34,
+      "step": 1945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1918604651162788e-07,
+      "loss": 0.2925,
+      "step": 1946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1914451827242524e-07,
+      "loss": 0.3107,
+      "step": 1947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1910299003322258e-07,
+      "loss": 0.2617,
+      "step": 1948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1906146179401992e-07,
+      "loss": 0.3044,
+      "step": 1949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1901993355481725e-07,
+      "loss": 0.2861,
+      "step": 1950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1897840531561462e-07,
+      "loss": 0.2761,
+      "step": 1951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1893687707641195e-07,
+      "loss": 0.3257,
+      "step": 1952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.188953488372093e-07,
+      "loss": 0.3708,
+      "step": 1953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1885382059800663e-07,
+      "loss": 0.2496,
+      "step": 1954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.18812292358804e-07,
+      "loss": 0.2889,
+      "step": 1955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.187707641196013e-07,
+      "loss": 0.266,
+      "step": 1956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1872923588039867e-07,
+      "loss": 0.3158,
+      "step": 1957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.18687707641196e-07,
+      "loss": 0.2532,
+      "step": 1958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1864617940199337e-07,
+      "loss": 0.2798,
+      "step": 1959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1860465116279068e-07,
+      "loss": 0.2897,
+      "step": 1960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1856312292358804e-07,
+      "loss": 0.2991,
+      "step": 1961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1852159468438538e-07,
+      "loss": 0.239,
+      "step": 1962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1848006644518269e-07,
+      "loss": 0.3309,
+      "step": 1963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1843853820598005e-07,
+      "loss": 0.3528,
+      "step": 1964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.183970099667774e-07,
+      "loss": 0.2637,
+      "step": 1965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1835548172757475e-07,
+      "loss": 0.2929,
+      "step": 1966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1831395348837206e-07,
+      "loss": 0.2594,
+      "step": 1967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1827242524916942e-07,
+      "loss": 0.3055,
+      "step": 1968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1823089700996679e-07,
+      "loss": 0.2818,
+      "step": 1969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1818936877076412e-07,
+      "loss": 0.2362,
+      "step": 1970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1814784053156143e-07,
+      "loss": 0.3015,
+      "step": 1971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.181063122923588e-07,
+      "loss": 0.2906,
+      "step": 1972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1806478405315616e-07,
+      "loss": 0.2703,
+      "step": 1973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1802325581395347e-07,
+      "loss": 0.2117,
+      "step": 1974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.179817275747508e-07,
+      "loss": 0.2795,
+      "step": 1975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1794019933554817e-07,
+      "loss": 0.2895,
+      "step": 1976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1789867109634553e-07,
+      "loss": 0.2831,
+      "step": 1977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1785714285714284e-07,
+      "loss": 0.2821,
+      "step": 1978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1781561461794018e-07,
+      "loss": 0.2957,
+      "step": 1979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1777408637873754e-07,
+      "loss": 0.2804,
+      "step": 1980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1773255813953485e-07,
+      "loss": 0.3235,
+      "step": 1981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1769102990033222e-07,
+      "loss": 0.2787,
+      "step": 1982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1764950166112955e-07,
+      "loss": 0.3265,
+      "step": 1983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1760797342192692e-07,
+      "loss": 0.2466,
+      "step": 1984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1756644518272423e-07,
+      "loss": 0.3064,
+      "step": 1985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.175249169435216e-07,
+      "loss": 0.3149,
+      "step": 1986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1748338870431893e-07,
+      "loss": 0.2799,
+      "step": 1987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1744186046511626e-07,
+      "loss": 0.2717,
+      "step": 1988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.174003322259136e-07,
+      "loss": 0.2919,
+      "step": 1989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1735880398671096e-07,
+      "loss": 0.3264,
+      "step": 1990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.173172757475083e-07,
+      "loss": 0.3153,
+      "step": 1991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1727574750830564e-07,
+      "loss": 0.3048,
+      "step": 1992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1723421926910297e-07,
+      "loss": 0.3588,
+      "step": 1993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1719269102990034e-07,
+      "loss": 0.3211,
+      "step": 1994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1715116279069765e-07,
+      "loss": 0.3284,
+      "step": 1995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.17109634551495e-07,
+      "loss": 0.2993,
+      "step": 1996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1706810631229235e-07,
+      "loss": 0.3172,
+      "step": 1997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.170265780730897e-07,
+      "loss": 0.3148,
+      "step": 1998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1698504983388702e-07,
+      "loss": 0.232,
+      "step": 1999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1694352159468439e-07,
+      "loss": 0.2387,
+      "step": 2000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1690199335548172e-07,
+      "loss": 0.2976,
+      "step": 2001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1686046511627906e-07,
+      "loss": 0.2795,
+      "step": 2002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.168189368770764e-07,
+      "loss": 0.2856,
+      "step": 2003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1677740863787376e-07,
+      "loss": 0.255,
+      "step": 2004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.167358803986711e-07,
+      "loss": 0.2632,
+      "step": 2005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.166943521594684e-07,
+      "loss": 0.3164,
+      "step": 2006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1665282392026577e-07,
+      "loss": 0.3437,
+      "step": 2007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1661129568106313e-07,
+      "loss": 0.2572,
+      "step": 2008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1656976744186044e-07,
+      "loss": 0.35,
+      "step": 2009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1652823920265778e-07,
+      "loss": 0.2755,
+      "step": 2010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1648671096345514e-07,
+      "loss": 0.247,
+      "step": 2011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.164451827242525e-07,
+      "loss": 0.2747,
+      "step": 2012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1640365448504982e-07,
+      "loss": 0.2759,
+      "step": 2013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1636212624584715e-07,
+      "loss": 0.2562,
+      "step": 2014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1632059800664452e-07,
+      "loss": 0.2644,
+      "step": 2015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1627906976744183e-07,
+      "loss": 0.2861,
+      "step": 2016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.162375415282392e-07,
+      "loss": 0.2316,
+      "step": 2017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1619601328903653e-07,
+      "loss": 0.2691,
+      "step": 2018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.161544850498339e-07,
+      "loss": 0.2852,
+      "step": 2019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.161129568106312e-07,
+      "loss": 0.2915,
+      "step": 2020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1607142857142856e-07,
+      "loss": 0.2908,
+      "step": 2021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.160299003322259e-07,
+      "loss": 0.3288,
+      "step": 2022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1598837209302324e-07,
+      "loss": 0.3049,
+      "step": 2023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1594684385382057e-07,
+      "loss": 0.2805,
+      "step": 2024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1590531561461794e-07,
+      "loss": 0.4086,
+      "step": 2025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1586378737541527e-07,
+      "loss": 0.2403,
+      "step": 2026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.158222591362126e-07,
+      "loss": 0.3137,
+      "step": 2027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1578073089700995e-07,
+      "loss": 0.2625,
+      "step": 2028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.157392026578073e-07,
+      "loss": 0.3371,
+      "step": 2029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1569767441860462e-07,
+      "loss": 0.2683,
+      "step": 2030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1565614617940198e-07,
+      "loss": 0.2751,
+      "step": 2031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1561461794019932e-07,
+      "loss": 0.2657,
+      "step": 2032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1557308970099668e-07,
+      "loss": 0.3224,
+      "step": 2033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.15531561461794e-07,
+      "loss": 0.2991,
+      "step": 2034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1549003322259136e-07,
+      "loss": 0.2727,
+      "step": 2035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.154485049833887e-07,
+      "loss": 0.266,
+      "step": 2036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1540697674418606e-07,
+      "loss": 0.275,
+      "step": 2037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1536544850498337e-07,
+      "loss": 0.3116,
+      "step": 2038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1532392026578073e-07,
+      "loss": 0.2032,
+      "step": 2039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.1528239202657807e-07,
+      "loss": 0.3118,
+      "step": 2040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.152408637873754e-07,
+      "loss": 0.284,
+      "step": 2041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1519933554817274e-07,
+      "loss": 0.314,
+      "step": 2042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.151578073089701e-07,
+      "loss": 0.2745,
+      "step": 2043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1511627906976744e-07,
+      "loss": 0.3461,
+      "step": 2044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1507475083056478e-07,
+      "loss": 0.2552,
+      "step": 2045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1503322259136212e-07,
+      "loss": 0.3022,
+      "step": 2046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1499169435215948e-07,
+      "loss": 0.266,
+      "step": 2047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.149501661129568e-07,
+      "loss": 0.2472,
+      "step": 2048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1490863787375415e-07,
+      "loss": 0.2813,
+      "step": 2049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.148671096345515e-07,
+      "loss": 0.2781,
+      "step": 2050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1482558139534885e-07,
+      "loss": 0.2553,
+      "step": 2051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1478405315614616e-07,
+      "loss": 0.2782,
+      "step": 2052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.147425249169435e-07,
+      "loss": 0.3149,
+      "step": 2053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1470099667774086e-07,
+      "loss": 0.3076,
+      "step": 2054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1465946843853817e-07,
+      "loss": 0.3279,
+      "step": 2055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1461794019933554e-07,
+      "loss": 0.3288,
+      "step": 2056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1457641196013287e-07,
+      "loss": 0.2984,
+      "step": 2057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1453488372093024e-07,
+      "loss": 0.5397,
+      "step": 2058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1449335548172755e-07,
+      "loss": 0.2837,
+      "step": 2059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.144518272425249e-07,
+      "loss": 0.3242,
+      "step": 2060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1441029900332225e-07,
+      "loss": 0.3023,
+      "step": 2061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1436877076411958e-07,
+      "loss": 0.3451,
+      "step": 2062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1432724252491692e-07,
+      "loss": 0.286,
+      "step": 2063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1428571428571428e-07,
+      "loss": 0.2576,
+      "step": 2064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1424418604651162e-07,
+      "loss": 0.2856,
+      "step": 2065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1420265780730896e-07,
+      "loss": 0.2778,
+      "step": 2066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.141611295681063e-07,
+      "loss": 0.306,
+      "step": 2067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1411960132890366e-07,
+      "loss": 0.3229,
+      "step": 2068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1407807308970097e-07,
+      "loss": 0.288,
+      "step": 2069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1403654485049833e-07,
+      "loss": 0.2608,
+      "step": 2070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1399501661129567e-07,
+      "loss": 0.2358,
+      "step": 2071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1395348837209303e-07,
+      "loss": 0.286,
+      "step": 2072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1391196013289034e-07,
+      "loss": 0.2988,
+      "step": 2073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.138704318936877e-07,
+      "loss": 0.3371,
+      "step": 2074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1382890365448504e-07,
+      "loss": 0.2754,
+      "step": 2075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.1378737541528238e-07,
+      "loss": 0.2353,
+      "step": 2076
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1374584717607971e-07,
+      "loss": 0.3322,
+      "step": 2077
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1370431893687708e-07,
+      "loss": 0.2446,
+      "step": 2078
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1366279069767441e-07,
+      "loss": 0.247,
+      "step": 2079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1362126245847175e-07,
+      "loss": 0.2805,
+      "step": 2080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.135797342192691e-07,
+      "loss": 0.3225,
+      "step": 2081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1353820598006645e-07,
+      "loss": 0.2962,
+      "step": 2082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1349667774086376e-07,
+      "loss": 0.3165,
+      "step": 2083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1345514950166113e-07,
+      "loss": 0.3213,
+      "step": 2084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1341362126245846e-07,
+      "loss": 0.2842,
+      "step": 2085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1337209302325583e-07,
+      "loss": 0.3078,
+      "step": 2086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1333056478405314e-07,
+      "loss": 0.3051,
+      "step": 2087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.132890365448505e-07,
+      "loss": 0.2583,
+      "step": 2088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1324750830564784e-07,
+      "loss": 0.2989,
+      "step": 2089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1320598006644515e-07,
+      "loss": 0.2623,
+      "step": 2090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.131644518272425e-07,
+      "loss": 0.2873,
+      "step": 2091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1312292358803987e-07,
+      "loss": 0.318,
+      "step": 2092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.130813953488372e-07,
+      "loss": 0.2543,
+      "step": 2093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1303986710963452e-07,
+      "loss": 0.3854,
+      "step": 2094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1299833887043188e-07,
+      "loss": 0.2894,
+      "step": 2095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1295681063122925e-07,
+      "loss": 0.2387,
+      "step": 2096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1291528239202658e-07,
+      "loss": 0.2641,
+      "step": 2097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.128737541528239e-07,
+      "loss": 0.3458,
+      "step": 2098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1283222591362126e-07,
+      "loss": 0.3076,
+      "step": 2099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1279069767441862e-07,
+      "loss": 0.2521,
+      "step": 2100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1274916943521593e-07,
+      "loss": 0.3022,
+      "step": 2101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1270764119601327e-07,
+      "loss": 0.2532,
+      "step": 2102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1266611295681063e-07,
+      "loss": 0.2617,
+      "step": 2103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1262458471760797e-07,
+      "loss": 0.3033,
+      "step": 2104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.125830564784053e-07,
+      "loss": 0.2406,
+      "step": 2105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1254152823920264e-07,
+      "loss": 0.2735,
+      "step": 2106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.125e-07,
+      "loss": 0.309,
+      "step": 2107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1245847176079731e-07,
+      "loss": 0.3173,
+      "step": 2108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1241694352159468e-07,
+      "loss": 0.2818,
+      "step": 2109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1237541528239201e-07,
+      "loss": 0.2861,
+      "step": 2110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1233388704318938e-07,
+      "loss": 0.2517,
+      "step": 2111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.122923588039867e-07,
+      "loss": 0.286,
+      "step": 2112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.1225083056478405e-07,
+      "loss": 0.2941,
+      "step": 2113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.122093023255814e-07,
+      "loss": 0.2905,
+      "step": 2114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1216777408637872e-07,
+      "loss": 0.3127,
+      "step": 2115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1212624584717606e-07,
+      "loss": 0.284,
+      "step": 2116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1208471760797342e-07,
+      "loss": 0.2732,
+      "step": 2117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1204318936877076e-07,
+      "loss": 0.2542,
+      "step": 2118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.120016611295681e-07,
+      "loss": 0.2575,
+      "step": 2119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1196013289036543e-07,
+      "loss": 0.3297,
+      "step": 2120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.119186046511628e-07,
+      "loss": 0.2753,
+      "step": 2121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.118770764119601e-07,
+      "loss": 0.2299,
+      "step": 2122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1183554817275747e-07,
+      "loss": 0.3239,
+      "step": 2123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.117940199335548e-07,
+      "loss": 0.3156,
+      "step": 2124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1175249169435217e-07,
+      "loss": 0.3006,
+      "step": 2125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1171096345514948e-07,
+      "loss": 0.2647,
+      "step": 2126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1166943521594685e-07,
+      "loss": 0.3483,
+      "step": 2127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1162790697674418e-07,
+      "loss": 0.2814,
+      "step": 2128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.115863787375415e-07,
+      "loss": 0.3198,
+      "step": 2129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1154485049833886e-07,
+      "loss": 0.307,
+      "step": 2130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1150332225913622e-07,
+      "loss": 0.2748,
+      "step": 2131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1146179401993356e-07,
+      "loss": 0.2992,
+      "step": 2132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1142026578073087e-07,
+      "loss": 0.35,
+      "step": 2133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1137873754152823e-07,
+      "loss": 0.2853,
+      "step": 2134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.113372093023256e-07,
+      "loss": 0.2229,
+      "step": 2135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.112956810631229e-07,
+      "loss": 0.3338,
+      "step": 2136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1125415282392024e-07,
+      "loss": 0.2611,
+      "step": 2137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.112126245847176e-07,
+      "loss": 0.2781,
+      "step": 2138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1117109634551497e-07,
+      "loss": 0.3076,
+      "step": 2139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1112956810631228e-07,
+      "loss": 0.2657,
+      "step": 2140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.110880398671096e-07,
+      "loss": 0.3807,
+      "step": 2141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1104651162790698e-07,
+      "loss": 0.3275,
+      "step": 2142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1100498338870429e-07,
+      "loss": 0.372,
+      "step": 2143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1096345514950165e-07,
+      "loss": 0.2704,
+      "step": 2144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.10921926910299e-07,
+      "loss": 0.2788,
+      "step": 2145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1088039867109635e-07,
+      "loss": 0.2574,
+      "step": 2146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1083887043189366e-07,
+      "loss": 0.285,
+      "step": 2147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1079734219269102e-07,
+      "loss": 0.3001,
+      "step": 2148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.1075581395348836e-07,
+      "loss": 0.3063,
+      "step": 2149
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.107142857142857e-07,
+      "loss": 0.3157,
+      "step": 2150
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1067275747508303e-07,
+      "loss": 0.3007,
+      "step": 2151
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.106312292358804e-07,
+      "loss": 0.3343,
+      "step": 2152
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1058970099667773e-07,
+      "loss": 0.2426,
+      "step": 2153
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1054817275747507e-07,
+      "loss": 0.3067,
+      "step": 2154
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.105066445182724e-07,
+      "loss": 0.2974,
+      "step": 2155
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1046511627906977e-07,
+      "loss": 0.2536,
+      "step": 2156
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1042358803986708e-07,
+      "loss": 0.2896,
+      "step": 2157
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1038205980066444e-07,
+      "loss": 0.317,
+      "step": 2158
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1034053156146178e-07,
+      "loss": 0.3162,
+      "step": 2159
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1029900332225914e-07,
+      "loss": 0.2708,
+      "step": 2160
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1025747508305645e-07,
+      "loss": 0.3221,
+      "step": 2161
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1021594684385382e-07,
+      "loss": 0.2755,
+      "step": 2162
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1017441860465115e-07,
+      "loss": 0.2698,
+      "step": 2163
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1013289036544852e-07,
+      "loss": 0.3707,
+      "step": 2164
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1009136212624583e-07,
+      "loss": 0.2461,
+      "step": 2165
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.100498338870432e-07,
+      "loss": 0.2763,
+      "step": 2166
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.1000830564784053e-07,
+      "loss": 0.3176,
+      "step": 2167
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0996677740863784e-07,
+      "loss": 0.2964,
+      "step": 2168
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.099252491694352e-07,
+      "loss": 0.3185,
+      "step": 2169
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0988372093023257e-07,
+      "loss": 0.3435,
+      "step": 2170
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.098421926910299e-07,
+      "loss": 0.2491,
+      "step": 2171
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.098006644518272e-07,
+      "loss": 0.3218,
+      "step": 2172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0975913621262458e-07,
+      "loss": 0.3643,
+      "step": 2173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0971760797342194e-07,
+      "loss": 0.2626,
+      "step": 2174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0967607973421925e-07,
+      "loss": 0.3159,
+      "step": 2175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0963455149501659e-07,
+      "loss": 0.3411,
+      "step": 2176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0959302325581395e-07,
+      "loss": 0.265,
+      "step": 2177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.095514950166113e-07,
+      "loss": 0.3156,
+      "step": 2178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0950996677740862e-07,
+      "loss": 0.2892,
+      "step": 2179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0946843853820596e-07,
+      "loss": 0.2246,
+      "step": 2180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0942691029900332e-07,
+      "loss": 0.2208,
+      "step": 2181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0938538205980063e-07,
+      "loss": 0.3156,
+      "step": 2182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.09343853820598e-07,
+      "loss": 0.2839,
+      "step": 2183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.0930232558139533e-07,
+      "loss": 0.2956,
+      "step": 2184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.092607973421927e-07,
+      "loss": 0.2783,
+      "step": 2185
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0921926910299e-07,
+      "loss": 0.2979,
+      "step": 2186
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0917774086378737e-07,
+      "loss": 0.3382,
+      "step": 2187
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.091362126245847e-07,
+      "loss": 0.3413,
+      "step": 2188
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0909468438538204e-07,
+      "loss": 0.2996,
+      "step": 2189
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0905315614617938e-07,
+      "loss": 0.3004,
+      "step": 2190
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0901162790697674e-07,
+      "loss": 0.2592,
+      "step": 2191
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0897009966777408e-07,
+      "loss": 0.2321,
+      "step": 2192
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0892857142857142e-07,
+      "loss": 0.4057,
+      "step": 2193
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0888704318936875e-07,
+      "loss": 0.3233,
+      "step": 2194
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0884551495016612e-07,
+      "loss": 0.2628,
+      "step": 2195
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0880398671096343e-07,
+      "loss": 0.275,
+      "step": 2196
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.087624584717608e-07,
+      "loss": 0.2796,
+      "step": 2197
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0872093023255813e-07,
+      "loss": 0.2606,
+      "step": 2198
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.086794019933555e-07,
+      "loss": 0.2671,
+      "step": 2199
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.086378737541528e-07,
+      "loss": 0.3325,
+      "step": 2200
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0859634551495016e-07,
+      "loss": 0.352,
+      "step": 2201
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.085548172757475e-07,
+      "loss": 0.3138,
+      "step": 2202
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0851328903654484e-07,
+      "loss": 0.2831,
+      "step": 2203
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0847176079734217e-07,
+      "loss": 0.273,
+      "step": 2204
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0843023255813954e-07,
+      "loss": 0.2592,
+      "step": 2205
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0838870431893687e-07,
+      "loss": 0.2799,
+      "step": 2206
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.083471760797342e-07,
+      "loss": 0.2228,
+      "step": 2207
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0830564784053155e-07,
+      "loss": 0.2654,
+      "step": 2208
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.082641196013289e-07,
+      "loss": 0.3069,
+      "step": 2209
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0822259136212622e-07,
+      "loss": 0.2341,
+      "step": 2210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0818106312292359e-07,
+      "loss": 0.2443,
+      "step": 2211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0813953488372092e-07,
+      "loss": 0.2587,
+      "step": 2212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0809800664451829e-07,
+      "loss": 0.2881,
+      "step": 2213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.080564784053156e-07,
+      "loss": 0.2761,
+      "step": 2214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0801495016611296e-07,
+      "loss": 0.261,
+      "step": 2215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.079734219269103e-07,
+      "loss": 0.2972,
+      "step": 2216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.079318936877076e-07,
+      "loss": 0.2916,
+      "step": 2217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0789036544850497e-07,
+      "loss": 0.2848,
+      "step": 2218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.078488372093023e-07,
+      "loss": 0.3074,
+      "step": 2219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0780730897009967e-07,
+      "loss": 0.2772,
+      "step": 2220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.0776578073089698e-07,
+      "loss": 0.3006,
+      "step": 2221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0772425249169434e-07,
+      "loss": 0.3681,
+      "step": 2222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0768272425249168e-07,
+      "loss": 0.2446,
+      "step": 2223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0764119601328902e-07,
+      "loss": 0.2833,
+      "step": 2224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0759966777408635e-07,
+      "loss": 0.2197,
+      "step": 2225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0755813953488372e-07,
+      "loss": 0.2499,
+      "step": 2226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0751661129568105e-07,
+      "loss": 0.2973,
+      "step": 2227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.074750830564784e-07,
+      "loss": 0.2762,
+      "step": 2228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0743355481727573e-07,
+      "loss": 0.2996,
+      "step": 2229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.073920265780731e-07,
+      "loss": 0.2713,
+      "step": 2230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0735049833887043e-07,
+      "loss": 0.2695,
+      "step": 2231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0730897009966776e-07,
+      "loss": 0.2265,
+      "step": 2232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.072674418604651e-07,
+      "loss": 0.3203,
+      "step": 2233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0722591362126246e-07,
+      "loss": 0.279,
+      "step": 2234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0718438538205977e-07,
+      "loss": 0.2812,
+      "step": 2235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0714285714285714e-07,
+      "loss": 0.3844,
+      "step": 2236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0710132890365447e-07,
+      "loss": 0.2992,
+      "step": 2237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0705980066445184e-07,
+      "loss": 0.2907,
+      "step": 2238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0701827242524915e-07,
+      "loss": 0.3551,
+      "step": 2239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.069767441860465e-07,
+      "loss": 0.2944,
+      "step": 2240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0693521594684385e-07,
+      "loss": 0.2337,
+      "step": 2241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0689368770764118e-07,
+      "loss": 0.2954,
+      "step": 2242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0685215946843852e-07,
+      "loss": 0.3247,
+      "step": 2243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0681063122923588e-07,
+      "loss": 0.314,
+      "step": 2244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0676910299003322e-07,
+      "loss": 0.2364,
+      "step": 2245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0672757475083056e-07,
+      "loss": 0.3046,
+      "step": 2246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.066860465116279e-07,
+      "loss": 0.2712,
+      "step": 2247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0664451827242526e-07,
+      "loss": 0.326,
+      "step": 2248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0660299003322257e-07,
+      "loss": 0.3179,
+      "step": 2249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0656146179401993e-07,
+      "loss": 0.2813,
+      "step": 2250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0651993355481727e-07,
+      "loss": 0.2864,
+      "step": 2251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0647840531561463e-07,
+      "loss": 0.3268,
+      "step": 2252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0643687707641194e-07,
+      "loss": 0.2725,
+      "step": 2253
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.063953488372093e-07,
+      "loss": 0.3105,
+      "step": 2254
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0635382059800664e-07,
+      "loss": 0.2928,
+      "step": 2255
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0631229235880395e-07,
+      "loss": 0.2795,
+      "step": 2256
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.0627076411960132e-07,
+      "loss": 0.3539,
+      "step": 2257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0622923588039868e-07,
+      "loss": 0.3134,
+      "step": 2258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0618770764119602e-07,
+      "loss": 0.2968,
+      "step": 2259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0614617940199333e-07,
+      "loss": 0.2829,
+      "step": 2260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.061046511627907e-07,
+      "loss": 0.284,
+      "step": 2261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0606312292358805e-07,
+      "loss": 0.3173,
+      "step": 2262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0602159468438536e-07,
+      "loss": 0.2839,
+      "step": 2263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.059800664451827e-07,
+      "loss": 0.2879,
+      "step": 2264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0593853820598006e-07,
+      "loss": 0.3166,
+      "step": 2265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0589700996677743e-07,
+      "loss": 0.3472,
+      "step": 2266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0585548172757474e-07,
+      "loss": 0.3155,
+      "step": 2267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0581395348837207e-07,
+      "loss": 0.257,
+      "step": 2268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0577242524916944e-07,
+      "loss": 0.3081,
+      "step": 2269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0573089700996675e-07,
+      "loss": 0.2857,
+      "step": 2270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.056893687707641e-07,
+      "loss": 0.2565,
+      "step": 2271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0564784053156145e-07,
+      "loss": 0.2756,
+      "step": 2272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.056063122923588e-07,
+      "loss": 0.2985,
+      "step": 2273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0556478405315612e-07,
+      "loss": 0.2784,
+      "step": 2274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0552325581395348e-07,
+      "loss": 0.2811,
+      "step": 2275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0548172757475082e-07,
+      "loss": 0.3117,
+      "step": 2276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0544019933554816e-07,
+      "loss": 0.2744,
+      "step": 2277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.053986710963455e-07,
+      "loss": 0.3251,
+      "step": 2278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0535714285714286e-07,
+      "loss": 0.2822,
+      "step": 2279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.053156146179402e-07,
+      "loss": 0.2635,
+      "step": 2280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0527408637873753e-07,
+      "loss": 0.2532,
+      "step": 2281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0523255813953487e-07,
+      "loss": 0.2937,
+      "step": 2282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0519102990033223e-07,
+      "loss": 0.2318,
+      "step": 2283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0514950166112954e-07,
+      "loss": 0.2752,
+      "step": 2284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.051079734219269e-07,
+      "loss": 0.2829,
+      "step": 2285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0506644518272424e-07,
+      "loss": 0.274,
+      "step": 2286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.050249169435216e-07,
+      "loss": 0.321,
+      "step": 2287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0498338870431891e-07,
+      "loss": 0.2731,
+      "step": 2288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0494186046511628e-07,
+      "loss": 0.2562,
+      "step": 2289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0490033222591361e-07,
+      "loss": 0.3157,
+      "step": 2290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0485880398671098e-07,
+      "loss": 0.2615,
+      "step": 2291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.048172757475083e-07,
+      "loss": 0.2957,
+      "step": 2292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.0477574750830565e-07,
+      "loss": 0.3054,
+      "step": 2293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.04734219269103e-07,
+      "loss": 0.3285,
+      "step": 2294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.046926910299003e-07,
+      "loss": 0.3144,
+      "step": 2295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0465116279069766e-07,
+      "loss": 0.2526,
+      "step": 2296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0460963455149503e-07,
+      "loss": 0.264,
+      "step": 2297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0456810631229236e-07,
+      "loss": 0.2857,
+      "step": 2298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0452657807308967e-07,
+      "loss": 0.2781,
+      "step": 2299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0448504983388704e-07,
+      "loss": 0.3187,
+      "step": 2300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.044435215946844e-07,
+      "loss": 0.2906,
+      "step": 2301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.044019933554817e-07,
+      "loss": 0.3184,
+      "step": 2302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0436046511627905e-07,
+      "loss": 0.2889,
+      "step": 2303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.043189368770764e-07,
+      "loss": 0.2505,
+      "step": 2304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0427740863787377e-07,
+      "loss": 0.2875,
+      "step": 2305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0423588039867108e-07,
+      "loss": 0.2632,
+      "step": 2306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0419435215946842e-07,
+      "loss": 0.2686,
+      "step": 2307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0415282392026578e-07,
+      "loss": 0.2726,
+      "step": 2308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.041112956810631e-07,
+      "loss": 0.2936,
+      "step": 2309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0406976744186046e-07,
+      "loss": 0.323,
+      "step": 2310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.040282392026578e-07,
+      "loss": 0.3101,
+      "step": 2311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0398671096345516e-07,
+      "loss": 0.3595,
+      "step": 2312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0394518272425247e-07,
+      "loss": 0.2375,
+      "step": 2313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0390365448504983e-07,
+      "loss": 0.3043,
+      "step": 2314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0386212624584717e-07,
+      "loss": 0.2807,
+      "step": 2315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.038205980066445e-07,
+      "loss": 0.2753,
+      "step": 2316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0377906976744184e-07,
+      "loss": 0.3199,
+      "step": 2317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.037375415282392e-07,
+      "loss": 0.3057,
+      "step": 2318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0369601328903654e-07,
+      "loss": 0.2929,
+      "step": 2319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0365448504983388e-07,
+      "loss": 0.3957,
+      "step": 2320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0361295681063121e-07,
+      "loss": 0.3049,
+      "step": 2321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0357142857142858e-07,
+      "loss": 0.2731,
+      "step": 2322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.035299003322259e-07,
+      "loss": 0.3134,
+      "step": 2323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0348837209302325e-07,
+      "loss": 0.3323,
+      "step": 2324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.034468438538206e-07,
+      "loss": 0.2956,
+      "step": 2325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0340531561461795e-07,
+      "loss": 0.2776,
+      "step": 2326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0336378737541526e-07,
+      "loss": 0.3255,
+      "step": 2327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0332225913621262e-07,
+      "loss": 0.2797,
+      "step": 2328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.0328073089700996e-07,
+      "loss": 0.3291,
+      "step": 2329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.032392026578073e-07,
+      "loss": 0.3445,
+      "step": 2330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0319767441860463e-07,
+      "loss": 0.3,
+      "step": 2331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.03156146179402e-07,
+      "loss": 0.3655,
+      "step": 2332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0311461794019933e-07,
+      "loss": 0.2837,
+      "step": 2333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0307308970099665e-07,
+      "loss": 0.3289,
+      "step": 2334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.03031561461794e-07,
+      "loss": 0.337,
+      "step": 2335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0299003322259137e-07,
+      "loss": 0.2932,
+      "step": 2336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0294850498338868e-07,
+      "loss": 0.2467,
+      "step": 2337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0290697674418602e-07,
+      "loss": 0.2816,
+      "step": 2338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0286544850498338e-07,
+      "loss": 0.3309,
+      "step": 2339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0282392026578075e-07,
+      "loss": 0.3231,
+      "step": 2340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0278239202657806e-07,
+      "loss": 0.2678,
+      "step": 2341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.027408637873754e-07,
+      "loss": 0.3056,
+      "step": 2342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0269933554817276e-07,
+      "loss": 0.2432,
+      "step": 2343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0265780730897007e-07,
+      "loss": 0.2992,
+      "step": 2344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0261627906976743e-07,
+      "loss": 0.2723,
+      "step": 2345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0257475083056477e-07,
+      "loss": 0.2867,
+      "step": 2346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0253322259136213e-07,
+      "loss": 0.2855,
+      "step": 2347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0249169435215944e-07,
+      "loss": 0.3065,
+      "step": 2348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.024501661129568e-07,
+      "loss": 0.2875,
+      "step": 2349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0240863787375414e-07,
+      "loss": 0.2028,
+      "step": 2350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0236710963455148e-07,
+      "loss": 0.2633,
+      "step": 2351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0232558139534881e-07,
+      "loss": 0.3556,
+      "step": 2352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0228405315614618e-07,
+      "loss": 0.29,
+      "step": 2353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0224252491694351e-07,
+      "loss": 0.3316,
+      "step": 2354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0220099667774085e-07,
+      "loss": 0.2488,
+      "step": 2355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.021594684385382e-07,
+      "loss": 0.3499,
+      "step": 2356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0211794019933555e-07,
+      "loss": 0.3292,
+      "step": 2357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.020764119601329e-07,
+      "loss": 0.3467,
+      "step": 2358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0203488372093022e-07,
+      "loss": 0.3429,
+      "step": 2359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0199335548172756e-07,
+      "loss": 0.263,
+      "step": 2360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0195182724252492e-07,
+      "loss": 0.2874,
+      "step": 2361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0191029900332223e-07,
+      "loss": 0.352,
+      "step": 2362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.018687707641196e-07,
+      "loss": 0.3013,
+      "step": 2363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.0182724252491693e-07,
+      "loss": 0.2939,
+      "step": 2364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.017857142857143e-07,
+      "loss": 0.2893,
+      "step": 2365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.017441860465116e-07,
+      "loss": 0.24,
+      "step": 2366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0170265780730897e-07,
+      "loss": 0.2727,
+      "step": 2367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.016611295681063e-07,
+      "loss": 0.2222,
+      "step": 2368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0161960132890364e-07,
+      "loss": 0.277,
+      "step": 2369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0157807308970098e-07,
+      "loss": 0.2649,
+      "step": 2370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0153654485049834e-07,
+      "loss": 0.2803,
+      "step": 2371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0149501661129568e-07,
+      "loss": 0.2162,
+      "step": 2372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0145348837209302e-07,
+      "loss": 0.2648,
+      "step": 2373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0141196013289035e-07,
+      "loss": 0.2032,
+      "step": 2374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0137043189368772e-07,
+      "loss": 0.2944,
+      "step": 2375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0132890365448503e-07,
+      "loss": 0.304,
+      "step": 2376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.012873754152824e-07,
+      "loss": 0.3529,
+      "step": 2377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0124584717607973e-07,
+      "loss": 0.2798,
+      "step": 2378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.012043189368771e-07,
+      "loss": 0.2976,
+      "step": 2379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.011627906976744e-07,
+      "loss": 0.246,
+      "step": 2380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0112126245847174e-07,
+      "loss": 0.2537,
+      "step": 2381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.010797342192691e-07,
+      "loss": 0.3211,
+      "step": 2382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.010382059800664e-07,
+      "loss": 0.2704,
+      "step": 2383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0099667774086378e-07,
+      "loss": 0.3044,
+      "step": 2384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.009551495016611e-07,
+      "loss": 0.2785,
+      "step": 2385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0091362126245848e-07,
+      "loss": 0.2505,
+      "step": 2386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0087209302325579e-07,
+      "loss": 0.2884,
+      "step": 2387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0083056478405315e-07,
+      "loss": 0.2699,
+      "step": 2388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0078903654485049e-07,
+      "loss": 0.2424,
+      "step": 2389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0074750830564782e-07,
+      "loss": 0.2296,
+      "step": 2390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0070598006644516e-07,
+      "loss": 0.2613,
+      "step": 2391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0066445182724252e-07,
+      "loss": 0.2954,
+      "step": 2392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0062292358803986e-07,
+      "loss": 0.2935,
+      "step": 2393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.005813953488372e-07,
+      "loss": 0.4085,
+      "step": 2394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0053986710963453e-07,
+      "loss": 0.2798,
+      "step": 2395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.004983388704319e-07,
+      "loss": 0.3081,
+      "step": 2396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.004568106312292e-07,
+      "loss": 0.268,
+      "step": 2397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0041528239202657e-07,
+      "loss": 0.2531,
+      "step": 2398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.003737541528239e-07,
+      "loss": 0.3297,
+      "step": 2399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0033222591362127e-07,
+      "loss": 0.2889,
+      "step": 2400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.0029069767441858e-07,
+      "loss": 0.2809,
+      "step": 2401
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0024916943521594e-07,
+      "loss": 0.3348,
+      "step": 2402
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0020764119601328e-07,
+      "loss": 0.3457,
+      "step": 2403
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0016611295681062e-07,
+      "loss": 0.2926,
+      "step": 2404
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0012458471760795e-07,
+      "loss": 0.3197,
+      "step": 2405
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0008305647840532e-07,
+      "loss": 0.2734,
+      "step": 2406
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.0004152823920265e-07,
+      "loss": 0.3166,
+      "step": 2407
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2e-07,
+      "loss": 0.3245,
+      "step": 2408
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9995847176079733e-07,
+      "loss": 0.2623,
+      "step": 2409
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.999169435215947e-07,
+      "loss": 0.2906,
+      "step": 2410
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.99875415282392e-07,
+      "loss": 0.3407,
+      "step": 2411
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9983388704318936e-07,
+      "loss": 0.3483,
+      "step": 2412
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.997923588039867e-07,
+      "loss": 0.2255,
+      "step": 2413
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9975083056478406e-07,
+      "loss": 0.2656,
+      "step": 2414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9970930232558137e-07,
+      "loss": 0.3283,
+      "step": 2415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9966777408637874e-07,
+      "loss": 0.3747,
+      "step": 2416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9962624584717607e-07,
+      "loss": 0.292,
+      "step": 2417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9958471760797339e-07,
+      "loss": 0.2903,
+      "step": 2418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9954318936877075e-07,
+      "loss": 0.3235,
+      "step": 2419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.995016611295681e-07,
+      "loss": 0.2167,
+      "step": 2420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9946013289036545e-07,
+      "loss": 0.2816,
+      "step": 2421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9941860465116276e-07,
+      "loss": 0.2344,
+      "step": 2422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9937707641196012e-07,
+      "loss": 0.3164,
+      "step": 2423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9933554817275749e-07,
+      "loss": 0.24,
+      "step": 2424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929401993355482e-07,
+      "loss": 0.298,
+      "step": 2425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9925249169435213e-07,
+      "loss": 0.2706,
+      "step": 2426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.992109634551495e-07,
+      "loss": 0.2662,
+      "step": 2427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9916943521594686e-07,
+      "loss": 0.2559,
+      "step": 2428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9912790697674417e-07,
+      "loss": 0.279,
+      "step": 2429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.990863787375415e-07,
+      "loss": 0.2036,
+      "step": 2430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9904485049833887e-07,
+      "loss": 0.3039,
+      "step": 2431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.990033222591362e-07,
+      "loss": 0.2754,
+      "step": 2432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9896179401993354e-07,
+      "loss": 0.3299,
+      "step": 2433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9892026578073088e-07,
+      "loss": 0.3017,
+      "step": 2434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9887873754152824e-07,
+      "loss": 0.265,
+      "step": 2435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9883720930232555e-07,
+      "loss": 0.2615,
+      "step": 2436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9879568106312292e-07,
+      "loss": 0.2462,
+      "step": 2437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9875415282392025e-07,
+      "loss": 0.2499,
+      "step": 2438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9871262458471762e-07,
+      "loss": 0.2435,
+      "step": 2439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9867109634551493e-07,
+      "loss": 0.3046,
+      "step": 2440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.986295681063123e-07,
+      "loss": 0.2866,
+      "step": 2441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9858803986710963e-07,
+      "loss": 0.2817,
+      "step": 2442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9854651162790696e-07,
+      "loss": 0.3347,
+      "step": 2443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.985049833887043e-07,
+      "loss": 0.2954,
+      "step": 2444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9846345514950166e-07,
+      "loss": 0.2373,
+      "step": 2445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.98421926910299e-07,
+      "loss": 0.3091,
+      "step": 2446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9838039867109634e-07,
+      "loss": 0.3324,
+      "step": 2447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9833887043189367e-07,
+      "loss": 0.2512,
+      "step": 2448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9829734219269104e-07,
+      "loss": 0.2387,
+      "step": 2449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9825581395348835e-07,
+      "loss": 0.306,
+      "step": 2450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.982142857142857e-07,
+      "loss": 0.3091,
+      "step": 2451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9817275747508305e-07,
+      "loss": 0.2893,
+      "step": 2452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.981312292358804e-07,
+      "loss": 0.3669,
+      "step": 2453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9808970099667772e-07,
+      "loss": 0.3074,
+      "step": 2454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9804817275747508e-07,
+      "loss": 0.2699,
+      "step": 2455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9800664451827242e-07,
+      "loss": 0.361,
+      "step": 2456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9796511627906973e-07,
+      "loss": 0.254,
+      "step": 2457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.979235880398671e-07,
+      "loss": 0.2785,
+      "step": 2458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9788205980066446e-07,
+      "loss": 0.2765,
+      "step": 2459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.978405315614618e-07,
+      "loss": 0.3002,
+      "step": 2460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.977990033222591e-07,
+      "loss": 0.2516,
+      "step": 2461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9775747508305647e-07,
+      "loss": 0.3224,
+      "step": 2462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9771594684385383e-07,
+      "loss": 0.3001,
+      "step": 2463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9767441860465114e-07,
+      "loss": 0.2886,
+      "step": 2464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9763289036544848e-07,
+      "loss": 0.2959,
+      "step": 2465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9759136212624584e-07,
+      "loss": 0.3236,
+      "step": 2466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.975498338870432e-07,
+      "loss": 0.3028,
+      "step": 2467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9750830564784052e-07,
+      "loss": 0.2333,
+      "step": 2468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9746677740863785e-07,
+      "loss": 0.3089,
+      "step": 2469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9742524916943522e-07,
+      "loss": 0.2795,
+      "step": 2470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9738372093023253e-07,
+      "loss": 0.2771,
+      "step": 2471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.973421926910299e-07,
+      "loss": 0.2904,
+      "step": 2472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9730066445182723e-07,
+      "loss": 0.3024,
+      "step": 2473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.972591362126246e-07,
+      "loss": 0.32,
+      "step": 2474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.972176079734219e-07,
+      "loss": 0.3122,
+      "step": 2475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9717607973421926e-07,
+      "loss": 0.297,
+      "step": 2476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.971345514950166e-07,
+      "loss": 0.3087,
+      "step": 2477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9709302325581394e-07,
+      "loss": 0.2868,
+      "step": 2478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9705149501661127e-07,
+      "loss": 0.274,
+      "step": 2479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9700996677740864e-07,
+      "loss": 0.3115,
+      "step": 2480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9696843853820597e-07,
+      "loss": 0.2587,
+      "step": 2481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.969269102990033e-07,
+      "loss": 0.2665,
+      "step": 2482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9688538205980065e-07,
+      "loss": 0.3068,
+      "step": 2483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.96843853820598e-07,
+      "loss": 0.3005,
+      "step": 2484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9680232558139535e-07,
+      "loss": 0.3312,
+      "step": 2485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9676079734219268e-07,
+      "loss": 0.2729,
+      "step": 2486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9671926910299002e-07,
+      "loss": 0.3143,
+      "step": 2487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9667774086378738e-07,
+      "loss": 0.381,
+      "step": 2488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.966362126245847e-07,
+      "loss": 0.3023,
+      "step": 2489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9659468438538206e-07,
+      "loss": 0.2532,
+      "step": 2490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.965531561461794e-07,
+      "loss": 0.2986,
+      "step": 2491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9651162790697676e-07,
+      "loss": 0.3534,
+      "step": 2492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9647009966777407e-07,
+      "loss": 0.2636,
+      "step": 2493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9642857142857143e-07,
+      "loss": 0.2719,
+      "step": 2494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9638704318936877e-07,
+      "loss": 0.3401,
+      "step": 2495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9634551495016608e-07,
+      "loss": 0.284,
+      "step": 2496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9630398671096344e-07,
+      "loss": 0.2944,
+      "step": 2497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.962624584717608e-07,
+      "loss": 0.2474,
+      "step": 2498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9622093023255814e-07,
+      "loss": 0.2906,
+      "step": 2499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9617940199335545e-07,
+      "loss": 0.2699,
+      "step": 2500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9613787375415282e-07,
+      "loss": 0.2335,
+      "step": 2501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9609634551495018e-07,
+      "loss": 0.2977,
+      "step": 2502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.960548172757475e-07,
+      "loss": 0.2873,
+      "step": 2503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9601328903654483e-07,
+      "loss": 0.3171,
+      "step": 2504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.959717607973422e-07,
+      "loss": 0.2748,
+      "step": 2505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9593023255813955e-07,
+      "loss": 0.3185,
+      "step": 2506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9588870431893686e-07,
+      "loss": 0.2948,
+      "step": 2507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.958471760797342e-07,
+      "loss": 0.3177,
+      "step": 2508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9580564784053156e-07,
+      "loss": 0.2733,
+      "step": 2509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9576411960132887e-07,
+      "loss": 0.279,
+      "step": 2510
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9572259136212624e-07,
+      "loss": 0.2706,
+      "step": 2511
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9568106312292357e-07,
+      "loss": 0.3088,
+      "step": 2512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9563953488372094e-07,
+      "loss": 0.3235,
+      "step": 2513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9559800664451825e-07,
+      "loss": 0.2855,
+      "step": 2514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.955564784053156e-07,
+      "loss": 0.3416,
+      "step": 2515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9551495016611295e-07,
+      "loss": 0.2758,
+      "step": 2516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9547342192691028e-07,
+      "loss": 0.2543,
+      "step": 2517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9543189368770762e-07,
+      "loss": 0.2723,
+      "step": 2518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9539036544850498e-07,
+      "loss": 0.2852,
+      "step": 2519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9534883720930232e-07,
+      "loss": 0.3163,
+      "step": 2520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9530730897009966e-07,
+      "loss": 0.3297,
+      "step": 2521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.95265780730897e-07,
+      "loss": 0.2922,
+      "step": 2522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9522425249169436e-07,
+      "loss": 0.2639,
+      "step": 2523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9518272425249167e-07,
+      "loss": 0.2954,
+      "step": 2524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9514119601328903e-07,
+      "loss": 0.2686,
+      "step": 2525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9509966777408637e-07,
+      "loss": 0.2537,
+      "step": 2526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9505813953488373e-07,
+      "loss": 0.2814,
+      "step": 2527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9501661129568104e-07,
+      "loss": 0.2935,
+      "step": 2528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.949750830564784e-07,
+      "loss": 0.3086,
+      "step": 2529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9493355481727574e-07,
+      "loss": 0.3069,
+      "step": 2530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9489202657807308e-07,
+      "loss": 0.3545,
+      "step": 2531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9485049833887041e-07,
+      "loss": 0.3251,
+      "step": 2532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9480897009966778e-07,
+      "loss": 0.2811,
+      "step": 2533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9476744186046511e-07,
+      "loss": 0.2345,
+      "step": 2534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9472591362126245e-07,
+      "loss": 0.2702,
+      "step": 2535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.946843853820598e-07,
+      "loss": 0.2999,
+      "step": 2536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9464285714285715e-07,
+      "loss": 0.2375,
+      "step": 2537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9460132890365446e-07,
+      "loss": 0.3002,
+      "step": 2538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9455980066445182e-07,
+      "loss": 0.2668,
+      "step": 2539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9451827242524916e-07,
+      "loss": 0.2894,
+      "step": 2540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9447674418604652e-07,
+      "loss": 0.3105,
+      "step": 2541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9443521594684383e-07,
+      "loss": 0.3012,
+      "step": 2542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.943936877076412e-07,
+      "loss": 0.3024,
+      "step": 2543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9435215946843854e-07,
+      "loss": 0.3053,
+      "step": 2544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9431063122923585e-07,
+      "loss": 0.2685,
+      "step": 2545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.942691029900332e-07,
+      "loss": 0.2884,
+      "step": 2546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9422757475083055e-07,
+      "loss": 0.2939,
+      "step": 2547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.941860465116279e-07,
+      "loss": 0.2522,
+      "step": 2548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9414451827242522e-07,
+      "loss": 0.2677,
+      "step": 2549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9410299003322258e-07,
+      "loss": 0.2914,
+      "step": 2550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9406146179401992e-07,
+      "loss": 0.2957,
+      "step": 2551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9401993355481728e-07,
+      "loss": 0.2843,
+      "step": 2552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.939784053156146e-07,
+      "loss": 0.3105,
+      "step": 2553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9393687707641196e-07,
+      "loss": 0.2758,
+      "step": 2554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.938953488372093e-07,
+      "loss": 0.2474,
+      "step": 2555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9385382059800663e-07,
+      "loss": 0.3014,
+      "step": 2556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9381229235880397e-07,
+      "loss": 0.2512,
+      "step": 2557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9377076411960133e-07,
+      "loss": 0.2905,
+      "step": 2558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9372923588039867e-07,
+      "loss": 0.2905,
+      "step": 2559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.93687707641196e-07,
+      "loss": 0.2721,
+      "step": 2560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9364617940199334e-07,
+      "loss": 0.2844,
+      "step": 2561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.936046511627907e-07,
+      "loss": 0.3012,
+      "step": 2562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9356312292358801e-07,
+      "loss": 0.2918,
+      "step": 2563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9352159468438538e-07,
+      "loss": 0.2815,
+      "step": 2564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9348006644518271e-07,
+      "loss": 0.2586,
+      "step": 2565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9343853820598008e-07,
+      "loss": 0.2495,
+      "step": 2566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933970099667774e-07,
+      "loss": 0.3217,
+      "step": 2567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9335548172757475e-07,
+      "loss": 0.3418,
+      "step": 2568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.933139534883721e-07,
+      "loss": 0.3453,
+      "step": 2569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9327242524916942e-07,
+      "loss": 0.24,
+      "step": 2570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9323089700996676e-07,
+      "loss": 0.2577,
+      "step": 2571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9318936877076412e-07,
+      "loss": 0.2597,
+      "step": 2572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9314784053156146e-07,
+      "loss": 0.2948,
+      "step": 2573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.931063122923588e-07,
+      "loss": 0.2964,
+      "step": 2574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9306478405315613e-07,
+      "loss": 0.2757,
+      "step": 2575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.930232558139535e-07,
+      "loss": 0.2623,
+      "step": 2576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.929817275747508e-07,
+      "loss": 0.2837,
+      "step": 2577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9294019933554817e-07,
+      "loss": 0.2955,
+      "step": 2578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.928986710963455e-07,
+      "loss": 0.2437,
+      "step": 2579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9285714285714287e-07,
+      "loss": 0.3129,
+      "step": 2580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9281561461794018e-07,
+      "loss": 0.286,
+      "step": 2581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9277408637873754e-07,
+      "loss": 0.2883,
+      "step": 2582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9273255813953488e-07,
+      "loss": 0.3278,
+      "step": 2583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.926910299003322e-07,
+      "loss": 0.2292,
+      "step": 2584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9264950166112956e-07,
+      "loss": 0.3173,
+      "step": 2585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9260797342192692e-07,
+      "loss": 0.2713,
+      "step": 2586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9256644518272426e-07,
+      "loss": 0.2637,
+      "step": 2587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9252491694352157e-07,
+      "loss": 0.2586,
+      "step": 2588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9248338870431893e-07,
+      "loss": 0.2646,
+      "step": 2589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924418604651163e-07,
+      "loss": 0.2929,
+      "step": 2590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924003322259136e-07,
+      "loss": 0.2948,
+      "step": 2591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9235880398671094e-07,
+      "loss": 0.2759,
+      "step": 2592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.923172757475083e-07,
+      "loss": 0.3119,
+      "step": 2593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9227574750830567e-07,
+      "loss": 0.2837,
+      "step": 2594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9223421926910298e-07,
+      "loss": 0.2493,
+      "step": 2595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.921926910299003e-07,
+      "loss": 0.2527,
+      "step": 2596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9215116279069768e-07,
+      "loss": 0.3131,
+      "step": 2597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9210963455149499e-07,
+      "loss": 0.3042,
+      "step": 2598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9206810631229235e-07,
+      "loss": 0.2839,
+      "step": 2599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9202657807308969e-07,
+      "loss": 0.3044,
+      "step": 2600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9198504983388705e-07,
+      "loss": 0.3193,
+      "step": 2601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9194352159468436e-07,
+      "loss": 0.3362,
+      "step": 2602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9190199335548172e-07,
+      "loss": 0.2707,
+      "step": 2603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9186046511627906e-07,
+      "loss": 0.3201,
+      "step": 2604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.918189368770764e-07,
+      "loss": 0.2703,
+      "step": 2605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9177740863787373e-07,
+      "loss": 0.29,
+      "step": 2606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.917358803986711e-07,
+      "loss": 0.2044,
+      "step": 2607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9169435215946843e-07,
+      "loss": 0.3077,
+      "step": 2608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9165282392026577e-07,
+      "loss": 0.2828,
+      "step": 2609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.916112956810631e-07,
+      "loss": 0.2995,
+      "step": 2610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9156976744186047e-07,
+      "loss": 0.2738,
+      "step": 2611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9152823920265778e-07,
+      "loss": 0.2815,
+      "step": 2612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9148671096345514e-07,
+      "loss": 0.2576,
+      "step": 2613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9144518272425248e-07,
+      "loss": 0.2932,
+      "step": 2614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9140365448504984e-07,
+      "loss": 0.28,
+      "step": 2615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9136212624584715e-07,
+      "loss": 0.32,
+      "step": 2616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9132059800664452e-07,
+      "loss": 0.2609,
+      "step": 2617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9127906976744185e-07,
+      "loss": 0.342,
+      "step": 2618
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9123754152823922e-07,
+      "loss": 0.2961,
+      "step": 2619
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9119601328903653e-07,
+      "loss": 0.3121,
+      "step": 2620
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.911544850498339e-07,
+      "loss": 0.3041,
+      "step": 2621
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9111295681063123e-07,
+      "loss": 0.3128,
+      "step": 2622
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9107142857142854e-07,
+      "loss": 0.3131,
+      "step": 2623
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.910299003322259e-07,
+      "loss": 0.3044,
+      "step": 2624
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9098837209302326e-07,
+      "loss": 0.2717,
+      "step": 2625
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.909468438538206e-07,
+      "loss": 0.3092,
+      "step": 2626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.909053156146179e-07,
+      "loss": 0.2882,
+      "step": 2627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9086378737541528e-07,
+      "loss": 0.2718,
+      "step": 2628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9082225913621264e-07,
+      "loss": 0.2586,
+      "step": 2629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9078073089700995e-07,
+      "loss": 0.3057,
+      "step": 2630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9073920265780729e-07,
+      "loss": 0.2786,
+      "step": 2631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9069767441860465e-07,
+      "loss": 0.312,
+      "step": 2632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.90656146179402e-07,
+      "loss": 0.3373,
+      "step": 2633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9061461794019932e-07,
+      "loss": 0.2944,
+      "step": 2634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9057308970099666e-07,
+      "loss": 0.3056,
+      "step": 2635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9053156146179402e-07,
+      "loss": 0.2829,
+      "step": 2636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9049003322259133e-07,
+      "loss": 0.2902,
+      "step": 2637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.904485049833887e-07,
+      "loss": 0.2709,
+      "step": 2638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9040697674418603e-07,
+      "loss": 0.2873,
+      "step": 2639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.903654485049834e-07,
+      "loss": 0.2579,
+      "step": 2640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.903239202657807e-07,
+      "loss": 0.2841,
+      "step": 2641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9028239202657807e-07,
+      "loss": 0.2712,
+      "step": 2642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.902408637873754e-07,
+      "loss": 0.2856,
+      "step": 2643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9019933554817274e-07,
+      "loss": 0.3053,
+      "step": 2644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9015780730897008e-07,
+      "loss": 0.291,
+      "step": 2645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9011627906976744e-07,
+      "loss": 0.3609,
+      "step": 2646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9007475083056478e-07,
+      "loss": 0.2959,
+      "step": 2647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9003322259136212e-07,
+      "loss": 0.2667,
+      "step": 2648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8999169435215945e-07,
+      "loss": 0.2798,
+      "step": 2649
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8995016611295682e-07,
+      "loss": 0.323,
+      "step": 2650
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8990863787375413e-07,
+      "loss": 0.2067,
+      "step": 2651
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.898671096345515e-07,
+      "loss": 0.2853,
+      "step": 2652
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8982558139534883e-07,
+      "loss": 0.2536,
+      "step": 2653
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.897840531561462e-07,
+      "loss": 0.2607,
+      "step": 2654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.897425249169435e-07,
+      "loss": 0.2992,
+      "step": 2655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8970099667774086e-07,
+      "loss": 0.3233,
+      "step": 2656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.896594684385382e-07,
+      "loss": 0.2233,
+      "step": 2657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8961794019933554e-07,
+      "loss": 0.299,
+      "step": 2658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8957641196013287e-07,
+      "loss": 0.2402,
+      "step": 2659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8953488372093024e-07,
+      "loss": 0.3154,
+      "step": 2660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8949335548172757e-07,
+      "loss": 0.2673,
+      "step": 2661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8945182724252488e-07,
+      "loss": 0.2867,
+      "step": 2662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8941029900332225e-07,
+      "loss": 0.2911,
+      "step": 2663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.893687707641196e-07,
+      "loss": 0.3567,
+      "step": 2664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8932724252491692e-07,
+      "loss": 0.3059,
+      "step": 2665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8928571428571426e-07,
+      "loss": 0.2578,
+      "step": 2666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8924418604651162e-07,
+      "loss": 0.3264,
+      "step": 2667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8920265780730898e-07,
+      "loss": 0.3262,
+      "step": 2668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.891611295681063e-07,
+      "loss": 0.34,
+      "step": 2669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8911960132890363e-07,
+      "loss": 0.3248,
+      "step": 2670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.89078073089701e-07,
+      "loss": 0.287,
+      "step": 2671
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.890365448504983e-07,
+      "loss": 0.2525,
+      "step": 2672
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8899501661129567e-07,
+      "loss": 0.2466,
+      "step": 2673
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.88953488372093e-07,
+      "loss": 0.2657,
+      "step": 2674
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8891196013289037e-07,
+      "loss": 0.3352,
+      "step": 2675
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8887043189368768e-07,
+      "loss": 0.2607,
+      "step": 2676
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8882890365448504e-07,
+      "loss": 0.3297,
+      "step": 2677
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8878737541528238e-07,
+      "loss": 0.3771,
+      "step": 2678
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8874584717607974e-07,
+      "loss": 0.2845,
+      "step": 2679
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8870431893687705e-07,
+      "loss": 0.3061,
+      "step": 2680
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8866279069767442e-07,
+      "loss": 0.3309,
+      "step": 2681
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8862126245847175e-07,
+      "loss": 0.2878,
+      "step": 2682
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.885797342192691e-07,
+      "loss": 0.2654,
+      "step": 2683
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8853820598006643e-07,
+      "loss": 0.2674,
+      "step": 2684
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.884966777408638e-07,
+      "loss": 0.3748,
+      "step": 2685
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8845514950166113e-07,
+      "loss": 0.3132,
+      "step": 2686
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8841362126245846e-07,
+      "loss": 0.2544,
+      "step": 2687
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.883720930232558e-07,
+      "loss": 0.2941,
+      "step": 2688
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8833056478405316e-07,
+      "loss": 0.2768,
+      "step": 2689
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8828903654485047e-07,
+      "loss": 0.3177,
+      "step": 2690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8824750830564784e-07,
+      "loss": 0.3273,
+      "step": 2691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8820598006644517e-07,
+      "loss": 0.2951,
+      "step": 2692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8816445182724254e-07,
+      "loss": 0.3721,
+      "step": 2693
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8812292358803985e-07,
+      "loss": 0.3067,
+      "step": 2694
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.880813953488372e-07,
+      "loss": 0.3141,
+      "step": 2695
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8803986710963455e-07,
+      "loss": 0.2668,
+      "step": 2696
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8799833887043188e-07,
+      "loss": 0.264,
+      "step": 2697
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8795681063122922e-07,
+      "loss": 0.2918,
+      "step": 2698
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8791528239202658e-07,
+      "loss": 0.2588,
+      "step": 2699
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8787375415282392e-07,
+      "loss": 0.3801,
+      "step": 2700
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8783222591362126e-07,
+      "loss": 0.2483,
+      "step": 2701
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.877906976744186e-07,
+      "loss": 0.3018,
+      "step": 2702
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8774916943521596e-07,
+      "loss": 0.3037,
+      "step": 2703
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8770764119601327e-07,
+      "loss": 0.3379,
+      "step": 2704
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8766611295681063e-07,
+      "loss": 0.3089,
+      "step": 2705
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8762458471760797e-07,
+      "loss": 0.2717,
+      "step": 2706
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8758305647840533e-07,
+      "loss": 0.3405,
+      "step": 2707
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8754152823920264e-07,
+      "loss": 0.2742,
+      "step": 2708
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.875e-07,
+      "loss": 0.2675,
+      "step": 2709
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8745847176079734e-07,
+      "loss": 0.2496,
+      "step": 2710
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8741694352159465e-07,
+      "loss": 0.2666,
+      "step": 2711
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8737541528239202e-07,
+      "loss": 0.2912,
+      "step": 2712
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8733388704318935e-07,
+      "loss": 0.3223,
+      "step": 2713
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8729235880398672e-07,
+      "loss": 0.3086,
+      "step": 2714
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8725083056478403e-07,
+      "loss": 0.2538,
+      "step": 2715
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.872093023255814e-07,
+      "loss": 0.2546,
+      "step": 2716
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8716777408637873e-07,
+      "loss": 0.3803,
+      "step": 2717
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8712624584717606e-07,
+      "loss": 0.2639,
+      "step": 2718
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.870847176079734e-07,
+      "loss": 0.218,
+      "step": 2719
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8704318936877076e-07,
+      "loss": 0.2583,
+      "step": 2720
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.870016611295681e-07,
+      "loss": 0.2927,
+      "step": 2721
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8696013289036544e-07,
+      "loss": 0.2627,
+      "step": 2722
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8691860465116277e-07,
+      "loss": 0.2592,
+      "step": 2723
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8687707641196014e-07,
+      "loss": 0.2553,
+      "step": 2724
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8683554817275745e-07,
+      "loss": 0.2767,
+      "step": 2725
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.867940199335548e-07,
+      "loss": 0.2731,
+      "step": 2726
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8675249169435215e-07,
+      "loss": 0.2508,
+      "step": 2727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.867109634551495e-07,
+      "loss": 0.2746,
+      "step": 2728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8666943521594682e-07,
+      "loss": 0.263,
+      "step": 2729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8662790697674418e-07,
+      "loss": 0.2777,
+      "step": 2730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8658637873754152e-07,
+      "loss": 0.2801,
+      "step": 2731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8654485049833886e-07,
+      "loss": 0.2732,
+      "step": 2732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.865033222591362e-07,
+      "loss": 0.2493,
+      "step": 2733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8646179401993356e-07,
+      "loss": 0.2954,
+      "step": 2734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.864202657807309e-07,
+      "loss": 0.2566,
+      "step": 2735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8637873754152823e-07,
+      "loss": 0.3723,
+      "step": 2736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8633720930232557e-07,
+      "loss": 0.282,
+      "step": 2737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8629568106312293e-07,
+      "loss": 0.2733,
+      "step": 2738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8625415282392024e-07,
+      "loss": 0.3014,
+      "step": 2739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.862126245847176e-07,
+      "loss": 0.2955,
+      "step": 2740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8617109634551494e-07,
+      "loss": 0.3044,
+      "step": 2741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.861295681063123e-07,
+      "loss": 0.2888,
+      "step": 2742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8608803986710961e-07,
+      "loss": 0.3094,
+      "step": 2743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8604651162790698e-07,
+      "loss": 0.2932,
+      "step": 2744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8600498338870431e-07,
+      "loss": 0.2689,
+      "step": 2745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8596345514950168e-07,
+      "loss": 0.2731,
+      "step": 2746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.85921926910299e-07,
+      "loss": 0.2884,
+      "step": 2747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8588039867109635e-07,
+      "loss": 0.3697,
+      "step": 2748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.858388704318937e-07,
+      "loss": 0.2549,
+      "step": 2749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.85797342192691e-07,
+      "loss": 0.3512,
+      "step": 2750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8575581395348836e-07,
+      "loss": 0.3083,
+      "step": 2751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8571428571428572e-07,
+      "loss": 0.339,
+      "step": 2752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8567275747508306e-07,
+      "loss": 0.3005,
+      "step": 2753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8563122923588037e-07,
+      "loss": 0.2825,
+      "step": 2754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8558970099667774e-07,
+      "loss": 0.2544,
+      "step": 2755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.855481727574751e-07,
+      "loss": 0.2371,
+      "step": 2756
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.855066445182724e-07,
+      "loss": 0.2808,
+      "step": 2757
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8546511627906975e-07,
+      "loss": 0.2425,
+      "step": 2758
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.854235880398671e-07,
+      "loss": 0.2879,
+      "step": 2759
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8538205980066445e-07,
+      "loss": 0.3294,
+      "step": 2760
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8534053156146178e-07,
+      "loss": 0.2958,
+      "step": 2761
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8529900332225912e-07,
+      "loss": 0.3001,
+      "step": 2762
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8525747508305648e-07,
+      "loss": 0.3228,
+      "step": 2763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.852159468438538e-07,
+      "loss": 0.291,
+      "step": 2764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8517441860465116e-07,
+      "loss": 0.2858,
+      "step": 2765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.851328903654485e-07,
+      "loss": 0.3383,
+      "step": 2766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8509136212624586e-07,
+      "loss": 0.3417,
+      "step": 2767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8504983388704317e-07,
+      "loss": 0.2862,
+      "step": 2768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8500830564784053e-07,
+      "loss": 0.2805,
+      "step": 2769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8496677740863787e-07,
+      "loss": 0.2378,
+      "step": 2770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.849252491694352e-07,
+      "loss": 0.3194,
+      "step": 2771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8488372093023254e-07,
+      "loss": 0.3587,
+      "step": 2772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.848421926910299e-07,
+      "loss": 0.3435,
+      "step": 2773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8480066445182724e-07,
+      "loss": 0.3338,
+      "step": 2774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8475913621262458e-07,
+      "loss": 0.2479,
+      "step": 2775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8471760797342191e-07,
+      "loss": 0.3113,
+      "step": 2776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8467607973421928e-07,
+      "loss": 0.2819,
+      "step": 2777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.846345514950166e-07,
+      "loss": 0.2361,
+      "step": 2778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8459302325581395e-07,
+      "loss": 0.2884,
+      "step": 2779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.845514950166113e-07,
+      "loss": 0.3315,
+      "step": 2780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8450996677740865e-07,
+      "loss": 0.3068,
+      "step": 2781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8446843853820596e-07,
+      "loss": 0.265,
+      "step": 2782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8442691029900332e-07,
+      "loss": 0.2601,
+      "step": 2783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8438538205980066e-07,
+      "loss": 0.3503,
+      "step": 2784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8434385382059797e-07,
+      "loss": 0.2666,
+      "step": 2785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8430232558139533e-07,
+      "loss": 0.344,
+      "step": 2786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.842607973421927e-07,
+      "loss": 0.3303,
+      "step": 2787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8421926910299003e-07,
+      "loss": 0.2882,
+      "step": 2788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8417774086378734e-07,
+      "loss": 0.2491,
+      "step": 2789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.841362126245847e-07,
+      "loss": 0.3052,
+      "step": 2790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8409468438538207e-07,
+      "loss": 0.3487,
+      "step": 2791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8405315614617938e-07,
+      "loss": 0.3001,
+      "step": 2792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8401162790697672e-07,
+      "loss": 0.3405,
+      "step": 2793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8397009966777408e-07,
+      "loss": 0.2887,
+      "step": 2794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8392857142857144e-07,
+      "loss": 0.2715,
+      "step": 2795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8388704318936876e-07,
+      "loss": 0.368,
+      "step": 2796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.838455149501661e-07,
+      "loss": 0.2783,
+      "step": 2797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8380398671096346e-07,
+      "loss": 0.279,
+      "step": 2798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8376245847176077e-07,
+      "loss": 0.3022,
+      "step": 2799
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8372093023255813e-07,
+      "loss": 0.2759,
+      "step": 2800
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8367940199335547e-07,
+      "loss": 0.318,
+      "step": 2801
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8363787375415283e-07,
+      "loss": 0.2825,
+      "step": 2802
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8359634551495014e-07,
+      "loss": 0.2746,
+      "step": 2803
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.835548172757475e-07,
+      "loss": 0.3084,
+      "step": 2804
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8351328903654484e-07,
+      "loss": 0.3109,
+      "step": 2805
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.834717607973422e-07,
+      "loss": 0.2768,
+      "step": 2806
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.834302325581395e-07,
+      "loss": 0.2772,
+      "step": 2807
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8338870431893688e-07,
+      "loss": 0.3073,
+      "step": 2808
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.833471760797342e-07,
+      "loss": 0.3111,
+      "step": 2809
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8330564784053155e-07,
+      "loss": 0.2922,
+      "step": 2810
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8326411960132889e-07,
+      "loss": 0.2707,
+      "step": 2811
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8322259136212625e-07,
+      "loss": 0.2586,
+      "step": 2812
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8318106312292359e-07,
+      "loss": 0.2702,
+      "step": 2813
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8313953488372092e-07,
+      "loss": 0.3037,
+      "step": 2814
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8309800664451826e-07,
+      "loss": 0.2608,
+      "step": 2815
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8305647840531562e-07,
+      "loss": 0.3129,
+      "step": 2816
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8301495016611293e-07,
+      "loss": 0.2737,
+      "step": 2817
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.829734219269103e-07,
+      "loss": 0.2441,
+      "step": 2818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8293189368770763e-07,
+      "loss": 0.2851,
+      "step": 2819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.82890365448505e-07,
+      "loss": 0.2625,
+      "step": 2820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.828488372093023e-07,
+      "loss": 0.3185,
+      "step": 2821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8280730897009967e-07,
+      "loss": 0.2447,
+      "step": 2822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.82765780730897e-07,
+      "loss": 0.2646,
+      "step": 2823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8272425249169432e-07,
+      "loss": 0.355,
+      "step": 2824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8268272425249168e-07,
+      "loss": 0.2141,
+      "step": 2825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8264119601328904e-07,
+      "loss": 0.3392,
+      "step": 2826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8259966777408638e-07,
+      "loss": 0.3154,
+      "step": 2827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.825581395348837e-07,
+      "loss": 0.2952,
+      "step": 2828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8251661129568105e-07,
+      "loss": 0.318,
+      "step": 2829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8247508305647842e-07,
+      "loss": 0.2805,
+      "step": 2830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8243355481727573e-07,
+      "loss": 0.2528,
+      "step": 2831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8239202657807306e-07,
+      "loss": 0.2507,
+      "step": 2832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8235049833887043e-07,
+      "loss": 0.2603,
+      "step": 2833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.823089700996678e-07,
+      "loss": 0.2645,
+      "step": 2834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.822674418604651e-07,
+      "loss": 0.2819,
+      "step": 2835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8222591362126244e-07,
+      "loss": 0.2765,
+      "step": 2836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.821843853820598e-07,
+      "loss": 0.2995,
+      "step": 2837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.821428571428571e-07,
+      "loss": 0.3194,
+      "step": 2838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8210132890365448e-07,
+      "loss": 0.3395,
+      "step": 2839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.820598006644518e-07,
+      "loss": 0.3216,
+      "step": 2840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8201827242524918e-07,
+      "loss": 0.2731,
+      "step": 2841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8197674418604649e-07,
+      "loss": 0.262,
+      "step": 2842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8193521594684385e-07,
+      "loss": 0.3528,
+      "step": 2843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8189368770764119e-07,
+      "loss": 0.2825,
+      "step": 2844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8185215946843852e-07,
+      "loss": 0.27,
+      "step": 2845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8181063122923586e-07,
+      "loss": 0.3011,
+      "step": 2846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8176910299003322e-07,
+      "loss": 0.3087,
+      "step": 2847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8172757475083056e-07,
+      "loss": 0.3,
+      "step": 2848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.816860465116279e-07,
+      "loss": 0.2901,
+      "step": 2849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8164451827242523e-07,
+      "loss": 0.3224,
+      "step": 2850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.816029900332226e-07,
+      "loss": 0.2743,
+      "step": 2851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.815614617940199e-07,
+      "loss": 0.2781,
+      "step": 2852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8151993355481727e-07,
+      "loss": 0.3212,
+      "step": 2853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.814784053156146e-07,
+      "loss": 0.3098,
+      "step": 2854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8143687707641197e-07,
+      "loss": 0.2813,
+      "step": 2855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8139534883720928e-07,
+      "loss": 0.3798,
+      "step": 2856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8135382059800664e-07,
+      "loss": 0.3216,
+      "step": 2857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8131229235880398e-07,
+      "loss": 0.2462,
+      "step": 2858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8127076411960132e-07,
+      "loss": 0.2519,
+      "step": 2859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8122923588039865e-07,
+      "loss": 0.342,
+      "step": 2860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8118770764119602e-07,
+      "loss": 0.2914,
+      "step": 2861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8114617940199335e-07,
+      "loss": 0.2667,
+      "step": 2862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.811046511627907e-07,
+      "loss": 0.2319,
+      "step": 2863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8106312292358803e-07,
+      "loss": 0.3161,
+      "step": 2864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.810215946843854e-07,
+      "loss": 0.3164,
+      "step": 2865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.809800664451827e-07,
+      "loss": 0.2928,
+      "step": 2866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8093853820598006e-07,
+      "loss": 0.2649,
+      "step": 2867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.808970099667774e-07,
+      "loss": 0.277,
+      "step": 2868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8085548172757476e-07,
+      "loss": 0.3492,
+      "step": 2869
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8081395348837207e-07,
+      "loss": 0.2751,
+      "step": 2870
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8077242524916944e-07,
+      "loss": 0.2184,
+      "step": 2871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8073089700996677e-07,
+      "loss": 0.2669,
+      "step": 2872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8068936877076414e-07,
+      "loss": 0.3472,
+      "step": 2873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8064784053156145e-07,
+      "loss": 0.3164,
+      "step": 2874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8060631229235878e-07,
+      "loss": 0.3157,
+      "step": 2875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8056478405315615e-07,
+      "loss": 0.3399,
+      "step": 2876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8052325581395346e-07,
+      "loss": 0.2671,
+      "step": 2877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8048172757475082e-07,
+      "loss": 0.2846,
+      "step": 2878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8044019933554816e-07,
+      "loss": 0.3175,
+      "step": 2879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8039867109634552e-07,
+      "loss": 0.2825,
+      "step": 2880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8035714285714283e-07,
+      "loss": 0.2895,
+      "step": 2881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.803156146179402e-07,
+      "loss": 0.3203,
+      "step": 2882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8027408637873753e-07,
+      "loss": 0.2558,
+      "step": 2883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8023255813953487e-07,
+      "loss": 0.3594,
+      "step": 2884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.801910299003322e-07,
+      "loss": 0.3056,
+      "step": 2885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8014950166112957e-07,
+      "loss": 0.2859,
+      "step": 2886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.801079734219269e-07,
+      "loss": 0.2927,
+      "step": 2887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8006644518272424e-07,
+      "loss": 0.3418,
+      "step": 2888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8002491694352158e-07,
+      "loss": 0.3146,
+      "step": 2889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7998338870431894e-07,
+      "loss": 0.2481,
+      "step": 2890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7994186046511625e-07,
+      "loss": 0.2615,
+      "step": 2891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7990033222591362e-07,
+      "loss": 0.2895,
+      "step": 2892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7985880398671095e-07,
+      "loss": 0.2511,
+      "step": 2893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7981727574750832e-07,
+      "loss": 0.2387,
+      "step": 2894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7977574750830563e-07,
+      "loss": 0.2728,
+      "step": 2895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.79734219269103e-07,
+      "loss": 0.2742,
+      "step": 2896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7969269102990033e-07,
+      "loss": 0.2947,
+      "step": 2897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7965116279069766e-07,
+      "loss": 0.2093,
+      "step": 2898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.79609634551495e-07,
+      "loss": 0.2603,
+      "step": 2899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7956810631229236e-07,
+      "loss": 0.2736,
+      "step": 2900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.795265780730897e-07,
+      "loss": 0.2869,
+      "step": 2901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7948504983388704e-07,
+      "loss": 0.2474,
+      "step": 2902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7944352159468437e-07,
+      "loss": 0.341,
+      "step": 2903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7940199335548174e-07,
+      "loss": 0.287,
+      "step": 2904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7936046511627905e-07,
+      "loss": 0.2769,
+      "step": 2905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.793189368770764e-07,
+      "loss": 0.2637,
+      "step": 2906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7927740863787375e-07,
+      "loss": 0.2528,
+      "step": 2907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792358803986711e-07,
+      "loss": 0.3197,
+      "step": 2908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7919435215946842e-07,
+      "loss": 0.3928,
+      "step": 2909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7915282392026578e-07,
+      "loss": 0.3065,
+      "step": 2910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7911129568106312e-07,
+      "loss": 0.2226,
+      "step": 2911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7906976744186043e-07,
+      "loss": 0.272,
+      "step": 2912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.790282392026578e-07,
+      "loss": 0.3406,
+      "step": 2913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7898671096345516e-07,
+      "loss": 0.2671,
+      "step": 2914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789451827242525e-07,
+      "loss": 0.2904,
+      "step": 2915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.789036544850498e-07,
+      "loss": 0.3418,
+      "step": 2916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7886212624584717e-07,
+      "loss": 0.2671,
+      "step": 2917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7882059800664453e-07,
+      "loss": 0.3472,
+      "step": 2918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7877906976744184e-07,
+      "loss": 0.2908,
+      "step": 2919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7873754152823918e-07,
+      "loss": 0.2981,
+      "step": 2920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7869601328903654e-07,
+      "loss": 0.2882,
+      "step": 2921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.786544850498339e-07,
+      "loss": 0.3462,
+      "step": 2922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7861295681063122e-07,
+      "loss": 0.2923,
+      "step": 2923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7857142857142855e-07,
+      "loss": 0.3122,
+      "step": 2924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7852990033222592e-07,
+      "loss": 0.3086,
+      "step": 2925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7848837209302323e-07,
+      "loss": 0.2636,
+      "step": 2926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784468438538206e-07,
+      "loss": 0.2542,
+      "step": 2927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7840531561461793e-07,
+      "loss": 0.2746,
+      "step": 2928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783637873754153e-07,
+      "loss": 0.3654,
+      "step": 2929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783222591362126e-07,
+      "loss": 0.2693,
+      "step": 2930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7828073089700996e-07,
+      "loss": 0.2993,
+      "step": 2931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782392026578073e-07,
+      "loss": 0.2676,
+      "step": 2932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7819767441860464e-07,
+      "loss": 0.3023,
+      "step": 2933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7815614617940197e-07,
+      "loss": 0.2806,
+      "step": 2934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7811461794019934e-07,
+      "loss": 0.2577,
+      "step": 2935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807308970099667e-07,
+      "loss": 0.2802,
+      "step": 2936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.78031561461794e-07,
+      "loss": 0.3201,
+      "step": 2937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7799003322259135e-07,
+      "loss": 0.3111,
+      "step": 2938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.779485049833887e-07,
+      "loss": 0.3452,
+      "step": 2939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7790697674418605e-07,
+      "loss": 0.23,
+      "step": 2940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7786544850498338e-07,
+      "loss": 0.3275,
+      "step": 2941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7782392026578072e-07,
+      "loss": 0.2886,
+      "step": 2942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7778239202657808e-07,
+      "loss": 0.3435,
+      "step": 2943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.777408637873754e-07,
+      "loss": 0.2577,
+      "step": 2944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7769933554817276e-07,
+      "loss": 0.2432,
+      "step": 2945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.776578073089701e-07,
+      "loss": 0.3579,
+      "step": 2946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7761627906976746e-07,
+      "loss": 0.3141,
+      "step": 2947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7757475083056477e-07,
+      "loss": 0.3514,
+      "step": 2948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7753322259136213e-07,
+      "loss": 0.2766,
+      "step": 2949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7749169435215947e-07,
+      "loss": 0.2596,
+      "step": 2950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7745016611295678e-07,
+      "loss": 0.3319,
+      "step": 2951
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7740863787375414e-07,
+      "loss": 0.3041,
+      "step": 2952
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.773671096345515e-07,
+      "loss": 0.2205,
+      "step": 2953
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7732558139534884e-07,
+      "loss": 0.329,
+      "step": 2954
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7728405315614615e-07,
+      "loss": 0.2786,
+      "step": 2955
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7724252491694351e-07,
+      "loss": 0.3151,
+      "step": 2956
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7720099667774088e-07,
+      "loss": 0.3116,
+      "step": 2957
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.771594684385382e-07,
+      "loss": 0.258,
+      "step": 2958
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7711794019933552e-07,
+      "loss": 0.292,
+      "step": 2959
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.770764119601329e-07,
+      "loss": 0.2713,
+      "step": 2960
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7703488372093025e-07,
+      "loss": 0.2996,
+      "step": 2961
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7699335548172756e-07,
+      "loss": 0.2992,
+      "step": 2962
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.769518272425249e-07,
+      "loss": 0.3003,
+      "step": 2963
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7691029900332226e-07,
+      "loss": 0.2447,
+      "step": 2964
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7686877076411957e-07,
+      "loss": 0.3039,
+      "step": 2965
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7682724252491694e-07,
+      "loss": 0.2652,
+      "step": 2966
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7678571428571427e-07,
+      "loss": 0.23,
+      "step": 2967
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7674418604651164e-07,
+      "loss": 0.2838,
+      "step": 2968
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7670265780730895e-07,
+      "loss": 0.2827,
+      "step": 2969
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.766611295681063e-07,
+      "loss": 0.3275,
+      "step": 2970
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7661960132890365e-07,
+      "loss": 0.3592,
+      "step": 2971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7657807308970098e-07,
+      "loss": 0.283,
+      "step": 2972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7653654485049832e-07,
+      "loss": 0.3087,
+      "step": 2973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7649501661129568e-07,
+      "loss": 0.2976,
+      "step": 2974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7645348837209302e-07,
+      "loss": 0.3425,
+      "step": 2975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7641196013289036e-07,
+      "loss": 0.2508,
+      "step": 2976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.763704318936877e-07,
+      "loss": 0.3065,
+      "step": 2977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7632890365448506e-07,
+      "loss": 0.2266,
+      "step": 2978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7628737541528237e-07,
+      "loss": 0.2533,
+      "step": 2979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7624584717607973e-07,
+      "loss": 0.26,
+      "step": 2980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7620431893687707e-07,
+      "loss": 0.2731,
+      "step": 2981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7616279069767443e-07,
+      "loss": 0.2552,
+      "step": 2982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7612126245847174e-07,
+      "loss": 0.3203,
+      "step": 2983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.760797342192691e-07,
+      "loss": 0.2409,
+      "step": 2984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7603820598006644e-07,
+      "loss": 0.3527,
+      "step": 2985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7599667774086378e-07,
+      "loss": 0.3083,
+      "step": 2986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7595514950166111e-07,
+      "loss": 0.3521,
+      "step": 2987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7591362126245848e-07,
+      "loss": 0.375,
+      "step": 2988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7587209302325581e-07,
+      "loss": 0.2382,
+      "step": 2989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7583056478405312e-07,
+      "loss": 0.3256,
+      "step": 2990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.757890365448505e-07,
+      "loss": 0.2762,
+      "step": 2991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7574750830564785e-07,
+      "loss": 0.2863,
+      "step": 2992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7570598006644516e-07,
+      "loss": 0.2629,
+      "step": 2993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.756644518272425e-07,
+      "loss": 0.381,
+      "step": 2994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7562292358803986e-07,
+      "loss": 0.3093,
+      "step": 2995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7558139534883722e-07,
+      "loss": 0.2842,
+      "step": 2996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7553986710963453e-07,
+      "loss": 0.2554,
+      "step": 2997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7549833887043187e-07,
+      "loss": 0.275,
+      "step": 2998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7545681063122923e-07,
+      "loss": 0.3642,
+      "step": 2999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.754152823920266e-07,
+      "loss": 0.2393,
+      "step": 3000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.753737541528239e-07,
+      "loss": 0.2979,
+      "step": 3001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7533222591362124e-07,
+      "loss": 0.2594,
+      "step": 3002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.752906976744186e-07,
+      "loss": 0.31,
+      "step": 3003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7524916943521592e-07,
+      "loss": 0.2639,
+      "step": 3004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7520764119601328e-07,
+      "loss": 0.2864,
+      "step": 3005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7516611295681062e-07,
+      "loss": 0.3366,
+      "step": 3006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7512458471760798e-07,
+      "loss": 0.2859,
+      "step": 3007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.750830564784053e-07,
+      "loss": 0.3035,
+      "step": 3008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7504152823920266e-07,
+      "loss": 0.3084,
+      "step": 3009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.75e-07,
+      "loss": 0.3566,
+      "step": 3010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7495847176079733e-07,
+      "loss": 0.2573,
+      "step": 3011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7491694352159467e-07,
+      "loss": 0.3122,
+      "step": 3012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7487541528239203e-07,
+      "loss": 0.2921,
+      "step": 3013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7483388704318937e-07,
+      "loss": 0.3069,
+      "step": 3014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.747923588039867e-07,
+      "loss": 0.2999,
+      "step": 3015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7475083056478404e-07,
+      "loss": 0.252,
+      "step": 3016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.747093023255814e-07,
+      "loss": 0.3583,
+      "step": 3017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.746677740863787e-07,
+      "loss": 0.2375,
+      "step": 3018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7462624584717608e-07,
+      "loss": 0.2985,
+      "step": 3019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.745847176079734e-07,
+      "loss": 0.3929,
+      "step": 3020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7454318936877078e-07,
+      "loss": 0.2615,
+      "step": 3021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7450166112956809e-07,
+      "loss": 0.2604,
+      "step": 3022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7446013289036545e-07,
+      "loss": 0.2798,
+      "step": 3023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7441860465116279e-07,
+      "loss": 0.2738,
+      "step": 3024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7437707641196012e-07,
+      "loss": 0.2929,
+      "step": 3025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7433554817275746e-07,
+      "loss": 0.3547,
+      "step": 3026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7429401993355482e-07,
+      "loss": 0.2853,
+      "step": 3027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7425249169435216e-07,
+      "loss": 0.2626,
+      "step": 3028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.742109634551495e-07,
+      "loss": 0.255,
+      "step": 3029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7416943521594683e-07,
+      "loss": 0.2543,
+      "step": 3030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.741279069767442e-07,
+      "loss": 0.2063,
+      "step": 3031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.740863787375415e-07,
+      "loss": 0.3089,
+      "step": 3032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7404485049833887e-07,
+      "loss": 0.2825,
+      "step": 3033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.740033222591362e-07,
+      "loss": 0.2413,
+      "step": 3034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7396179401993357e-07,
+      "loss": 0.331,
+      "step": 3035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7392026578073088e-07,
+      "loss": 0.3154,
+      "step": 3036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7387873754152824e-07,
+      "loss": 0.3139,
+      "step": 3037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7383720930232558e-07,
+      "loss": 0.3266,
+      "step": 3038
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.737956810631229e-07,
+      "loss": 0.2339,
+      "step": 3039
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7375415282392025e-07,
+      "loss": 0.282,
+      "step": 3040
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.737126245847176e-07,
+      "loss": 0.2913,
+      "step": 3041
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7367109634551495e-07,
+      "loss": 0.2876,
+      "step": 3042
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7362956810631226e-07,
+      "loss": 0.2442,
+      "step": 3043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7358803986710963e-07,
+      "loss": 0.3183,
+      "step": 3044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7354651162790696e-07,
+      "loss": 0.2633,
+      "step": 3045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.735049833887043e-07,
+      "loss": 0.2864,
+      "step": 3046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7346345514950164e-07,
+      "loss": 0.2762,
+      "step": 3047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.73421926910299e-07,
+      "loss": 0.2972,
+      "step": 3048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7338039867109634e-07,
+      "loss": 0.352,
+      "step": 3049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7333887043189368e-07,
+      "loss": 0.2936,
+      "step": 3050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.73297342192691e-07,
+      "loss": 0.3605,
+      "step": 3051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.7325581395348838e-07,
+      "loss": 0.2477,
+      "step": 3052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7321428571428569e-07,
+      "loss": 0.2537,
+      "step": 3053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7317275747508305e-07,
+      "loss": 0.2873,
+      "step": 3054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7313122923588039e-07,
+      "loss": 0.2813,
+      "step": 3055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7308970099667775e-07,
+      "loss": 0.2671,
+      "step": 3056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7304817275747506e-07,
+      "loss": 0.272,
+      "step": 3057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7300664451827242e-07,
+      "loss": 0.2333,
+      "step": 3058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7296511627906976e-07,
+      "loss": 0.3103,
+      "step": 3059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.729235880398671e-07,
+      "loss": 0.352,
+      "step": 3060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7288205980066443e-07,
+      "loss": 0.3262,
+      "step": 3061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.728405315614618e-07,
+      "loss": 0.302,
+      "step": 3062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7279900332225913e-07,
+      "loss": 0.2775,
+      "step": 3063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7275747508305647e-07,
+      "loss": 0.3446,
+      "step": 3064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.727159468438538e-07,
+      "loss": 0.3161,
+      "step": 3065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7267441860465117e-07,
+      "loss": 0.2776,
+      "step": 3066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.726328903654485e-07,
+      "loss": 0.2859,
+      "step": 3067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7259136212624584e-07,
+      "loss": 0.3248,
+      "step": 3068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7254983388704318e-07,
+      "loss": 0.2753,
+      "step": 3069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7250830564784054e-07,
+      "loss": 0.2442,
+      "step": 3070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7246677740863785e-07,
+      "loss": 0.2732,
+      "step": 3071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7242524916943522e-07,
+      "loss": 0.2996,
+      "step": 3072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7238372093023255e-07,
+      "loss": 0.2944,
+      "step": 3073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7234219269102992e-07,
+      "loss": 0.2596,
+      "step": 3074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7230066445182723e-07,
+      "loss": 0.2525,
+      "step": 3075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.722591362126246e-07,
+      "loss": 0.2312,
+      "step": 3076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7221760797342193e-07,
+      "loss": 0.2854,
+      "step": 3077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7217607973421924e-07,
+      "loss": 0.3015,
+      "step": 3078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.721345514950166e-07,
+      "loss": 0.2883,
+      "step": 3079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7209302325581396e-07,
+      "loss": 0.3048,
+      "step": 3080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.720514950166113e-07,
+      "loss": 0.3087,
+      "step": 3081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.720099667774086e-07,
+      "loss": 0.3459,
+      "step": 3082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7196843853820597e-07,
+      "loss": 0.3394,
+      "step": 3083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7192691029900334e-07,
+      "loss": 0.2916,
+      "step": 3084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7188538205980065e-07,
+      "loss": 0.344,
+      "step": 3085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7184385382059798e-07,
+      "loss": 0.2997,
+      "step": 3086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7180232558139535e-07,
+      "loss": 0.2948,
+      "step": 3087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.7176079734219268e-07,
+      "loss": 0.2832,
+      "step": 3088
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7171926910299002e-07,
+      "loss": 0.3292,
+      "step": 3089
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7167774086378736e-07,
+      "loss": 0.3556,
+      "step": 3090
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7163621262458472e-07,
+      "loss": 0.3216,
+      "step": 3091
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7159468438538203e-07,
+      "loss": 0.2947,
+      "step": 3092
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.715531561461794e-07,
+      "loss": 0.2619,
+      "step": 3093
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7151162790697673e-07,
+      "loss": 0.2772,
+      "step": 3094
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.714700996677741e-07,
+      "loss": 0.3044,
+      "step": 3095
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.714285714285714e-07,
+      "loss": 0.2978,
+      "step": 3096
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7138704318936877e-07,
+      "loss": 0.3063,
+      "step": 3097
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.713455149501661e-07,
+      "loss": 0.2731,
+      "step": 3098
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7130398671096344e-07,
+      "loss": 0.2965,
+      "step": 3099
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7126245847176078e-07,
+      "loss": 0.2864,
+      "step": 3100
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7122093023255814e-07,
+      "loss": 0.2663,
+      "step": 3101
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7117940199335548e-07,
+      "loss": 0.2974,
+      "step": 3102
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7113787375415282e-07,
+      "loss": 0.2686,
+      "step": 3103
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7109634551495015e-07,
+      "loss": 0.2771,
+      "step": 3104
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7105481727574752e-07,
+      "loss": 0.3154,
+      "step": 3105
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7101328903654483e-07,
+      "loss": 0.304,
+      "step": 3106
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.709717607973422e-07,
+      "loss": 0.2689,
+      "step": 3107
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7093023255813953e-07,
+      "loss": 0.2675,
+      "step": 3108
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.708887043189369e-07,
+      "loss": 0.3057,
+      "step": 3109
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.708471760797342e-07,
+      "loss": 0.2392,
+      "step": 3110
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7080564784053156e-07,
+      "loss": 0.2785,
+      "step": 3111
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.707641196013289e-07,
+      "loss": 0.272,
+      "step": 3112
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.707225913621262e-07,
+      "loss": 0.3154,
+      "step": 3113
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7068106312292357e-07,
+      "loss": 0.293,
+      "step": 3114
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7063953488372094e-07,
+      "loss": 0.303,
+      "step": 3115
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7059800664451827e-07,
+      "loss": 0.2657,
+      "step": 3116
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7055647840531558e-07,
+      "loss": 0.26,
+      "step": 3117
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7051495016611295e-07,
+      "loss": 0.3134,
+      "step": 3118
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.704734219269103e-07,
+      "loss": 0.3137,
+      "step": 3119
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7043189368770762e-07,
+      "loss": 0.3293,
+      "step": 3120
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7039036544850496e-07,
+      "loss": 0.3193,
+      "step": 3121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7034883720930232e-07,
+      "loss": 0.2832,
+      "step": 3122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.7030730897009968e-07,
+      "loss": 0.3049,
+      "step": 3123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.70265780730897e-07,
+      "loss": 0.2637,
+      "step": 3124
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7022425249169433e-07,
+      "loss": 0.2718,
+      "step": 3125
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.701827242524917e-07,
+      "loss": 0.2461,
+      "step": 3126
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.70141196013289e-07,
+      "loss": 0.2515,
+      "step": 3127
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7009966777408637e-07,
+      "loss": 0.2746,
+      "step": 3128
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.700581395348837e-07,
+      "loss": 0.3168,
+      "step": 3129
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.7001661129568107e-07,
+      "loss": 0.3133,
+      "step": 3130
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6997508305647838e-07,
+      "loss": 0.2986,
+      "step": 3131
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6993355481727574e-07,
+      "loss": 0.3227,
+      "step": 3132
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6989202657807308e-07,
+      "loss": 0.3545,
+      "step": 3133
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6985049833887044e-07,
+      "loss": 0.3094,
+      "step": 3134
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6980897009966775e-07,
+      "loss": 0.2694,
+      "step": 3135
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6976744186046512e-07,
+      "loss": 0.2794,
+      "step": 3136
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6972591362126245e-07,
+      "loss": 0.2488,
+      "step": 3137
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.696843853820598e-07,
+      "loss": 0.2662,
+      "step": 3138
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6964285714285713e-07,
+      "loss": 0.2738,
+      "step": 3139
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.696013289036545e-07,
+      "loss": 0.2917,
+      "step": 3140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6955980066445183e-07,
+      "loss": 0.3254,
+      "step": 3141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6951827242524916e-07,
+      "loss": 0.2828,
+      "step": 3142
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.694767441860465e-07,
+      "loss": 0.2739,
+      "step": 3143
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6943521594684386e-07,
+      "loss": 0.2774,
+      "step": 3144
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6939368770764117e-07,
+      "loss": 0.3149,
+      "step": 3145
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6935215946843854e-07,
+      "loss": 0.3068,
+      "step": 3146
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6931063122923587e-07,
+      "loss": 0.2296,
+      "step": 3147
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6926910299003324e-07,
+      "loss": 0.3233,
+      "step": 3148
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6922757475083055e-07,
+      "loss": 0.2701,
+      "step": 3149
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.691860465116279e-07,
+      "loss": 0.2973,
+      "step": 3150
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6914451827242525e-07,
+      "loss": 0.2614,
+      "step": 3151
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6910299003322256e-07,
+      "loss": 0.1955,
+      "step": 3152
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6906146179401992e-07,
+      "loss": 0.2707,
+      "step": 3153
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6901993355481728e-07,
+      "loss": 0.2537,
+      "step": 3154
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6897840531561462e-07,
+      "loss": 0.2911,
+      "step": 3155
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6893687707641193e-07,
+      "loss": 0.3295,
+      "step": 3156
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.688953488372093e-07,
+      "loss": 0.2831,
+      "step": 3157
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6885382059800666e-07,
+      "loss": 0.2648,
+      "step": 3158
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6881229235880397e-07,
+      "loss": 0.3319,
+      "step": 3159
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.687707641196013e-07,
+      "loss": 0.3461,
+      "step": 3160
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6872923588039867e-07,
+      "loss": 0.2861,
+      "step": 3161
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6868770764119603e-07,
+      "loss": 0.2323,
+      "step": 3162
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6864617940199334e-07,
+      "loss": 0.2381,
+      "step": 3163
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6860465116279068e-07,
+      "loss": 0.2643,
+      "step": 3164
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6856312292358804e-07,
+      "loss": 0.2921,
+      "step": 3165
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6852159468438535e-07,
+      "loss": 0.3065,
+      "step": 3166
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6848006644518271e-07,
+      "loss": 0.3732,
+      "step": 3167
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6843853820598005e-07,
+      "loss": 0.2529,
+      "step": 3168
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6839700996677741e-07,
+      "loss": 0.2937,
+      "step": 3169
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6835548172757472e-07,
+      "loss": 0.2608,
+      "step": 3170
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.683139534883721e-07,
+      "loss": 0.3041,
+      "step": 3171
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6827242524916942e-07,
+      "loss": 0.29,
+      "step": 3172
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6823089700996676e-07,
+      "loss": 0.3256,
+      "step": 3173
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.681893687707641e-07,
+      "loss": 0.2705,
+      "step": 3174
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6814784053156146e-07,
+      "loss": 0.3421,
+      "step": 3175
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.681063122923588e-07,
+      "loss": 0.3039,
+      "step": 3176
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6806478405315614e-07,
+      "loss": 0.2937,
+      "step": 3177
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6802325581395347e-07,
+      "loss": 0.3147,
+      "step": 3178
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6798172757475084e-07,
+      "loss": 0.3233,
+      "step": 3179
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6794019933554815e-07,
+      "loss": 0.2375,
+      "step": 3180
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.678986710963455e-07,
+      "loss": 0.2767,
+      "step": 3181
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6785714285714285e-07,
+      "loss": 0.3293,
+      "step": 3182
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.678156146179402e-07,
+      "loss": 0.3145,
+      "step": 3183
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6777408637873752e-07,
+      "loss": 0.2594,
+      "step": 3184
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6773255813953488e-07,
+      "loss": 0.3643,
+      "step": 3185
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6769102990033222e-07,
+      "loss": 0.2799,
+      "step": 3186
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6764950166112956e-07,
+      "loss": 0.3127,
+      "step": 3187
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.676079734219269e-07,
+      "loss": 0.2976,
+      "step": 3188
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6756644518272426e-07,
+      "loss": 0.3058,
+      "step": 3189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.675249169435216e-07,
+      "loss": 0.2901,
+      "step": 3190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6748338870431893e-07,
+      "loss": 0.296,
+      "step": 3191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6744186046511627e-07,
+      "loss": 0.3495,
+      "step": 3192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6740033222591363e-07,
+      "loss": 0.278,
+      "step": 3193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6735880398671097e-07,
+      "loss": 0.3474,
+      "step": 3194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.673172757475083e-07,
+      "loss": 0.3327,
+      "step": 3195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6727574750830564e-07,
+      "loss": 0.2855,
+      "step": 3196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.67234219269103e-07,
+      "loss": 0.2829,
+      "step": 3197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6719269102990031e-07,
+      "loss": 0.2855,
+      "step": 3198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6715116279069768e-07,
+      "loss": 0.3086,
+      "step": 3199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6710963455149501e-07,
+      "loss": 0.2848,
+      "step": 3200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6706810631229238e-07,
+      "loss": 0.2711,
+      "step": 3201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.670265780730897e-07,
+      "loss": 0.2117,
+      "step": 3202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6698504983388702e-07,
+      "loss": 0.3228,
+      "step": 3203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.669435215946844e-07,
+      "loss": 0.2963,
+      "step": 3204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.669019933554817e-07,
+      "loss": 0.318,
+      "step": 3205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6686046511627906e-07,
+      "loss": 0.2783,
+      "step": 3206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.668189368770764e-07,
+      "loss": 0.2105,
+      "step": 3207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6677740863787376e-07,
+      "loss": 0.2461,
+      "step": 3208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6673588039867107e-07,
+      "loss": 0.2586,
+      "step": 3209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6669435215946843e-07,
+      "loss": 0.248,
+      "step": 3210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6665282392026577e-07,
+      "loss": 0.2347,
+      "step": 3211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.666112956810631e-07,
+      "loss": 0.2961,
+      "step": 3212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6656976744186044e-07,
+      "loss": 0.2558,
+      "step": 3213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.665282392026578e-07,
+      "loss": 0.2948,
+      "step": 3214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6648671096345515e-07,
+      "loss": 0.3181,
+      "step": 3215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6644518272425248e-07,
+      "loss": 0.2492,
+      "step": 3216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6640365448504982e-07,
+      "loss": 0.2873,
+      "step": 3217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6636212624584718e-07,
+      "loss": 0.2701,
+      "step": 3218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.663205980066445e-07,
+      "loss": 0.2935,
+      "step": 3219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6627906976744186e-07,
+      "loss": 0.3448,
+      "step": 3220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.662375415282392e-07,
+      "loss": 0.2713,
+      "step": 3221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6619601328903656e-07,
+      "loss": 0.3185,
+      "step": 3222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6615448504983387e-07,
+      "loss": 0.2645,
+      "step": 3223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6611295681063123e-07,
+      "loss": 0.2922,
+      "step": 3224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6607142857142857e-07,
+      "loss": 0.2681,
+      "step": 3225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.660299003322259e-07,
+      "loss": 0.3139,
+      "step": 3226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6598837209302324e-07,
+      "loss": 0.2566,
+      "step": 3227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.659468438538206e-07,
+      "loss": 0.2831,
+      "step": 3228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6590531561461794e-07,
+      "loss": 0.1998,
+      "step": 3229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6586378737541528e-07,
+      "loss": 0.3043,
+      "step": 3230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.658222591362126e-07,
+      "loss": 0.2768,
+      "step": 3231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6578073089700998e-07,
+      "loss": 0.2979,
+      "step": 3232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6573920265780729e-07,
+      "loss": 0.241,
+      "step": 3233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6569767441860465e-07,
+      "loss": 0.3046,
+      "step": 3234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6565614617940199e-07,
+      "loss": 0.375,
+      "step": 3235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6561461794019935e-07,
+      "loss": 0.2511,
+      "step": 3236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6557308970099666e-07,
+      "loss": 0.2544,
+      "step": 3237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6553156146179402e-07,
+      "loss": 0.2634,
+      "step": 3238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6549003322259136e-07,
+      "loss": 0.2486,
+      "step": 3239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6544850498338867e-07,
+      "loss": 0.3107,
+      "step": 3240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6540697674418603e-07,
+      "loss": 0.2734,
+      "step": 3241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.653654485049834e-07,
+      "loss": 0.3333,
+      "step": 3242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6532392026578073e-07,
+      "loss": 0.2932,
+      "step": 3243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6528239202657804e-07,
+      "loss": 0.3166,
+      "step": 3244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.652408637873754e-07,
+      "loss": 0.2729,
+      "step": 3245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6519933554817277e-07,
+      "loss": 0.3241,
+      "step": 3246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6515780730897008e-07,
+      "loss": 0.3147,
+      "step": 3247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6511627906976742e-07,
+      "loss": 0.3022,
+      "step": 3248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6507475083056478e-07,
+      "loss": 0.2448,
+      "step": 3249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6503322259136214e-07,
+      "loss": 0.289,
+      "step": 3250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6499169435215945e-07,
+      "loss": 0.2939,
+      "step": 3251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.649501661129568e-07,
+      "loss": 0.3051,
+      "step": 3252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6490863787375415e-07,
+      "loss": 0.2707,
+      "step": 3253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6486710963455146e-07,
+      "loss": 0.3263,
+      "step": 3254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6482558139534883e-07,
+      "loss": 0.3054,
+      "step": 3255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6478405315614616e-07,
+      "loss": 0.3284,
+      "step": 3256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6474252491694353e-07,
+      "loss": 0.2856,
+      "step": 3257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6470099667774084e-07,
+      "loss": 0.271,
+      "step": 3258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.646594684385382e-07,
+      "loss": 0.3058,
+      "step": 3259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6461794019933554e-07,
+      "loss": 0.2763,
+      "step": 3260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645764119601329e-07,
+      "loss": 0.2439,
+      "step": 3261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645348837209302e-07,
+      "loss": 0.2842,
+      "step": 3262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449335548172758e-07,
+      "loss": 0.2663,
+      "step": 3263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644518272425249e-07,
+      "loss": 0.2655,
+      "step": 3264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441029900332225e-07,
+      "loss": 0.311,
+      "step": 3265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6436877076411959e-07,
+      "loss": 0.2967,
+      "step": 3266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432724252491695e-07,
+      "loss": 0.308,
+      "step": 3267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6428571428571429e-07,
+      "loss": 0.2946,
+      "step": 3268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6424418604651162e-07,
+      "loss": 0.299,
+      "step": 3269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6420265780730896e-07,
+      "loss": 0.264,
+      "step": 3270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6416112956810632e-07,
+      "loss": 0.2926,
+      "step": 3271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6411960132890363e-07,
+      "loss": 0.3094,
+      "step": 3272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.64078073089701e-07,
+      "loss": 0.3339,
+      "step": 3273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6403654485049833e-07,
+      "loss": 0.2594,
+      "step": 3274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.639950166112957e-07,
+      "loss": 0.2878,
+      "step": 3275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63953488372093e-07,
+      "loss": 0.3318,
+      "step": 3276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6391196013289037e-07,
+      "loss": 0.2941,
+      "step": 3277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.638704318936877e-07,
+      "loss": 0.2837,
+      "step": 3278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6382890365448502e-07,
+      "loss": 0.3102,
+      "step": 3279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6378737541528238e-07,
+      "loss": 0.2946,
+      "step": 3280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6374584717607974e-07,
+      "loss": 0.3183,
+      "step": 3281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6370431893687708e-07,
+      "loss": 0.2884,
+      "step": 3282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636627906976744e-07,
+      "loss": 0.2929,
+      "step": 3283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362126245847175e-07,
+      "loss": 0.3236,
+      "step": 3284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357973421926912e-07,
+      "loss": 0.3693,
+      "step": 3285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353820598006643e-07,
+      "loss": 0.3312,
+      "step": 3286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6349667774086376e-07,
+      "loss": 0.3085,
+      "step": 3287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345514950166113e-07,
+      "loss": 0.3155,
+      "step": 3288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634136212624585e-07,
+      "loss": 0.2522,
+      "step": 3289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633720930232558e-07,
+      "loss": 0.3357,
+      "step": 3290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333056478405314e-07,
+      "loss": 0.3018,
+      "step": 3291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632890365448505e-07,
+      "loss": 0.3207,
+      "step": 3292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632475083056478e-07,
+      "loss": 0.2811,
+      "step": 3293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320598006644517e-07,
+      "loss": 0.3051,
+      "step": 3294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631644518272425e-07,
+      "loss": 0.2896,
+      "step": 3295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6312292358803987e-07,
+      "loss": 0.2722,
+      "step": 3296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308139534883718e-07,
+      "loss": 0.2485,
+      "step": 3297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6303986710963455e-07,
+      "loss": 0.2643,
+      "step": 3298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6299833887043189e-07,
+      "loss": 0.2808,
+      "step": 3299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295681063122922e-07,
+      "loss": 0.3305,
+      "step": 3300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291528239202656e-07,
+      "loss": 0.3241,
+      "step": 3301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6287375415282392e-07,
+      "loss": 0.3126,
+      "step": 3302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6283222591362126e-07,
+      "loss": 0.2919,
+      "step": 3303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.627906976744186e-07,
+      "loss": 0.3052,
+      "step": 3304
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274916943521593e-07,
+      "loss": 0.2693,
+      "step": 3305
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627076411960133e-07,
+      "loss": 0.2875,
+      "step": 3306
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626661129568106e-07,
+      "loss": 0.2962,
+      "step": 3307
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262458471760797e-07,
+      "loss": 0.3093,
+      "step": 3308
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625830564784053e-07,
+      "loss": 0.264,
+      "step": 3309
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6254152823920267e-07,
+      "loss": 0.3132,
+      "step": 3310
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249999999999998e-07,
+      "loss": 0.2933,
+      "step": 3311
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245847176079734e-07,
+      "loss": 0.3063,
+      "step": 3312
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241694352159468e-07,
+      "loss": 0.3516,
+      "step": 3313
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237541528239202e-07,
+      "loss": 0.2786,
+      "step": 3314
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233388704318935e-07,
+      "loss": 0.3701,
+      "step": 3315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229235880398672e-07,
+      "loss": 0.248,
+      "step": 3316
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225083056478405e-07,
+      "loss": 0.2775,
+      "step": 3317
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6220930232558136e-07,
+      "loss": 0.2987,
+      "step": 3318
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216777408637873e-07,
+      "loss": 0.2839,
+      "step": 3319
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621262458471761e-07,
+      "loss": 0.335,
+      "step": 3320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620847176079734e-07,
+      "loss": 0.2908,
+      "step": 3321
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6204318936877074e-07,
+      "loss": 0.3026,
+      "step": 3322
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620016611295681e-07,
+      "loss": 0.3112,
+      "step": 3323
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6196013289036546e-07,
+      "loss": 0.2745,
+      "step": 3324
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6191860465116277e-07,
+      "loss": 0.2755,
+      "step": 3325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.618770764119601e-07,
+      "loss": 0.2899,
+      "step": 3326
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6183554817275747e-07,
+      "loss": 0.3221,
+      "step": 3327
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6179401993355484e-07,
+      "loss": 0.3332,
+      "step": 3328
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6175249169435215e-07,
+      "loss": 0.3033,
+      "step": 3329
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6171096345514948e-07,
+      "loss": 0.3643,
+      "step": 3330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6166943521594685e-07,
+      "loss": 0.3151,
+      "step": 3331
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6162790697674416e-07,
+      "loss": 0.3581,
+      "step": 3332
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6158637873754152e-07,
+      "loss": 0.3153,
+      "step": 3333
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6154485049833886e-07,
+      "loss": 0.2664,
+      "step": 3334
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6150332225913622e-07,
+      "loss": 0.2755,
+      "step": 3335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6146179401993353e-07,
+      "loss": 0.3003,
+      "step": 3336
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.614202657807309e-07,
+      "loss": 0.294,
+      "step": 3337
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6137873754152823e-07,
+      "loss": 0.2629,
+      "step": 3338
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6133720930232557e-07,
+      "loss": 0.2214,
+      "step": 3339
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.612956810631229e-07,
+      "loss": 0.2485,
+      "step": 3340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6125415282392027e-07,
+      "loss": 0.3187,
+      "step": 3341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612126245847176e-07,
+      "loss": 0.2913,
+      "step": 3342
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6117109634551494e-07,
+      "loss": 0.296,
+      "step": 3343
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6112956810631228e-07,
+      "loss": 0.2797,
+      "step": 3344
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6108803986710964e-07,
+      "loss": 0.2903,
+      "step": 3345
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6104651162790695e-07,
+      "loss": 0.2734,
+      "step": 3346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6100498338870432e-07,
+      "loss": 0.2433,
+      "step": 3347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6096345514950165e-07,
+      "loss": 0.2867,
+      "step": 3348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6092192691029902e-07,
+      "loss": 0.2357,
+      "step": 3349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6088039867109633e-07,
+      "loss": 0.3198,
+      "step": 3350
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.608388704318937e-07,
+      "loss": 0.285,
+      "step": 3351
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6079734219269103e-07,
+      "loss": 0.3075,
+      "step": 3352
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6075581395348836e-07,
+      "loss": 0.2735,
+      "step": 3353
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.607142857142857e-07,
+      "loss": 0.3083,
+      "step": 3354
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6067275747508306e-07,
+      "loss": 0.2681,
+      "step": 3355
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.606312292358804e-07,
+      "loss": 0.2957,
+      "step": 3356
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6058970099667774e-07,
+      "loss": 0.2719,
+      "step": 3357
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6054817275747507e-07,
+      "loss": 0.3063,
+      "step": 3358
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6050664451827244e-07,
+      "loss": 0.2906,
+      "step": 3359
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6046511627906975e-07,
+      "loss": 0.316,
+      "step": 3360
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.604235880398671e-07,
+      "loss": 0.2946,
+      "step": 3361
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6038205980066445e-07,
+      "loss": 0.2741,
+      "step": 3362
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.603405315614618e-07,
+      "loss": 0.3166,
+      "step": 3363
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6029900332225912e-07,
+      "loss": 0.229,
+      "step": 3364
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6025747508305648e-07,
+      "loss": 0.3093,
+      "step": 3365
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6021594684385382e-07,
+      "loss": 0.2937,
+      "step": 3366
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6017441860465113e-07,
+      "loss": 0.3334,
+      "step": 3367
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.601328903654485e-07,
+      "loss": 0.3264,
+      "step": 3368
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6009136212624583e-07,
+      "loss": 0.2732,
+      "step": 3369
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.600498338870432e-07,
+      "loss": 0.3019,
+      "step": 3370
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.600083056478405e-07,
+      "loss": 0.305,
+      "step": 3371
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5996677740863787e-07,
+      "loss": 0.2027,
+      "step": 3372
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.599252491694352e-07,
+      "loss": 0.2177,
+      "step": 3373
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5988372093023254e-07,
+      "loss": 0.2372,
+      "step": 3374
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5984219269102988e-07,
+      "loss": 0.3122,
+      "step": 3375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5980066445182724e-07,
+      "loss": 0.2614,
+      "step": 3376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5975913621262458e-07,
+      "loss": 0.2932,
+      "step": 3377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5971760797342191e-07,
+      "loss": 0.3163,
+      "step": 3378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5967607973421925e-07,
+      "loss": 0.2634,
+      "step": 3379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5963455149501661e-07,
+      "loss": 0.2636,
+      "step": 3380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5959302325581393e-07,
+      "loss": 0.375,
+      "step": 3381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595514950166113e-07,
+      "loss": 0.306,
+      "step": 3382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5950996677740863e-07,
+      "loss": 0.2879,
+      "step": 3383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.59468438538206e-07,
+      "loss": 0.2241,
+      "step": 3384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.594269102990033e-07,
+      "loss": 0.2899,
+      "step": 3385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5938538205980066e-07,
+      "loss": 0.3636,
+      "step": 3386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.59343853820598e-07,
+      "loss": 0.3043,
+      "step": 3387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5930232558139536e-07,
+      "loss": 0.2649,
+      "step": 3388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5926079734219267e-07,
+      "loss": 0.346,
+      "step": 3389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5921926910299004e-07,
+      "loss": 0.2888,
+      "step": 3390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5917774086378737e-07,
+      "loss": 0.3243,
+      "step": 3391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.591362126245847e-07,
+      "loss": 0.3202,
+      "step": 3392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5909468438538205e-07,
+      "loss": 0.2844,
+      "step": 3393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.590531561461794e-07,
+      "loss": 0.2744,
+      "step": 3394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5901162790697675e-07,
+      "loss": 0.3046,
+      "step": 3395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5897009966777408e-07,
+      "loss": 0.2832,
+      "step": 3396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5892857142857142e-07,
+      "loss": 0.2878,
+      "step": 3397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5888704318936878e-07,
+      "loss": 0.2956,
+      "step": 3398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.588455149501661e-07,
+      "loss": 0.2928,
+      "step": 3399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5880398671096346e-07,
+      "loss": 0.325,
+      "step": 3400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.587624584717608e-07,
+      "loss": 0.226,
+      "step": 3401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5872093023255816e-07,
+      "loss": 0.2489,
+      "step": 3402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5867940199335547e-07,
+      "loss": 0.2299,
+      "step": 3403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5863787375415283e-07,
+      "loss": 0.2986,
+      "step": 3404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5859634551495017e-07,
+      "loss": 0.2546,
+      "step": 3405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5855481727574748e-07,
+      "loss": 0.2882,
+      "step": 3406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5851328903654484e-07,
+      "loss": 0.262,
+      "step": 3407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.584717607973422e-07,
+      "loss": 0.2421,
+      "step": 3408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5843023255813954e-07,
+      "loss": 0.3105,
+      "step": 3409
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5838870431893685e-07,
+      "loss": 0.3153,
+      "step": 3410
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5834717607973421e-07,
+      "loss": 0.3446,
+      "step": 3411
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5830564784053158e-07,
+      "loss": 0.2635,
+      "step": 3412
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.582641196013289e-07,
+      "loss": 0.2312,
+      "step": 3413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5822259136212622e-07,
+      "loss": 0.2746,
+      "step": 3414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.581810631229236e-07,
+      "loss": 0.3547,
+      "step": 3415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5813953488372092e-07,
+      "loss": 0.2929,
+      "step": 3416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5809800664451826e-07,
+      "loss": 0.2342,
+      "step": 3417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.580564784053156e-07,
+      "loss": 0.2982,
+      "step": 3418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5801495016611296e-07,
+      "loss": 0.2692,
+      "step": 3419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5797342192691027e-07,
+      "loss": 0.2435,
+      "step": 3420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5793189368770763e-07,
+      "loss": 0.3014,
+      "step": 3421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5789036544850497e-07,
+      "loss": 0.2562,
+      "step": 3422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5784883720930233e-07,
+      "loss": 0.2868,
+      "step": 3423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5780730897009965e-07,
+      "loss": 0.2592,
+      "step": 3424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.57765780730897e-07,
+      "loss": 0.2615,
+      "step": 3425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5772425249169435e-07,
+      "loss": 0.2528,
+      "step": 3426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5768272425249168e-07,
+      "loss": 0.2465,
+      "step": 3427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5764119601328902e-07,
+      "loss": 0.2267,
+      "step": 3428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5759966777408638e-07,
+      "loss": 0.2647,
+      "step": 3429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5755813953488372e-07,
+      "loss": 0.2315,
+      "step": 3430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5751661129568106e-07,
+      "loss": 0.3289,
+      "step": 3431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.574750830564784e-07,
+      "loss": 0.2577,
+      "step": 3432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5743355481727576e-07,
+      "loss": 0.2559,
+      "step": 3433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5739202657807307e-07,
+      "loss": 0.2933,
+      "step": 3434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5735049833887043e-07,
+      "loss": 0.2979,
+      "step": 3435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5730897009966777e-07,
+      "loss": 0.2116,
+      "step": 3436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5726744186046513e-07,
+      "loss": 0.3331,
+      "step": 3437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5722591362126244e-07,
+      "loss": 0.2937,
+      "step": 3438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.571843853820598e-07,
+      "loss": 0.2873,
+      "step": 3439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5714285714285714e-07,
+      "loss": 0.2756,
+      "step": 3440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5710132890365445e-07,
+      "loss": 0.3076,
+      "step": 3441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.570598006644518e-07,
+      "loss": 0.27,
+      "step": 3442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5701827242524918e-07,
+      "loss": 0.384,
+      "step": 3443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.569767441860465e-07,
+      "loss": 0.3098,
+      "step": 3444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5693521594684382e-07,
+      "loss": 0.2941,
+      "step": 3445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5689368770764119e-07,
+      "loss": 0.2884,
+      "step": 3446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5685215946843855e-07,
+      "loss": 0.2746,
+      "step": 3447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5681063122923586e-07,
+      "loss": 0.301,
+      "step": 3448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.567691029900332e-07,
+      "loss": 0.286,
+      "step": 3449
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5672757475083056e-07,
+      "loss": 0.3515,
+      "step": 3450
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5668604651162792e-07,
+      "loss": 0.258,
+      "step": 3451
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5664451827242523e-07,
+      "loss": 0.2416,
+      "step": 3452
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5660299003322257e-07,
+      "loss": 0.2724,
+      "step": 3453
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5656146179401993e-07,
+      "loss": 0.2683,
+      "step": 3454
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.565199335548173e-07,
+      "loss": 0.2722,
+      "step": 3455
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.564784053156146e-07,
+      "loss": 0.2665,
+      "step": 3456
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5643687707641194e-07,
+      "loss": 0.3114,
+      "step": 3457
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.563953488372093e-07,
+      "loss": 0.3451,
+      "step": 3458
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5635382059800662e-07,
+      "loss": 0.3251,
+      "step": 3459
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5631229235880398e-07,
+      "loss": 0.3377,
+      "step": 3460
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5627076411960132e-07,
+      "loss": 0.2587,
+      "step": 3461
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5622923588039868e-07,
+      "loss": 0.393,
+      "step": 3462
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.56187707641196e-07,
+      "loss": 0.2847,
+      "step": 3463
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5614617940199335e-07,
+      "loss": 0.2871,
+      "step": 3464
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.561046511627907e-07,
+      "loss": 0.2626,
+      "step": 3465
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5606312292358803e-07,
+      "loss": 0.2918,
+      "step": 3466
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5602159468438537e-07,
+      "loss": 0.2786,
+      "step": 3467
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5598006644518273e-07,
+      "loss": 0.3178,
+      "step": 3468
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5593853820598007e-07,
+      "loss": 0.2873,
+      "step": 3469
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.558970099667774e-07,
+      "loss": 0.3162,
+      "step": 3470
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5585548172757474e-07,
+      "loss": 0.2382,
+      "step": 3471
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.558139534883721e-07,
+      "loss": 0.3043,
+      "step": 3472
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.557724252491694e-07,
+      "loss": 0.2641,
+      "step": 3473
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5573089700996678e-07,
+      "loss": 0.3583,
+      "step": 3474
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.556893687707641e-07,
+      "loss": 0.3013,
+      "step": 3475
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5564784053156148e-07,
+      "loss": 0.2774,
+      "step": 3476
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5560631229235879e-07,
+      "loss": 0.4204,
+      "step": 3477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5556478405315615e-07,
+      "loss": 0.2867,
+      "step": 3478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5552325581395349e-07,
+      "loss": 0.2545,
+      "step": 3479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.554817275747508e-07,
+      "loss": 0.3199,
+      "step": 3480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5544019933554816e-07,
+      "loss": 0.2664,
+      "step": 3481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5539867109634552e-07,
+      "loss": 0.2991,
+      "step": 3482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5535714285714286e-07,
+      "loss": 0.2928,
+      "step": 3483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5531561461794017e-07,
+      "loss": 0.3477,
+      "step": 3484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5527408637873753e-07,
+      "loss": 0.3317,
+      "step": 3485
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.552325581395349e-07,
+      "loss": 0.3295,
+      "step": 3486
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.551910299003322e-07,
+      "loss": 0.3024,
+      "step": 3487
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5514950166112954e-07,
+      "loss": 0.2621,
+      "step": 3488
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.551079734219269e-07,
+      "loss": 0.2918,
+      "step": 3489
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5506644518272427e-07,
+      "loss": 0.3037,
+      "step": 3490
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5502491694352158e-07,
+      "loss": 0.3312,
+      "step": 3491
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5498338870431892e-07,
+      "loss": 0.2347,
+      "step": 3492
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5494186046511628e-07,
+      "loss": 0.267,
+      "step": 3493
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.549003322259136e-07,
+      "loss": 0.3179,
+      "step": 3494
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5485880398671095e-07,
+      "loss": 0.2736,
+      "step": 3495
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.548172757475083e-07,
+      "loss": 0.241,
+      "step": 3496
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5477574750830565e-07,
+      "loss": 0.2611,
+      "step": 3497
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5473421926910296e-07,
+      "loss": 0.2752,
+      "step": 3498
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5469269102990033e-07,
+      "loss": 0.2619,
+      "step": 3499
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5465116279069766e-07,
+      "loss": 0.2827,
+      "step": 3500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.54609634551495e-07,
+      "loss": 0.2999,
+      "step": 3501
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5456810631229234e-07,
+      "loss": 0.3001,
+      "step": 3502
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.545265780730897e-07,
+      "loss": 0.2998,
+      "step": 3503
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5448504983388704e-07,
+      "loss": 0.3256,
+      "step": 3504
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5444352159468437e-07,
+      "loss": 0.2833,
+      "step": 3505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.544019933554817e-07,
+      "loss": 0.2964,
+      "step": 3506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5436046511627907e-07,
+      "loss": 0.284,
+      "step": 3507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5431893687707639e-07,
+      "loss": 0.3376,
+      "step": 3508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5427740863787375e-07,
+      "loss": 0.2571,
+      "step": 3509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5423588039867109e-07,
+      "loss": 0.3161,
+      "step": 3510
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5419435215946845e-07,
+      "loss": 0.2751,
+      "step": 3511
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5415282392026576e-07,
+      "loss": 0.2936,
+      "step": 3512
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5411129568106312e-07,
+      "loss": 0.3273,
+      "step": 3513
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5406976744186046e-07,
+      "loss": 0.3058,
+      "step": 3514
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.540282392026578e-07,
+      "loss": 0.325,
+      "step": 3515
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5398671096345513e-07,
+      "loss": 0.2869,
+      "step": 3516
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.539451827242525e-07,
+      "loss": 0.3537,
+      "step": 3517
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5390365448504983e-07,
+      "loss": 0.2743,
+      "step": 3518
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5386212624584717e-07,
+      "loss": 0.2755,
+      "step": 3519
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.538205980066445e-07,
+      "loss": 0.2989,
+      "step": 3520
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5377906976744187e-07,
+      "loss": 0.2959,
+      "step": 3521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.537375415282392e-07,
+      "loss": 0.2745,
+      "step": 3522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5369601328903654e-07,
+      "loss": 0.286,
+      "step": 3523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5365448504983388e-07,
+      "loss": 0.2648,
+      "step": 3524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5361295681063124e-07,
+      "loss": 0.3233,
+      "step": 3525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5357142857142855e-07,
+      "loss": 0.2939,
+      "step": 3526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5352990033222592e-07,
+      "loss": 0.2647,
+      "step": 3527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5348837209302325e-07,
+      "loss": 0.3513,
+      "step": 3528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5344684385382062e-07,
+      "loss": 0.2703,
+      "step": 3529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5340531561461793e-07,
+      "loss": 0.3042,
+      "step": 3530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5336378737541526e-07,
+      "loss": 0.3211,
+      "step": 3531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5332225913621263e-07,
+      "loss": 0.2736,
+      "step": 3532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5328073089700994e-07,
+      "loss": 0.292,
+      "step": 3533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.532392026578073e-07,
+      "loss": 0.3255,
+      "step": 3534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5319767441860464e-07,
+      "loss": 0.385,
+      "step": 3535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.53156146179402e-07,
+      "loss": 0.2701,
+      "step": 3536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.531146179401993e-07,
+      "loss": 0.3261,
+      "step": 3537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5307308970099667e-07,
+      "loss": 0.2885,
+      "step": 3538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.53031561461794e-07,
+      "loss": 0.2858,
+      "step": 3539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5299003322259135e-07,
+      "loss": 0.284,
+      "step": 3540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5294850498338868e-07,
+      "loss": 0.3578,
+      "step": 3541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5290697674418605e-07,
+      "loss": 0.2513,
+      "step": 3542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5286544850498338e-07,
+      "loss": 0.3287,
+      "step": 3543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5282392026578072e-07,
+      "loss": 0.3783,
+      "step": 3544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5278239202657806e-07,
+      "loss": 0.3206,
+      "step": 3545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5274086378737542e-07,
+      "loss": 0.2526,
+      "step": 3546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5269933554817273e-07,
+      "loss": 0.2703,
+      "step": 3547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.526578073089701e-07,
+      "loss": 0.3309,
+      "step": 3548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5261627906976743e-07,
+      "loss": 0.2609,
+      "step": 3549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.525747508305648e-07,
+      "loss": 0.2665,
+      "step": 3550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.525332225913621e-07,
+      "loss": 0.2784,
+      "step": 3551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5249169435215947e-07,
+      "loss": 0.2897,
+      "step": 3552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.524501661129568e-07,
+      "loss": 0.3003,
+      "step": 3553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5240863787375414e-07,
+      "loss": 0.3394,
+      "step": 3554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5236710963455148e-07,
+      "loss": 0.2864,
+      "step": 3555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5232558139534884e-07,
+      "loss": 0.3099,
+      "step": 3556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5228405315614618e-07,
+      "loss": 0.3007,
+      "step": 3557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5224252491694352e-07,
+      "loss": 0.2812,
+      "step": 3558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5220099667774085e-07,
+      "loss": 0.2737,
+      "step": 3559
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5215946843853822e-07,
+      "loss": 0.2606,
+      "step": 3560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5211794019933553e-07,
+      "loss": 0.2894,
+      "step": 3561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.520764119601329e-07,
+      "loss": 0.2912,
+      "step": 3562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5203488372093023e-07,
+      "loss": 0.2815,
+      "step": 3563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.519933554817276e-07,
+      "loss": 0.2426,
+      "step": 3564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.519518272425249e-07,
+      "loss": 0.3008,
+      "step": 3565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5191029900332226e-07,
+      "loss": 0.2671,
+      "step": 3566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.518687707641196e-07,
+      "loss": 0.2828,
+      "step": 3567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.518272425249169e-07,
+      "loss": 0.2571,
+      "step": 3568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5178571428571427e-07,
+      "loss": 0.3425,
+      "step": 3569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5174418604651164e-07,
+      "loss": 0.2337,
+      "step": 3570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5170265780730897e-07,
+      "loss": 0.2676,
+      "step": 3571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5166112956810628e-07,
+      "loss": 0.2565,
+      "step": 3572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5161960132890365e-07,
+      "loss": 0.2751,
+      "step": 3573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.51578073089701e-07,
+      "loss": 0.279,
+      "step": 3574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5153654485049832e-07,
+      "loss": 0.2839,
+      "step": 3575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5149501661129566e-07,
+      "loss": 0.3075,
+      "step": 3576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5145348837209302e-07,
+      "loss": 0.2951,
+      "step": 3577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5141196013289038e-07,
+      "loss": 0.2735,
+      "step": 3578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.513704318936877e-07,
+      "loss": 0.2435,
+      "step": 3579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5132890365448503e-07,
+      "loss": 0.2482,
+      "step": 3580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.512873754152824e-07,
+      "loss": 0.3698,
+      "step": 3581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5124584717607973e-07,
+      "loss": 0.2802,
+      "step": 3582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5120431893687707e-07,
+      "loss": 0.2978,
+      "step": 3583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.511627906976744e-07,
+      "loss": 0.2742,
+      "step": 3584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5112126245847177e-07,
+      "loss": 0.2823,
+      "step": 3585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5107973421926908e-07,
+      "loss": 0.2463,
+      "step": 3586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5103820598006644e-07,
+      "loss": 0.3378,
+      "step": 3587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5099667774086378e-07,
+      "loss": 0.2422,
+      "step": 3588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5095514950166114e-07,
+      "loss": 0.2554,
+      "step": 3589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5091362126245845e-07,
+      "loss": 0.2437,
+      "step": 3590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5087209302325581e-07,
+      "loss": 0.2338,
+      "step": 3591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5083056478405315e-07,
+      "loss": 0.2749,
+      "step": 3592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.507890365448505e-07,
+      "loss": 0.2651,
+      "step": 3593
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5074750830564783e-07,
+      "loss": 0.3168,
+      "step": 3594
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.507059800664452e-07,
+      "loss": 0.2937,
+      "step": 3595
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5066445182724253e-07,
+      "loss": 0.302,
+      "step": 3596
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5062292358803986e-07,
+      "loss": 0.345,
+      "step": 3597
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.505813953488372e-07,
+      "loss": 0.3118,
+      "step": 3598
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5053986710963456e-07,
+      "loss": 0.3103,
+      "step": 3599
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5049833887043187e-07,
+      "loss": 0.2875,
+      "step": 3600
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5045681063122924e-07,
+      "loss": 0.2962,
+      "step": 3601
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5041528239202657e-07,
+      "loss": 0.2519,
+      "step": 3602
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5037375415282394e-07,
+      "loss": 0.2727,
+      "step": 3603
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5033222591362125e-07,
+      "loss": 0.388,
+      "step": 3604
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.502906976744186e-07,
+      "loss": 0.2804,
+      "step": 3605
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5024916943521595e-07,
+      "loss": 0.3139,
+      "step": 3606
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5020764119601326e-07,
+      "loss": 0.3245,
+      "step": 3607
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5016611295681062e-07,
+      "loss": 0.3482,
+      "step": 3608
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5012458471760798e-07,
+      "loss": 0.2964,
+      "step": 3609
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5008305647840532e-07,
+      "loss": 0.2746,
+      "step": 3610
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5004152823920263e-07,
+      "loss": 0.2991,
+      "step": 3611
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5e-07,
+      "loss": 0.2759,
+      "step": 3612
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.4813360273838043,
+      "eval_runtime": 3090.1552,
+      "eval_samples_per_second": 5.158,
+      "eval_steps_per_second": 0.081,
+      "eval_wer": 0.3922098334888747,
+      "step": 3612
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4995847176079733e-07,
+      "loss": 0.3069,
+      "step": 3613
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4991694352159467e-07,
+      "loss": 0.3021,
+      "step": 3614
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.49875415282392e-07,
+      "loss": 0.2993,
+      "step": 3615
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4983388704318937e-07,
+      "loss": 0.3228,
+      "step": 3616
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.497923588039867e-07,
+      "loss": 0.2863,
+      "step": 3617
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4975083056478404e-07,
+      "loss": 0.2691,
+      "step": 3618
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4970930232558138e-07,
+      "loss": 0.2776,
+      "step": 3619
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4966777408637871e-07,
+      "loss": 0.2849,
+      "step": 3620
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4962624584717608e-07,
+      "loss": 0.2891,
+      "step": 3621
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4958471760797341e-07,
+      "loss": 0.2975,
+      "step": 3622
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4954318936877075e-07,
+      "loss": 0.2805,
+      "step": 3623
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.495016611295681e-07,
+      "loss": 0.3236,
+      "step": 3624
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4946013289036545e-07,
+      "loss": 0.2926,
+      "step": 3625
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.494186046511628e-07,
+      "loss": 0.2506,
+      "step": 3626
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4937707641196012e-07,
+      "loss": 0.3236,
+      "step": 3627
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4933554817275746e-07,
+      "loss": 0.3211,
+      "step": 3628
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.492940199335548e-07,
+      "loss": 0.2639,
+      "step": 3629
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4925249169435216e-07,
+      "loss": 0.2343,
+      "step": 3630
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.492109634551495e-07,
+      "loss": 0.2502,
+      "step": 3631
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4916943521594683e-07,
+      "loss": 0.2736,
+      "step": 3632
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4912790697674417e-07,
+      "loss": 0.2872,
+      "step": 3633
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4908637873754153e-07,
+      "loss": 0.2909,
+      "step": 3634
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4904485049833887e-07,
+      "loss": 0.2625,
+      "step": 3635
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.490033222591362e-07,
+      "loss": 0.2659,
+      "step": 3636
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4896179401993355e-07,
+      "loss": 0.3476,
+      "step": 3637
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4892026578073088e-07,
+      "loss": 0.2723,
+      "step": 3638
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4887873754152825e-07,
+      "loss": 0.2425,
+      "step": 3639
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4883720930232558e-07,
+      "loss": 0.2368,
+      "step": 3640
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4879568106312292e-07,
+      "loss": 0.2825,
+      "step": 3641
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4875415282392026e-07,
+      "loss": 0.2878,
+      "step": 3642
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.487126245847176e-07,
+      "loss": 0.2896,
+      "step": 3643
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4867109634551496e-07,
+      "loss": 0.332,
+      "step": 3644
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.486295681063123e-07,
+      "loss": 0.2392,
+      "step": 3645
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4858803986710963e-07,
+      "loss": 0.3537,
+      "step": 3646
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4854651162790697e-07,
+      "loss": 0.3013,
+      "step": 3647
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4850498338870433e-07,
+      "loss": 0.3097,
+      "step": 3648
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4846345514950167e-07,
+      "loss": 0.3158,
+      "step": 3649
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4842192691029898e-07,
+      "loss": 0.3124,
+      "step": 3650
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4838039867109634e-07,
+      "loss": 0.2422,
+      "step": 3651
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4833887043189368e-07,
+      "loss": 0.2733,
+      "step": 3652
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4829734219269104e-07,
+      "loss": 0.2285,
+      "step": 3653
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4825581395348835e-07,
+      "loss": 0.2369,
+      "step": 3654
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4821428571428571e-07,
+      "loss": 0.3507,
+      "step": 3655
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4817275747508305e-07,
+      "loss": 0.2996,
+      "step": 3656
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.481312292358804e-07,
+      "loss": 0.3195,
+      "step": 3657
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4808970099667772e-07,
+      "loss": 0.2528,
+      "step": 3658
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4804817275747506e-07,
+      "loss": 0.2748,
+      "step": 3659
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4800664451827242e-07,
+      "loss": 0.2502,
+      "step": 3660
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4796511627906976e-07,
+      "loss": 0.2768,
+      "step": 3661
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.479235880398671e-07,
+      "loss": 0.2548,
+      "step": 3662
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4788205980066443e-07,
+      "loss": 0.287,
+      "step": 3663
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.478405315614618e-07,
+      "loss": 0.304,
+      "step": 3664
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4779900332225913e-07,
+      "loss": 0.2825,
+      "step": 3665
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.4775747508305647e-07,
+      "loss": 0.2892,
+      "step": 3666
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.477159468438538e-07,
+      "loss": 0.2966,
+      "step": 3667
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4767441860465114e-07,
+      "loss": 0.2968,
+      "step": 3668
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.476328903654485e-07,
+      "loss": 0.2667,
+      "step": 3669
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4759136212624584e-07,
+      "loss": 0.2423,
+      "step": 3670
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4754983388704318e-07,
+      "loss": 0.32,
+      "step": 3671
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4750830564784052e-07,
+      "loss": 0.2482,
+      "step": 3672
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4746677740863785e-07,
+      "loss": 0.2513,
+      "step": 3673
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4742524916943522e-07,
+      "loss": 0.2325,
+      "step": 3674
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4738372093023255e-07,
+      "loss": 0.282,
+      "step": 3675
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.473421926910299e-07,
+      "loss": 0.2495,
+      "step": 3676
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4730066445182723e-07,
+      "loss": 0.3213,
+      "step": 3677
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.472591362126246e-07,
+      "loss": 0.2587,
+      "step": 3678
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4721760797342193e-07,
+      "loss": 0.3131,
+      "step": 3679
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4717607973421927e-07,
+      "loss": 0.2979,
+      "step": 3680
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.471345514950166e-07,
+      "loss": 0.2837,
+      "step": 3681
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4709302325581394e-07,
+      "loss": 0.2583,
+      "step": 3682
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.470514950166113e-07,
+      "loss": 0.3258,
+      "step": 3683
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4700996677740864e-07,
+      "loss": 0.288,
+      "step": 3684
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4696843853820598e-07,
+      "loss": 0.2956,
+      "step": 3685
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.469269102990033e-07,
+      "loss": 0.281,
+      "step": 3686
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4688538205980065e-07,
+      "loss": 0.2941,
+      "step": 3687
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.46843853820598e-07,
+      "loss": 0.2481,
+      "step": 3688
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4680232558139535e-07,
+      "loss": 0.2988,
+      "step": 3689
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4676079734219269e-07,
+      "loss": 0.2707,
+      "step": 3690
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4671926910299002e-07,
+      "loss": 0.2824,
+      "step": 3691
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4667774086378739e-07,
+      "loss": 0.2436,
+      "step": 3692
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4663621262458472e-07,
+      "loss": 0.2484,
+      "step": 3693
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4659468438538206e-07,
+      "loss": 0.2543,
+      "step": 3694
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.465531561461794e-07,
+      "loss": 0.2703,
+      "step": 3695
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4651162790697673e-07,
+      "loss": 0.3006,
+      "step": 3696
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4647009966777407e-07,
+      "loss": 0.2927,
+      "step": 3697
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.464285714285714e-07,
+      "loss": 0.2358,
+      "step": 3698
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4638704318936877e-07,
+      "loss": 0.2661,
+      "step": 3699
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.463455149501661e-07,
+      "loss": 0.29,
+      "step": 3700
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4630398671096344e-07,
+      "loss": 0.2865,
+      "step": 3701
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.4626245847176078e-07,
+      "loss": 0.2813,
+      "step": 3702
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4622093023255812e-07,
+      "loss": 0.2921,
+      "step": 3703
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4617940199335548e-07,
+      "loss": 0.3454,
+      "step": 3704
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4613787375415282e-07,
+      "loss": 0.3391,
+      "step": 3705
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4609634551495015e-07,
+      "loss": 0.2532,
+      "step": 3706
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.460548172757475e-07,
+      "loss": 0.3017,
+      "step": 3707
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4601328903654485e-07,
+      "loss": 0.2817,
+      "step": 3708
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.459717607973422e-07,
+      "loss": 0.261,
+      "step": 3709
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4593023255813953e-07,
+      "loss": 0.2997,
+      "step": 3710
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4588870431893686e-07,
+      "loss": 0.3477,
+      "step": 3711
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.458471760797342e-07,
+      "loss": 0.3251,
+      "step": 3712
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4580564784053156e-07,
+      "loss": 0.2378,
+      "step": 3713
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.457641196013289e-07,
+      "loss": 0.2814,
+      "step": 3714
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4572259136212624e-07,
+      "loss": 0.2404,
+      "step": 3715
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4568106312292357e-07,
+      "loss": 0.3187,
+      "step": 3716
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.456395348837209e-07,
+      "loss": 0.2988,
+      "step": 3717
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4559800664451827e-07,
+      "loss": 0.2786,
+      "step": 3718
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.455564784053156e-07,
+      "loss": 0.297,
+      "step": 3719
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4551495016611295e-07,
+      "loss": 0.2496,
+      "step": 3720
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4547342192691029e-07,
+      "loss": 0.3778,
+      "step": 3721
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4543189368770765e-07,
+      "loss": 0.2518,
+      "step": 3722
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4539036544850499e-07,
+      "loss": 0.3759,
+      "step": 3723
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4534883720930232e-07,
+      "loss": 0.2656,
+      "step": 3724
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4530730897009966e-07,
+      "loss": 0.3246,
+      "step": 3725
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.45265780730897e-07,
+      "loss": 0.3057,
+      "step": 3726
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4522425249169436e-07,
+      "loss": 0.3419,
+      "step": 3727
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.451827242524917e-07,
+      "loss": 0.3471,
+      "step": 3728
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4514119601328903e-07,
+      "loss": 0.3173,
+      "step": 3729
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4509966777408637e-07,
+      "loss": 0.3036,
+      "step": 3730
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4505813953488373e-07,
+      "loss": 0.2532,
+      "step": 3731
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4501661129568107e-07,
+      "loss": 0.2609,
+      "step": 3732
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4497508305647838e-07,
+      "loss": 0.2168,
+      "step": 3733
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4493355481727574e-07,
+      "loss": 0.2537,
+      "step": 3734
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4489202657807308e-07,
+      "loss": 0.318,
+      "step": 3735
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4485049833887044e-07,
+      "loss": 0.2777,
+      "step": 3736
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4480897009966775e-07,
+      "loss": 0.1942,
+      "step": 3737
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.4476744186046512e-07,
+      "loss": 0.2662,
+      "step": 3738
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4472591362126245e-07,
+      "loss": 0.2378,
+      "step": 3739
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.446843853820598e-07,
+      "loss": 0.3803,
+      "step": 3740
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4464285714285713e-07,
+      "loss": 0.3011,
+      "step": 3741
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4460132890365446e-07,
+      "loss": 0.2294,
+      "step": 3742
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4455980066445183e-07,
+      "loss": 0.3158,
+      "step": 3743
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4451827242524916e-07,
+      "loss": 0.3246,
+      "step": 3744
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.444767441860465e-07,
+      "loss": 0.3165,
+      "step": 3745
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4443521594684384e-07,
+      "loss": 0.2342,
+      "step": 3746
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4439368770764117e-07,
+      "loss": 0.2295,
+      "step": 3747
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4435215946843854e-07,
+      "loss": 0.2331,
+      "step": 3748
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4431063122923587e-07,
+      "loss": 0.3209,
+      "step": 3749
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.442691029900332e-07,
+      "loss": 0.2971,
+      "step": 3750
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4422757475083055e-07,
+      "loss": 0.3287,
+      "step": 3751
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.441860465116279e-07,
+      "loss": 0.2822,
+      "step": 3752
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4414451827242525e-07,
+      "loss": 0.2698,
+      "step": 3753
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4410299003322258e-07,
+      "loss": 0.3376,
+      "step": 3754
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4406146179401992e-07,
+      "loss": 0.3394,
+      "step": 3755
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4401993355481726e-07,
+      "loss": 0.2684,
+      "step": 3756
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4397840531561462e-07,
+      "loss": 0.3308,
+      "step": 3757
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4393687707641196e-07,
+      "loss": 0.3414,
+      "step": 3758
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.438953488372093e-07,
+      "loss": 0.2492,
+      "step": 3759
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4385382059800663e-07,
+      "loss": 0.3006,
+      "step": 3760
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.43812292358804e-07,
+      "loss": 0.2982,
+      "step": 3761
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4377076411960133e-07,
+      "loss": 0.2977,
+      "step": 3762
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4372923588039867e-07,
+      "loss": 0.2674,
+      "step": 3763
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.43687707641196e-07,
+      "loss": 0.3136,
+      "step": 3764
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4364617940199334e-07,
+      "loss": 0.2884,
+      "step": 3765
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.436046511627907e-07,
+      "loss": 0.2945,
+      "step": 3766
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4356312292358804e-07,
+      "loss": 0.2825,
+      "step": 3767
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4352159468438538e-07,
+      "loss": 0.3248,
+      "step": 3768
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4348006644518272e-07,
+      "loss": 0.275,
+      "step": 3769
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4343853820598005e-07,
+      "loss": 0.2871,
+      "step": 3770
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4339700996677742e-07,
+      "loss": 0.3356,
+      "step": 3771
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4335548172757475e-07,
+      "loss": 0.281,
+      "step": 3772
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.433139534883721e-07,
+      "loss": 0.3098,
+      "step": 3773
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.4327242524916943e-07,
+      "loss": 0.3359,
+      "step": 3774
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.432308970099668e-07,
+      "loss": 0.2721,
+      "step": 3775
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4318936877076413e-07,
+      "loss": 0.3333,
+      "step": 3776
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4314784053156144e-07,
+      "loss": 0.2788,
+      "step": 3777
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.431063122923588e-07,
+      "loss": 0.2777,
+      "step": 3778
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4306478405315614e-07,
+      "loss": 0.3002,
+      "step": 3779
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4302325581395347e-07,
+      "loss": 0.3136,
+      "step": 3780
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.429817275747508e-07,
+      "loss": 0.3131,
+      "step": 3781
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4294019933554817e-07,
+      "loss": 0.3386,
+      "step": 3782
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.428986710963455e-07,
+      "loss": 0.2705,
+      "step": 3783
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4285714285714285e-07,
+      "loss": 0.2437,
+      "step": 3784
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4281561461794018e-07,
+      "loss": 0.3004,
+      "step": 3785
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4277408637873752e-07,
+      "loss": 0.3314,
+      "step": 3786
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4273255813953488e-07,
+      "loss": 0.3,
+      "step": 3787
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4269102990033222e-07,
+      "loss": 0.279,
+      "step": 3788
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4264950166112956e-07,
+      "loss": 0.2883,
+      "step": 3789
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.426079734219269e-07,
+      "loss": 0.3007,
+      "step": 3790
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4256644518272426e-07,
+      "loss": 0.2923,
+      "step": 3791
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.425249169435216e-07,
+      "loss": 0.2621,
+      "step": 3792
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4248338870431893e-07,
+      "loss": 0.3463,
+      "step": 3793
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4244186046511627e-07,
+      "loss": 0.3159,
+      "step": 3794
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.424003322259136e-07,
+      "loss": 0.2878,
+      "step": 3795
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4235880398671097e-07,
+      "loss": 0.2849,
+      "step": 3796
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.423172757475083e-07,
+      "loss": 0.2359,
+      "step": 3797
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4227574750830564e-07,
+      "loss": 0.2784,
+      "step": 3798
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4223421926910298e-07,
+      "loss": 0.2835,
+      "step": 3799
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4219269102990031e-07,
+      "loss": 0.2236,
+      "step": 3800
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4215116279069768e-07,
+      "loss": 0.2641,
+      "step": 3801
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4210963455149501e-07,
+      "loss": 0.2558,
+      "step": 3802
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4206810631229235e-07,
+      "loss": 0.248,
+      "step": 3803
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.420265780730897e-07,
+      "loss": 0.256,
+      "step": 3804
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4198504983388705e-07,
+      "loss": 0.2688,
+      "step": 3805
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.419435215946844e-07,
+      "loss": 0.238,
+      "step": 3806
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4190199335548173e-07,
+      "loss": 0.2487,
+      "step": 3807
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4186046511627906e-07,
+      "loss": 0.2706,
+      "step": 3808
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.418189368770764e-07,
+      "loss": 0.3038,
+      "step": 3809
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.4177740863787376e-07,
+      "loss": 0.2717,
+      "step": 3810
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.417358803986711e-07,
+      "loss": 0.2767,
+      "step": 3811
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4169435215946844e-07,
+      "loss": 0.2361,
+      "step": 3812
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4165282392026577e-07,
+      "loss": 0.2701,
+      "step": 3813
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.416112956810631e-07,
+      "loss": 0.3305,
+      "step": 3814
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4156976744186047e-07,
+      "loss": 0.2692,
+      "step": 3815
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4152823920265778e-07,
+      "loss": 0.3114,
+      "step": 3816
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4148671096345515e-07,
+      "loss": 0.2751,
+      "step": 3817
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4144518272425248e-07,
+      "loss": 0.2627,
+      "step": 3818
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4140365448504985e-07,
+      "loss": 0.324,
+      "step": 3819
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4136212624584716e-07,
+      "loss": 0.3198,
+      "step": 3820
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.413205980066445e-07,
+      "loss": 0.2911,
+      "step": 3821
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4127906976744186e-07,
+      "loss": 0.2664,
+      "step": 3822
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.412375415282392e-07,
+      "loss": 0.2613,
+      "step": 3823
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4119601328903653e-07,
+      "loss": 0.3356,
+      "step": 3824
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4115448504983387e-07,
+      "loss": 0.3177,
+      "step": 3825
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4111295681063123e-07,
+      "loss": 0.3179,
+      "step": 3826
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4107142857142857e-07,
+      "loss": 0.3168,
+      "step": 3827
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.410299003322259e-07,
+      "loss": 0.338,
+      "step": 3828
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4098837209302324e-07,
+      "loss": 0.3136,
+      "step": 3829
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4094684385382058e-07,
+      "loss": 0.2709,
+      "step": 3830
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4090531561461794e-07,
+      "loss": 0.2847,
+      "step": 3831
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4086378737541528e-07,
+      "loss": 0.2223,
+      "step": 3832
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4082225913621261e-07,
+      "loss": 0.283,
+      "step": 3833
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4078073089700995e-07,
+      "loss": 0.2708,
+      "step": 3834
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4073920265780731e-07,
+      "loss": 0.2869,
+      "step": 3835
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4069767441860465e-07,
+      "loss": 0.2916,
+      "step": 3836
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.40656146179402e-07,
+      "loss": 0.2711,
+      "step": 3837
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4061461794019932e-07,
+      "loss": 0.2531,
+      "step": 3838
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4057308970099666e-07,
+      "loss": 0.3345,
+      "step": 3839
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4053156146179402e-07,
+      "loss": 0.2168,
+      "step": 3840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4049003322259136e-07,
+      "loss": 0.2564,
+      "step": 3841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.404485049833887e-07,
+      "loss": 0.3133,
+      "step": 3842
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4040697674418603e-07,
+      "loss": 0.2635,
+      "step": 3843
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4036544850498337e-07,
+      "loss": 0.3464,
+      "step": 3844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4032392026578073e-07,
+      "loss": 0.3165,
+      "step": 3845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4028239202657807e-07,
+      "loss": 0.2649,
+      "step": 3846
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.402408637873754e-07,
+      "loss": 0.2279,
+      "step": 3847
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4019933554817275e-07,
+      "loss": 0.316,
+      "step": 3848
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.401578073089701e-07,
+      "loss": 0.2752,
+      "step": 3849
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4011627906976745e-07,
+      "loss": 0.2658,
+      "step": 3850
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4007475083056478e-07,
+      "loss": 0.3067,
+      "step": 3851
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4003322259136212e-07,
+      "loss": 0.2476,
+      "step": 3852
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3999169435215946e-07,
+      "loss": 0.2536,
+      "step": 3853
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3995016611295682e-07,
+      "loss": 0.2583,
+      "step": 3854
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3990863787375416e-07,
+      "loss": 0.2999,
+      "step": 3855
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.398671096345515e-07,
+      "loss": 0.2576,
+      "step": 3856
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3982558139534883e-07,
+      "loss": 0.2334,
+      "step": 3857
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.397840531561462e-07,
+      "loss": 0.2862,
+      "step": 3858
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.397425249169435e-07,
+      "loss": 0.3345,
+      "step": 3859
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3970099667774084e-07,
+      "loss": 0.3179,
+      "step": 3860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.396594684385382e-07,
+      "loss": 0.3015,
+      "step": 3861
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3961794019933554e-07,
+      "loss": 0.2812,
+      "step": 3862
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3957641196013288e-07,
+      "loss": 0.2722,
+      "step": 3863
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3953488372093021e-07,
+      "loss": 0.2616,
+      "step": 3864
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3949335548172758e-07,
+      "loss": 0.2655,
+      "step": 3865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3945182724252491e-07,
+      "loss": 0.2743,
+      "step": 3866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3941029900332225e-07,
+      "loss": 0.2938,
+      "step": 3867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.393687707641196e-07,
+      "loss": 0.2445,
+      "step": 3868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3932724252491692e-07,
+      "loss": 0.2793,
+      "step": 3869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.392857142857143e-07,
+      "loss": 0.302,
+      "step": 3870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3924418604651162e-07,
+      "loss": 0.2536,
+      "step": 3871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3920265780730896e-07,
+      "loss": 0.2735,
+      "step": 3872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.391611295681063e-07,
+      "loss": 0.3555,
+      "step": 3873
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3911960132890363e-07,
+      "loss": 0.2793,
+      "step": 3874
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.39078073089701e-07,
+      "loss": 0.2175,
+      "step": 3875
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3903654485049833e-07,
+      "loss": 0.2743,
+      "step": 3876
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3899501661129567e-07,
+      "loss": 0.2411,
+      "step": 3877
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.38953488372093e-07,
+      "loss": 0.3071,
+      "step": 3878
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3891196013289037e-07,
+      "loss": 0.3014,
+      "step": 3879
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.388704318936877e-07,
+      "loss": 0.2432,
+      "step": 3880
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3882890365448504e-07,
+      "loss": 0.3037,
+      "step": 3881
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.3878737541528238e-07,
+      "loss": 0.2608,
+      "step": 3882
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3874584717607972e-07,
+      "loss": 0.3008,
+      "step": 3883
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3870431893687708e-07,
+      "loss": 0.2378,
+      "step": 3884
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3866279069767442e-07,
+      "loss": 0.3278,
+      "step": 3885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3862126245847175e-07,
+      "loss": 0.301,
+      "step": 3886
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.385797342192691e-07,
+      "loss": 0.2257,
+      "step": 3887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3853820598006646e-07,
+      "loss": 0.2799,
+      "step": 3888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.384966777408638e-07,
+      "loss": 0.2656,
+      "step": 3889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3845514950166113e-07,
+      "loss": 0.2398,
+      "step": 3890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3841362126245847e-07,
+      "loss": 0.2184,
+      "step": 3891
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.383720930232558e-07,
+      "loss": 0.3386,
+      "step": 3892
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3833056478405317e-07,
+      "loss": 0.2995,
+      "step": 3893
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.382890365448505e-07,
+      "loss": 0.2732,
+      "step": 3894
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3824750830564784e-07,
+      "loss": 0.3286,
+      "step": 3895
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3820598006644518e-07,
+      "loss": 0.2371,
+      "step": 3896
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.381644518272425e-07,
+      "loss": 0.2969,
+      "step": 3897
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3812292358803988e-07,
+      "loss": 0.2494,
+      "step": 3898
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3808139534883719e-07,
+      "loss": 0.2873,
+      "step": 3899
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3803986710963455e-07,
+      "loss": 0.2454,
+      "step": 3900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3799833887043189e-07,
+      "loss": 0.2755,
+      "step": 3901
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3795681063122925e-07,
+      "loss": 0.2609,
+      "step": 3902
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3791528239202656e-07,
+      "loss": 0.2589,
+      "step": 3903
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.378737541528239e-07,
+      "loss": 0.2835,
+      "step": 3904
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3783222591362126e-07,
+      "loss": 0.3035,
+      "step": 3905
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.377906976744186e-07,
+      "loss": 0.2557,
+      "step": 3906
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3774916943521593e-07,
+      "loss": 0.2754,
+      "step": 3907
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3770764119601327e-07,
+      "loss": 0.2599,
+      "step": 3908
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3766611295681063e-07,
+      "loss": 0.2915,
+      "step": 3909
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3762458471760797e-07,
+      "loss": 0.2915,
+      "step": 3910
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.375830564784053e-07,
+      "loss": 0.3247,
+      "step": 3911
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3754152823920264e-07,
+      "loss": 0.2524,
+      "step": 3912
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3749999999999998e-07,
+      "loss": 0.2214,
+      "step": 3913
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3745847176079734e-07,
+      "loss": 0.2919,
+      "step": 3914
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3741694352159468e-07,
+      "loss": 0.2545,
+      "step": 3915
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3737541528239202e-07,
+      "loss": 0.2806,
+      "step": 3916
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3733388704318935e-07,
+      "loss": 0.3234,
+      "step": 3917
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.372923588039867e-07,
+      "loss": 0.3588,
+      "step": 3918
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.3725083056478405e-07,
+      "loss": 0.344,
+      "step": 3919
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.372093023255814e-07,
+      "loss": 0.3063,
+      "step": 3920
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3716777408637873e-07,
+      "loss": 0.2309,
+      "step": 3921
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3712624584717606e-07,
+      "loss": 0.2865,
+      "step": 3922
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3708471760797343e-07,
+      "loss": 0.2954,
+      "step": 3923
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3704318936877076e-07,
+      "loss": 0.267,
+      "step": 3924
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.370016611295681e-07,
+      "loss": 0.246,
+      "step": 3925
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3696013289036544e-07,
+      "loss": 0.2376,
+      "step": 3926
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3691860465116277e-07,
+      "loss": 0.2964,
+      "step": 3927
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3687707641196014e-07,
+      "loss": 0.263,
+      "step": 3928
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3683554817275748e-07,
+      "loss": 0.2926,
+      "step": 3929
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.367940199335548e-07,
+      "loss": 0.2775,
+      "step": 3930
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3675249169435215e-07,
+      "loss": 0.2527,
+      "step": 3931
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.367109634551495e-07,
+      "loss": 0.2881,
+      "step": 3932
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3666943521594685e-07,
+      "loss": 0.3697,
+      "step": 3933
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3662790697674419e-07,
+      "loss": 0.3318,
+      "step": 3934
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3658637873754152e-07,
+      "loss": 0.3228,
+      "step": 3935
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3654485049833886e-07,
+      "loss": 0.2969,
+      "step": 3936
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3650332225913622e-07,
+      "loss": 0.2613,
+      "step": 3937
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3646179401993356e-07,
+      "loss": 0.279,
+      "step": 3938
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.364202657807309e-07,
+      "loss": 0.3187,
+      "step": 3939
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3637873754152823e-07,
+      "loss": 0.2575,
+      "step": 3940
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3633720930232557e-07,
+      "loss": 0.2694,
+      "step": 3941
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.362956810631229e-07,
+      "loss": 0.2486,
+      "step": 3942
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3625415282392024e-07,
+      "loss": 0.2847,
+      "step": 3943
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.362126245847176e-07,
+      "loss": 0.2808,
+      "step": 3944
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3617109634551494e-07,
+      "loss": 0.2142,
+      "step": 3945
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3612956810631228e-07,
+      "loss": 0.3212,
+      "step": 3946
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3608803986710962e-07,
+      "loss": 0.2502,
+      "step": 3947
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3604651162790695e-07,
+      "loss": 0.2953,
+      "step": 3948
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3600498338870432e-07,
+      "loss": 0.2991,
+      "step": 3949
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3596345514950165e-07,
+      "loss": 0.2525,
+      "step": 3950
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.35921926910299e-07,
+      "loss": 0.2515,
+      "step": 3951
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3588039867109633e-07,
+      "loss": 0.2833,
+      "step": 3952
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.358388704318937e-07,
+      "loss": 0.2727,
+      "step": 3953
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3579734219269103e-07,
+      "loss": 0.2484,
+      "step": 3954
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.3575581395348836e-07,
+      "loss": 0.2756,
+      "step": 3955
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.357142857142857e-07,
+      "loss": 0.3182,
+      "step": 3956
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3567275747508304e-07,
+      "loss": 0.2519,
+      "step": 3957
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.356312292358804e-07,
+      "loss": 0.2885,
+      "step": 3958
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3558970099667774e-07,
+      "loss": 0.3014,
+      "step": 3959
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3554817275747507e-07,
+      "loss": 0.3569,
+      "step": 3960
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.355066445182724e-07,
+      "loss": 0.2365,
+      "step": 3961
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3546511627906977e-07,
+      "loss": 0.3033,
+      "step": 3962
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.354235880398671e-07,
+      "loss": 0.274,
+      "step": 3963
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3538205980066445e-07,
+      "loss": 0.2849,
+      "step": 3964
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3534053156146178e-07,
+      "loss": 0.3347,
+      "step": 3965
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3529900332225912e-07,
+      "loss": 0.2972,
+      "step": 3966
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3525747508305648e-07,
+      "loss": 0.2933,
+      "step": 3967
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3521594684385382e-07,
+      "loss": 0.314,
+      "step": 3968
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3517441860465116e-07,
+      "loss": 0.3055,
+      "step": 3969
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.351328903654485e-07,
+      "loss": 0.2333,
+      "step": 3970
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3509136212624583e-07,
+      "loss": 0.2676,
+      "step": 3971
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.350498338870432e-07,
+      "loss": 0.2664,
+      "step": 3972
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3500830564784053e-07,
+      "loss": 0.2253,
+      "step": 3973
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3496677740863787e-07,
+      "loss": 0.2581,
+      "step": 3974
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.349252491694352e-07,
+      "loss": 0.2586,
+      "step": 3975
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3488372093023257e-07,
+      "loss": 0.2994,
+      "step": 3976
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.348421926910299e-07,
+      "loss": 0.2993,
+      "step": 3977
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3480066445182722e-07,
+      "loss": 0.2951,
+      "step": 3978
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3475913621262458e-07,
+      "loss": 0.3081,
+      "step": 3979
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3471760797342192e-07,
+      "loss": 0.2623,
+      "step": 3980
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3467607973421928e-07,
+      "loss": 0.2536,
+      "step": 3981
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.346345514950166e-07,
+      "loss": 0.3077,
+      "step": 3982
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3459302325581395e-07,
+      "loss": 0.3211,
+      "step": 3983
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.345514950166113e-07,
+      "loss": 0.3719,
+      "step": 3984
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3450996677740865e-07,
+      "loss": 0.2534,
+      "step": 3985
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3446843853820596e-07,
+      "loss": 0.2977,
+      "step": 3986
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.344269102990033e-07,
+      "loss": 0.2672,
+      "step": 3987
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3438538205980066e-07,
+      "loss": 0.2985,
+      "step": 3988
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.34343853820598e-07,
+      "loss": 0.3431,
+      "step": 3989
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3430232558139534e-07,
+      "loss": 0.2219,
+      "step": 3990
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.3426079734219267e-07,
+      "loss": 0.2744,
+      "step": 3991
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3421926910299004e-07,
+      "loss": 0.3095,
+      "step": 3992
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3417774086378737e-07,
+      "loss": 0.3216,
+      "step": 3993
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.341362126245847e-07,
+      "loss": 0.2905,
+      "step": 3994
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3409468438538205e-07,
+      "loss": 0.296,
+      "step": 3995
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3405315614617938e-07,
+      "loss": 0.282,
+      "step": 3996
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3401162790697675e-07,
+      "loss": 0.277,
+      "step": 3997
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3397009966777408e-07,
+      "loss": 0.3042,
+      "step": 3998
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3392857142857142e-07,
+      "loss": 0.325,
+      "step": 3999
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3388704318936876e-07,
+      "loss": 0.2554,
+      "step": 4000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.338455149501661e-07,
+      "loss": 0.2594,
+      "step": 4001
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3380398671096346e-07,
+      "loss": 0.2923,
+      "step": 4002
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.337624584717608e-07,
+      "loss": 0.3037,
+      "step": 4003
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3372093023255813e-07,
+      "loss": 0.297,
+      "step": 4004
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3367940199335547e-07,
+      "loss": 0.4008,
+      "step": 4005
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3363787375415283e-07,
+      "loss": 0.2652,
+      "step": 4006
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3359634551495017e-07,
+      "loss": 0.3145,
+      "step": 4007
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.335548172757475e-07,
+      "loss": 0.2886,
+      "step": 4008
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3351328903654484e-07,
+      "loss": 0.2582,
+      "step": 4009
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3347176079734218e-07,
+      "loss": 0.2685,
+      "step": 4010
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3343023255813954e-07,
+      "loss": 0.307,
+      "step": 4011
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3338870431893688e-07,
+      "loss": 0.2891,
+      "step": 4012
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3334717607973422e-07,
+      "loss": 0.3155,
+      "step": 4013
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3330564784053155e-07,
+      "loss": 0.2219,
+      "step": 4014
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.332641196013289e-07,
+      "loss": 0.3354,
+      "step": 4015
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3322259136212625e-07,
+      "loss": 0.2346,
+      "step": 4016
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.331810631229236e-07,
+      "loss": 0.222,
+      "step": 4017
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3313953488372093e-07,
+      "loss": 0.3491,
+      "step": 4018
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3309800664451826e-07,
+      "loss": 0.2751,
+      "step": 4019
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3305647840531563e-07,
+      "loss": 0.329,
+      "step": 4020
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3301495016611296e-07,
+      "loss": 0.2957,
+      "step": 4021
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.329734219269103e-07,
+      "loss": 0.2502,
+      "step": 4022
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3293189368770764e-07,
+      "loss": 0.2698,
+      "step": 4023
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3289036544850497e-07,
+      "loss": 0.2908,
+      "step": 4024
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.328488372093023e-07,
+      "loss": 0.2887,
+      "step": 4025
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.3280730897009965e-07,
+      "loss": 0.37,
+      "step": 4026
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.32765780730897e-07,
+      "loss": 0.3037,
+      "step": 4027
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3272425249169435e-07,
+      "loss": 0.2694,
+      "step": 4028
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3268272425249168e-07,
+      "loss": 0.2332,
+      "step": 4029
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3264119601328902e-07,
+      "loss": 0.2267,
+      "step": 4030
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3259966777408636e-07,
+      "loss": 0.2415,
+      "step": 4031
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3255813953488372e-07,
+      "loss": 0.3032,
+      "step": 4032
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3251661129568106e-07,
+      "loss": 0.233,
+      "step": 4033
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.324750830564784e-07,
+      "loss": 0.3107,
+      "step": 4034
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3243355481727573e-07,
+      "loss": 0.2713,
+      "step": 4035
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.323920265780731e-07,
+      "loss": 0.3185,
+      "step": 4036
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3235049833887043e-07,
+      "loss": 0.3528,
+      "step": 4037
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3230897009966777e-07,
+      "loss": 0.2511,
+      "step": 4038
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.322674418604651e-07,
+      "loss": 0.2999,
+      "step": 4039
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3222591362126244e-07,
+      "loss": 0.3112,
+      "step": 4040
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.321843853820598e-07,
+      "loss": 0.2201,
+      "step": 4041
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3214285714285714e-07,
+      "loss": 0.2925,
+      "step": 4042
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3210132890365448e-07,
+      "loss": 0.2981,
+      "step": 4043
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3205980066445181e-07,
+      "loss": 0.3308,
+      "step": 4044
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3201827242524915e-07,
+      "loss": 0.2548,
+      "step": 4045
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3197674418604651e-07,
+      "loss": 0.2995,
+      "step": 4046
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3193521594684385e-07,
+      "loss": 0.278,
+      "step": 4047
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.318936877076412e-07,
+      "loss": 0.3409,
+      "step": 4048
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3185215946843852e-07,
+      "loss": 0.3042,
+      "step": 4049
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.318106312292359e-07,
+      "loss": 0.4372,
+      "step": 4050
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3176910299003322e-07,
+      "loss": 0.3561,
+      "step": 4051
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3172757475083056e-07,
+      "loss": 0.3033,
+      "step": 4052
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.316860465116279e-07,
+      "loss": 0.2876,
+      "step": 4053
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3164451827242524e-07,
+      "loss": 0.2572,
+      "step": 4054
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.316029900332226e-07,
+      "loss": 0.2481,
+      "step": 4055
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3156146179401994e-07,
+      "loss": 0.3189,
+      "step": 4056
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3151993355481727e-07,
+      "loss": 0.2678,
+      "step": 4057
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.314784053156146e-07,
+      "loss": 0.2867,
+      "step": 4058
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3143687707641197e-07,
+      "loss": 0.2541,
+      "step": 4059
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.313953488372093e-07,
+      "loss": 0.2582,
+      "step": 4060
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3135382059800662e-07,
+      "loss": 0.2983,
+      "step": 4061
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3131229235880398e-07,
+      "loss": 0.2864,
+      "step": 4062
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.3127076411960132e-07,
+      "loss": 0.2817,
+      "step": 4063
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3122923588039868e-07,
+      "loss": 0.2364,
+      "step": 4064
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.31187707641196e-07,
+      "loss": 0.2663,
+      "step": 4065
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3114617940199336e-07,
+      "loss": 0.3625,
+      "step": 4066
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.311046511627907e-07,
+      "loss": 0.3129,
+      "step": 4067
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3106312292358803e-07,
+      "loss": 0.2497,
+      "step": 4068
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3102159468438537e-07,
+      "loss": 0.3232,
+      "step": 4069
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.309800664451827e-07,
+      "loss": 0.2535,
+      "step": 4070
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3093853820598007e-07,
+      "loss": 0.3093,
+      "step": 4071
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.308970099667774e-07,
+      "loss": 0.2621,
+      "step": 4072
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3085548172757474e-07,
+      "loss": 0.2603,
+      "step": 4073
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3081395348837208e-07,
+      "loss": 0.2787,
+      "step": 4074
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3077242524916941e-07,
+      "loss": 0.3032,
+      "step": 4075
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3073089700996678e-07,
+      "loss": 0.2563,
+      "step": 4076
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3068936877076411e-07,
+      "loss": 0.2435,
+      "step": 4077
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3064784053156145e-07,
+      "loss": 0.3171,
+      "step": 4078
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.306063122923588e-07,
+      "loss": 0.2899,
+      "step": 4079
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3056478405315615e-07,
+      "loss": 0.2877,
+      "step": 4080
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.305232558139535e-07,
+      "loss": 0.2466,
+      "step": 4081
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3048172757475082e-07,
+      "loss": 0.3232,
+      "step": 4082
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3044019933554816e-07,
+      "loss": 0.2279,
+      "step": 4083
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.303986710963455e-07,
+      "loss": 0.3121,
+      "step": 4084
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3035714285714286e-07,
+      "loss": 0.2689,
+      "step": 4085
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.303156146179402e-07,
+      "loss": 0.3144,
+      "step": 4086
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3027408637873753e-07,
+      "loss": 0.3042,
+      "step": 4087
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3023255813953487e-07,
+      "loss": 0.2637,
+      "step": 4088
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3019102990033223e-07,
+      "loss": 0.2393,
+      "step": 4089
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3014950166112957e-07,
+      "loss": 0.2499,
+      "step": 4090
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.301079734219269e-07,
+      "loss": 0.3573,
+      "step": 4091
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3006644518272424e-07,
+      "loss": 0.2698,
+      "step": 4092
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.3002491694352158e-07,
+      "loss": 0.2706,
+      "step": 4093
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2998338870431894e-07,
+      "loss": 0.2822,
+      "step": 4094
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2994186046511628e-07,
+      "loss": 0.2594,
+      "step": 4095
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2990033222591362e-07,
+      "loss": 0.2725,
+      "step": 4096
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2985880398671096e-07,
+      "loss": 0.1797,
+      "step": 4097
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.298172757475083e-07,
+      "loss": 0.2857,
+      "step": 4098
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.2977574750830566e-07,
+      "loss": 0.2612,
+      "step": 4099
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.29734219269103e-07,
+      "loss": 0.2474,
+      "step": 4100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2969269102990033e-07,
+      "loss": 0.2523,
+      "step": 4101
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2965116279069767e-07,
+      "loss": 0.244,
+      "step": 4102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2960963455149503e-07,
+      "loss": 0.2889,
+      "step": 4103
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2956810631229237e-07,
+      "loss": 0.2874,
+      "step": 4104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2952657807308968e-07,
+      "loss": 0.3427,
+      "step": 4105
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2948504983388704e-07,
+      "loss": 0.374,
+      "step": 4106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2944352159468438e-07,
+      "loss": 0.3201,
+      "step": 4107
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.294019933554817e-07,
+      "loss": 0.3483,
+      "step": 4108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2936046511627905e-07,
+      "loss": 0.3226,
+      "step": 4109
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.293189368770764e-07,
+      "loss": 0.279,
+      "step": 4110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2927740863787375e-07,
+      "loss": 0.3247,
+      "step": 4111
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2923588039867109e-07,
+      "loss": 0.2575,
+      "step": 4112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2919435215946842e-07,
+      "loss": 0.2334,
+      "step": 4113
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2915282392026576e-07,
+      "loss": 0.2811,
+      "step": 4114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2911129568106312e-07,
+      "loss": 0.2948,
+      "step": 4115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2906976744186046e-07,
+      "loss": 0.2803,
+      "step": 4116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.290282392026578e-07,
+      "loss": 0.2559,
+      "step": 4117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2898671096345513e-07,
+      "loss": 0.2753,
+      "step": 4118
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.289451827242525e-07,
+      "loss": 0.2789,
+      "step": 4119
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2890365448504983e-07,
+      "loss": 0.3261,
+      "step": 4120
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2886212624584717e-07,
+      "loss": 0.2724,
+      "step": 4121
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.288205980066445e-07,
+      "loss": 0.2987,
+      "step": 4122
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2877906976744184e-07,
+      "loss": 0.2905,
+      "step": 4123
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.287375415282392e-07,
+      "loss": 0.3705,
+      "step": 4124
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2869601328903654e-07,
+      "loss": 0.269,
+      "step": 4125
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2865448504983388e-07,
+      "loss": 0.259,
+      "step": 4126
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2861295681063122e-07,
+      "loss": 0.3924,
+      "step": 4127
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2857142857142855e-07,
+      "loss": 0.3021,
+      "step": 4128
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2852990033222592e-07,
+      "loss": 0.2636,
+      "step": 4129
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2848837209302325e-07,
+      "loss": 0.2939,
+      "step": 4130
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.284468438538206e-07,
+      "loss": 0.3271,
+      "step": 4131
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2840531561461793e-07,
+      "loss": 0.2281,
+      "step": 4132
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.283637873754153e-07,
+      "loss": 0.2514,
+      "step": 4133
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2832225913621263e-07,
+      "loss": 0.2907,
+      "step": 4134
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.2828073089700996e-07,
+      "loss": 0.2981,
+      "step": 4135
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.282392026578073e-07,
+      "loss": 0.3062,
+      "step": 4136
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2819767441860464e-07,
+      "loss": 0.3123,
+      "step": 4137
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.28156146179402e-07,
+      "loss": 0.285,
+      "step": 4138
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2811461794019934e-07,
+      "loss": 0.2733,
+      "step": 4139
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2807308970099668e-07,
+      "loss": 0.2992,
+      "step": 4140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.28031561461794e-07,
+      "loss": 0.2655,
+      "step": 4141
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2799003322259135e-07,
+      "loss": 0.2586,
+      "step": 4142
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.279485049833887e-07,
+      "loss": 0.2503,
+      "step": 4143
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2790697674418602e-07,
+      "loss": 0.301,
+      "step": 4144
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2786544850498339e-07,
+      "loss": 0.3294,
+      "step": 4145
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2782392026578072e-07,
+      "loss": 0.2973,
+      "step": 4146
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2778239202657809e-07,
+      "loss": 0.2797,
+      "step": 4147
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.277408637873754e-07,
+      "loss": 0.3578,
+      "step": 4148
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2769933554817276e-07,
+      "loss": 0.2998,
+      "step": 4149
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.276578073089701e-07,
+      "loss": 0.3568,
+      "step": 4150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2761627906976743e-07,
+      "loss": 0.2692,
+      "step": 4151
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2757475083056477e-07,
+      "loss": 0.2488,
+      "step": 4152
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.275332225913621e-07,
+      "loss": 0.2519,
+      "step": 4153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2749169435215947e-07,
+      "loss": 0.2861,
+      "step": 4154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.274501661129568e-07,
+      "loss": 0.3237,
+      "step": 4155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2740863787375414e-07,
+      "loss": 0.2154,
+      "step": 4156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2736710963455148e-07,
+      "loss": 0.2969,
+      "step": 4157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2732558139534882e-07,
+      "loss": 0.2564,
+      "step": 4158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2728405315614618e-07,
+      "loss": 0.3136,
+      "step": 4159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2724252491694352e-07,
+      "loss": 0.2696,
+      "step": 4160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2720099667774085e-07,
+      "loss": 0.366,
+      "step": 4161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.271594684385382e-07,
+      "loss": 0.2859,
+      "step": 4162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2711794019933555e-07,
+      "loss": 0.2745,
+      "step": 4163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.270764119601329e-07,
+      "loss": 0.3133,
+      "step": 4164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2703488372093023e-07,
+      "loss": 0.2209,
+      "step": 4165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2699335548172756e-07,
+      "loss": 0.2383,
+      "step": 4166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.269518272425249e-07,
+      "loss": 0.2632,
+      "step": 4167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2691029900332226e-07,
+      "loss": 0.2787,
+      "step": 4168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.268687707641196e-07,
+      "loss": 0.3061,
+      "step": 4169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2682724252491694e-07,
+      "loss": 0.3098,
+      "step": 4170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.2678571428571427e-07,
+      "loss": 0.3256,
+      "step": 4171
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.267441860465116e-07,
+      "loss": 0.2877,
+      "step": 4172
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2670265780730897e-07,
+      "loss": 0.2931,
+      "step": 4173
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.266611295681063e-07,
+      "loss": 0.3494,
+      "step": 4174
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2661960132890365e-07,
+      "loss": 0.2335,
+      "step": 4175
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2657807308970098e-07,
+      "loss": 0.3631,
+      "step": 4176
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2653654485049835e-07,
+      "loss": 0.2185,
+      "step": 4177
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2649501661129568e-07,
+      "loss": 0.2475,
+      "step": 4178
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2645348837209302e-07,
+      "loss": 0.3216,
+      "step": 4179
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2641196013289036e-07,
+      "loss": 0.2754,
+      "step": 4180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.263704318936877e-07,
+      "loss": 0.2506,
+      "step": 4181
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2632890365448506e-07,
+      "loss": 0.3726,
+      "step": 4182
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.262873754152824e-07,
+      "loss": 0.2784,
+      "step": 4183
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2624584717607973e-07,
+      "loss": 0.292,
+      "step": 4184
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2620431893687707e-07,
+      "loss": 0.3085,
+      "step": 4185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2616279069767443e-07,
+      "loss": 0.3718,
+      "step": 4186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2612126245847174e-07,
+      "loss": 0.262,
+      "step": 4187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2607973421926908e-07,
+      "loss": 0.2396,
+      "step": 4188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2603820598006644e-07,
+      "loss": 0.2823,
+      "step": 4189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2599667774086378e-07,
+      "loss": 0.2587,
+      "step": 4190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2595514950166112e-07,
+      "loss": 0.2937,
+      "step": 4191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2591362126245845e-07,
+      "loss": 0.2834,
+      "step": 4192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2587209302325582e-07,
+      "loss": 0.3243,
+      "step": 4193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2583056478405315e-07,
+      "loss": 0.3099,
+      "step": 4194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.257890365448505e-07,
+      "loss": 0.2926,
+      "step": 4195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2574750830564783e-07,
+      "loss": 0.3213,
+      "step": 4196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2570598006644516e-07,
+      "loss": 0.2585,
+      "step": 4197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2566445182724253e-07,
+      "loss": 0.2125,
+      "step": 4198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2562292358803986e-07,
+      "loss": 0.2837,
+      "step": 4199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.255813953488372e-07,
+      "loss": 0.2379,
+      "step": 4200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2553986710963454e-07,
+      "loss": 0.3034,
+      "step": 4201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2549833887043187e-07,
+      "loss": 0.2264,
+      "step": 4202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2545681063122924e-07,
+      "loss": 0.3013,
+      "step": 4203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2541528239202657e-07,
+      "loss": 0.262,
+      "step": 4204
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.253737541528239e-07,
+      "loss": 0.2696,
+      "step": 4205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.2533222591362125e-07,
+      "loss": 0.3329,
+      "step": 4206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.252906976744186e-07,
+      "loss": 0.3353,
+      "step": 4207
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2524916943521595e-07,
+      "loss": 0.3267,
+      "step": 4208
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2520764119601328e-07,
+      "loss": 0.294,
+      "step": 4209
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2516611295681062e-07,
+      "loss": 0.2553,
+      "step": 4210
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2512458471760796e-07,
+      "loss": 0.3093,
+      "step": 4211
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2508305647840532e-07,
+      "loss": 0.2999,
+      "step": 4212
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2504152823920266e-07,
+      "loss": 0.3517,
+      "step": 4213
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.25e-07,
+      "loss": 0.2334,
+      "step": 4214
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2495847176079733e-07,
+      "loss": 0.3276,
+      "step": 4215
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.249169435215947e-07,
+      "loss": 0.2797,
+      "step": 4216
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2487541528239203e-07,
+      "loss": 0.307,
+      "step": 4217
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2483388704318937e-07,
+      "loss": 0.3301,
+      "step": 4218
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.247923588039867e-07,
+      "loss": 0.276,
+      "step": 4219
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2475083056478404e-07,
+      "loss": 0.3263,
+      "step": 4220
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.247093023255814e-07,
+      "loss": 0.2866,
+      "step": 4221
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2466777408637874e-07,
+      "loss": 0.2565,
+      "step": 4222
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2462624584717608e-07,
+      "loss": 0.3214,
+      "step": 4223
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2458471760797342e-07,
+      "loss": 0.3481,
+      "step": 4224
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2454318936877075e-07,
+      "loss": 0.3633,
+      "step": 4225
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2450166112956812e-07,
+      "loss": 0.2805,
+      "step": 4226
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2446013289036543e-07,
+      "loss": 0.3251,
+      "step": 4227
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.244186046511628e-07,
+      "loss": 0.3232,
+      "step": 4228
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2437707641196013e-07,
+      "loss": 0.2476,
+      "step": 4229
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.243355481727575e-07,
+      "loss": 0.2356,
+      "step": 4230
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.242940199335548e-07,
+      "loss": 0.3045,
+      "step": 4231
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2425249169435214e-07,
+      "loss": 0.2603,
+      "step": 4232
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.242109634551495e-07,
+      "loss": 0.299,
+      "step": 4233
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2416943521594684e-07,
+      "loss": 0.2813,
+      "step": 4234
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2412790697674417e-07,
+      "loss": 0.3125,
+      "step": 4235
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.240863787375415e-07,
+      "loss": 0.3171,
+      "step": 4236
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2404485049833887e-07,
+      "loss": 0.2852,
+      "step": 4237
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.240033222591362e-07,
+      "loss": 0.2791,
+      "step": 4238
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2396179401993355e-07,
+      "loss": 0.2635,
+      "step": 4239
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2392026578073088e-07,
+      "loss": 0.2379,
+      "step": 4240
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2387873754152822e-07,
+      "loss": 0.3297,
+      "step": 4241
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2383720930232558e-07,
+      "loss": 0.3121,
+      "step": 4242
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2379568106312292e-07,
+      "loss": 0.256,
+      "step": 4243
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.2375415282392026e-07,
+      "loss": 0.2807,
+      "step": 4244
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.237126245847176e-07,
+      "loss": 0.2903,
+      "step": 4245
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2367109634551496e-07,
+      "loss": 0.2844,
+      "step": 4246
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.236295681063123e-07,
+      "loss": 0.3429,
+      "step": 4247
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2358803986710963e-07,
+      "loss": 0.274,
+      "step": 4248
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2354651162790697e-07,
+      "loss": 0.2675,
+      "step": 4249
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.235049833887043e-07,
+      "loss": 0.2531,
+      "step": 4250
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2346345514950167e-07,
+      "loss": 0.2448,
+      "step": 4251
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.23421926910299e-07,
+      "loss": 0.236,
+      "step": 4252
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2338039867109634e-07,
+      "loss": 0.2755,
+      "step": 4253
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2333887043189368e-07,
+      "loss": 0.3255,
+      "step": 4254
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2329734219269101e-07,
+      "loss": 0.2749,
+      "step": 4255
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2325581395348838e-07,
+      "loss": 0.281,
+      "step": 4256
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2321428571428571e-07,
+      "loss": 0.3097,
+      "step": 4257
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2317275747508305e-07,
+      "loss": 0.3087,
+      "step": 4258
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.231312292358804e-07,
+      "loss": 0.297,
+      "step": 4259
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2308970099667775e-07,
+      "loss": 0.2342,
+      "step": 4260
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.230481727574751e-07,
+      "loss": 0.2641,
+      "step": 4261
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2300664451827242e-07,
+      "loss": 0.2472,
+      "step": 4262
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2296511627906976e-07,
+      "loss": 0.2847,
+      "step": 4263
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.229235880398671e-07,
+      "loss": 0.2925,
+      "step": 4264
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2288205980066446e-07,
+      "loss": 0.3262,
+      "step": 4265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.228405315614618e-07,
+      "loss": 0.2782,
+      "step": 4266
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2279900332225914e-07,
+      "loss": 0.2595,
+      "step": 4267
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2275747508305647e-07,
+      "loss": 0.3052,
+      "step": 4268
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.227159468438538e-07,
+      "loss": 0.3231,
+      "step": 4269
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2267441860465115e-07,
+      "loss": 0.2131,
+      "step": 4270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2263289036544848e-07,
+      "loss": 0.284,
+      "step": 4271
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2259136212624585e-07,
+      "loss": 0.3394,
+      "step": 4272
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2254983388704318e-07,
+      "loss": 0.3069,
+      "step": 4273
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2250830564784052e-07,
+      "loss": 0.2772,
+      "step": 4274
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2246677740863786e-07,
+      "loss": 0.3032,
+      "step": 4275
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2242524916943522e-07,
+      "loss": 0.2791,
+      "step": 4276
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2238372093023256e-07,
+      "loss": 0.3395,
+      "step": 4277
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.223421926910299e-07,
+      "loss": 0.2706,
+      "step": 4278
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2230066445182723e-07,
+      "loss": 0.2969,
+      "step": 4279
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.2225913621262457e-07,
+      "loss": 0.2939,
+      "step": 4280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2221760797342193e-07,
+      "loss": 0.3384,
+      "step": 4281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2217607973421927e-07,
+      "loss": 0.3022,
+      "step": 4282
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.221345514950166e-07,
+      "loss": 0.2541,
+      "step": 4283
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2209302325581394e-07,
+      "loss": 0.2921,
+      "step": 4284
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2205149501661128e-07,
+      "loss": 0.2944,
+      "step": 4285
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2200996677740864e-07,
+      "loss": 0.2795,
+      "step": 4286
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2196843853820598e-07,
+      "loss": 0.2281,
+      "step": 4287
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2192691029900331e-07,
+      "loss": 0.3093,
+      "step": 4288
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2188538205980065e-07,
+      "loss": 0.2944,
+      "step": 4289
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2184385382059801e-07,
+      "loss": 0.2685,
+      "step": 4290
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2180232558139535e-07,
+      "loss": 0.3187,
+      "step": 4291
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.217607973421927e-07,
+      "loss": 0.3122,
+      "step": 4292
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2171926910299002e-07,
+      "loss": 0.3006,
+      "step": 4293
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2167774086378736e-07,
+      "loss": 0.3249,
+      "step": 4294
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2163621262458472e-07,
+      "loss": 0.3025,
+      "step": 4295
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2159468438538206e-07,
+      "loss": 0.276,
+      "step": 4296
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.215531561461794e-07,
+      "loss": 0.3056,
+      "step": 4297
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2151162790697673e-07,
+      "loss": 0.3268,
+      "step": 4298
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2147009966777407e-07,
+      "loss": 0.339,
+      "step": 4299
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2142857142857143e-07,
+      "loss": 0.2594,
+      "step": 4300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2138704318936877e-07,
+      "loss": 0.2917,
+      "step": 4301
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.213455149501661e-07,
+      "loss": 0.273,
+      "step": 4302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2130398671096344e-07,
+      "loss": 0.2823,
+      "step": 4303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.212624584717608e-07,
+      "loss": 0.3653,
+      "step": 4304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2122093023255814e-07,
+      "loss": 0.2391,
+      "step": 4305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2117940199335548e-07,
+      "loss": 0.34,
+      "step": 4306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2113787375415282e-07,
+      "loss": 0.2508,
+      "step": 4307
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2109634551495016e-07,
+      "loss": 0.3088,
+      "step": 4308
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2105481727574752e-07,
+      "loss": 0.2478,
+      "step": 4309
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2101328903654483e-07,
+      "loss": 0.3637,
+      "step": 4310
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.209717607973422e-07,
+      "loss": 0.3443,
+      "step": 4311
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2093023255813953e-07,
+      "loss": 0.351,
+      "step": 4312
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.208887043189369e-07,
+      "loss": 0.2654,
+      "step": 4313
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.208471760797342e-07,
+      "loss": 0.2391,
+      "step": 4314
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.2080564784053154e-07,
+      "loss": 0.2819,
+      "step": 4315
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.207641196013289e-07,
+      "loss": 0.3345,
+      "step": 4316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2072259136212624e-07,
+      "loss": 0.2608,
+      "step": 4317
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2068106312292358e-07,
+      "loss": 0.2517,
+      "step": 4318
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.206395348837209e-07,
+      "loss": 0.285,
+      "step": 4319
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2059800664451828e-07,
+      "loss": 0.2687,
+      "step": 4320
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.205564784053156e-07,
+      "loss": 0.2643,
+      "step": 4321
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2051495016611295e-07,
+      "loss": 0.2377,
+      "step": 4322
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2047342192691029e-07,
+      "loss": 0.3624,
+      "step": 4323
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2043189368770762e-07,
+      "loss": 0.2656,
+      "step": 4324
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2039036544850499e-07,
+      "loss": 0.3025,
+      "step": 4325
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2034883720930232e-07,
+      "loss": 0.3106,
+      "step": 4326
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2030730897009966e-07,
+      "loss": 0.3111,
+      "step": 4327
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.20265780730897e-07,
+      "loss": 0.3034,
+      "step": 4328
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2022425249169433e-07,
+      "loss": 0.3006,
+      "step": 4329
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.201827242524917e-07,
+      "loss": 0.2623,
+      "step": 4330
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2014119601328903e-07,
+      "loss": 0.2733,
+      "step": 4331
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2009966777408637e-07,
+      "loss": 0.3001,
+      "step": 4332
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.200581395348837e-07,
+      "loss": 0.2322,
+      "step": 4333
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.2001661129568107e-07,
+      "loss": 0.2373,
+      "step": 4334
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.199750830564784e-07,
+      "loss": 0.2418,
+      "step": 4335
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1993355481727574e-07,
+      "loss": 0.281,
+      "step": 4336
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1989202657807308e-07,
+      "loss": 0.2779,
+      "step": 4337
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1985049833887042e-07,
+      "loss": 0.2965,
+      "step": 4338
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1980897009966778e-07,
+      "loss": 0.314,
+      "step": 4339
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1976744186046512e-07,
+      "loss": 0.2654,
+      "step": 4340
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1972591362126245e-07,
+      "loss": 0.2761,
+      "step": 4341
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.196843853820598e-07,
+      "loss": 0.2533,
+      "step": 4342
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1964285714285715e-07,
+      "loss": 0.3116,
+      "step": 4343
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.196013289036545e-07,
+      "loss": 0.2479,
+      "step": 4344
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1955980066445183e-07,
+      "loss": 0.3073,
+      "step": 4345
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1951827242524916e-07,
+      "loss": 0.3212,
+      "step": 4346
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.194767441860465e-07,
+      "loss": 0.2409,
+      "step": 4347
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1943521594684386e-07,
+      "loss": 0.3188,
+      "step": 4348
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.193936877076412e-07,
+      "loss": 0.2848,
+      "step": 4349
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1935215946843854e-07,
+      "loss": 0.2649,
+      "step": 4350
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.1931063122923588e-07,
+      "loss": 0.2877,
+      "step": 4351
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.192691029900332e-07,
+      "loss": 0.2937,
+      "step": 4352
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1922757475083055e-07,
+      "loss": 0.2703,
+      "step": 4353
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.191860465116279e-07,
+      "loss": 0.2894,
+      "step": 4354
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1914451827242525e-07,
+      "loss": 0.2345,
+      "step": 4355
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1910299003322259e-07,
+      "loss": 0.3572,
+      "step": 4356
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1906146179401994e-07,
+      "loss": 0.2982,
+      "step": 4357
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1901993355481727e-07,
+      "loss": 0.2503,
+      "step": 4358
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1897840531561461e-07,
+      "loss": 0.2977,
+      "step": 4359
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1893687707641196e-07,
+      "loss": 0.3528,
+      "step": 4360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.188953488372093e-07,
+      "loss": 0.2964,
+      "step": 4361
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1885382059800665e-07,
+      "loss": 0.2758,
+      "step": 4362
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1881229235880398e-07,
+      "loss": 0.3579,
+      "step": 4363
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1877076411960133e-07,
+      "loss": 0.3499,
+      "step": 4364
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1872923588039867e-07,
+      "loss": 0.3236,
+      "step": 4365
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.18687707641196e-07,
+      "loss": 0.2637,
+      "step": 4366
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1864617940199336e-07,
+      "loss": 0.2883,
+      "step": 4367
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1860465116279068e-07,
+      "loss": 0.2823,
+      "step": 4368
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1856312292358804e-07,
+      "loss": 0.3312,
+      "step": 4369
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1852159468438537e-07,
+      "loss": 0.2537,
+      "step": 4370
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1848006644518273e-07,
+      "loss": 0.2714,
+      "step": 4371
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1843853820598005e-07,
+      "loss": 0.3369,
+      "step": 4372
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1839700996677742e-07,
+      "loss": 0.3058,
+      "step": 4373
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1835548172757474e-07,
+      "loss": 0.2535,
+      "step": 4374
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1831395348837208e-07,
+      "loss": 0.3246,
+      "step": 4375
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1827242524916943e-07,
+      "loss": 0.301,
+      "step": 4376
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1823089700996676e-07,
+      "loss": 0.2541,
+      "step": 4377
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1818936877076411e-07,
+      "loss": 0.3243,
+      "step": 4378
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1814784053156145e-07,
+      "loss": 0.3235,
+      "step": 4379
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.181063122923588e-07,
+      "loss": 0.2673,
+      "step": 4380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1806478405315614e-07,
+      "loss": 0.2741,
+      "step": 4381
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1802325581395347e-07,
+      "loss": 0.2918,
+      "step": 4382
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1798172757475082e-07,
+      "loss": 0.3158,
+      "step": 4383
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1794019933554816e-07,
+      "loss": 0.2918,
+      "step": 4384
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1789867109634551e-07,
+      "loss": 0.3277,
+      "step": 4385
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1785714285714285e-07,
+      "loss": 0.3166,
+      "step": 4386
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.178156146179402e-07,
+      "loss": 0.2671,
+      "step": 4387
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.1777408637873753e-07,
+      "loss": 0.3288,
+      "step": 4388
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1773255813953487e-07,
+      "loss": 0.3389,
+      "step": 4389
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1769102990033222e-07,
+      "loss": 0.3851,
+      "step": 4390
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1764950166112956e-07,
+      "loss": 0.3025,
+      "step": 4391
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1760797342192691e-07,
+      "loss": 0.2801,
+      "step": 4392
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1756644518272425e-07,
+      "loss": 0.2952,
+      "step": 4393
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.175249169435216e-07,
+      "loss": 0.3179,
+      "step": 4394
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1748338870431893e-07,
+      "loss": 0.3163,
+      "step": 4395
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1744186046511627e-07,
+      "loss": 0.223,
+      "step": 4396
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1740033222591362e-07,
+      "loss": 0.3192,
+      "step": 4397
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1735880398671096e-07,
+      "loss": 0.2779,
+      "step": 4398
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.173172757475083e-07,
+      "loss": 0.2646,
+      "step": 4399
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1727574750830564e-07,
+      "loss": 0.3789,
+      "step": 4400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1723421926910299e-07,
+      "loss": 0.2606,
+      "step": 4401
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1719269102990033e-07,
+      "loss": 0.2511,
+      "step": 4402
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1715116279069768e-07,
+      "loss": 0.2897,
+      "step": 4403
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1710963455149502e-07,
+      "loss": 0.2911,
+      "step": 4404
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1706810631229235e-07,
+      "loss": 0.269,
+      "step": 4405
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.170265780730897e-07,
+      "loss": 0.2334,
+      "step": 4406
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1698504983388704e-07,
+      "loss": 0.3048,
+      "step": 4407
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1694352159468439e-07,
+      "loss": 0.3198,
+      "step": 4408
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1690199335548173e-07,
+      "loss": 0.2521,
+      "step": 4409
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1686046511627908e-07,
+      "loss": 0.2865,
+      "step": 4410
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1681893687707641e-07,
+      "loss": 0.2958,
+      "step": 4411
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1677740863787374e-07,
+      "loss": 0.2989,
+      "step": 4412
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.167358803986711e-07,
+      "loss": 0.2408,
+      "step": 4413
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1669435215946842e-07,
+      "loss": 0.2723,
+      "step": 4414
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1665282392026579e-07,
+      "loss": 0.3528,
+      "step": 4415
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1661129568106311e-07,
+      "loss": 0.3234,
+      "step": 4416
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1656976744186046e-07,
+      "loss": 0.2607,
+      "step": 4417
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.165282392026578e-07,
+      "loss": 0.2439,
+      "step": 4418
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1648671096345513e-07,
+      "loss": 0.2591,
+      "step": 4419
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1644518272425248e-07,
+      "loss": 0.2789,
+      "step": 4420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1640365448504982e-07,
+      "loss": 0.3015,
+      "step": 4421
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1636212624584717e-07,
+      "loss": 0.3086,
+      "step": 4422
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1632059800664451e-07,
+      "loss": 0.3098,
+      "step": 4423
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.1627906976744186e-07,
+      "loss": 0.3125,
+      "step": 4424
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.162375415282392e-07,
+      "loss": 0.3083,
+      "step": 4425
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1619601328903653e-07,
+      "loss": 0.2979,
+      "step": 4426
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1615448504983388e-07,
+      "loss": 0.2674,
+      "step": 4427
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1611295681063122e-07,
+      "loss": 0.2905,
+      "step": 4428
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1607142857142857e-07,
+      "loss": 0.3085,
+      "step": 4429
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.160299003322259e-07,
+      "loss": 0.1995,
+      "step": 4430
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1598837209302325e-07,
+      "loss": 0.3108,
+      "step": 4431
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1594684385382059e-07,
+      "loss": 0.3366,
+      "step": 4432
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1590531561461793e-07,
+      "loss": 0.2587,
+      "step": 4433
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1586378737541528e-07,
+      "loss": 0.2551,
+      "step": 4434
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1582225913621262e-07,
+      "loss": 0.2463,
+      "step": 4435
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1578073089700997e-07,
+      "loss": 0.294,
+      "step": 4436
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.157392026578073e-07,
+      "loss": 0.3264,
+      "step": 4437
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1569767441860465e-07,
+      "loss": 0.2834,
+      "step": 4438
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1565614617940199e-07,
+      "loss": 0.263,
+      "step": 4439
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1561461794019934e-07,
+      "loss": 0.2615,
+      "step": 4440
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1557308970099668e-07,
+      "loss": 0.2388,
+      "step": 4441
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1553156146179401e-07,
+      "loss": 0.3115,
+      "step": 4442
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1549003322259136e-07,
+      "loss": 0.2874,
+      "step": 4443
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.154485049833887e-07,
+      "loss": 0.3337,
+      "step": 4444
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1540697674418605e-07,
+      "loss": 0.2683,
+      "step": 4445
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1536544850498339e-07,
+      "loss": 0.2434,
+      "step": 4446
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1532392026578074e-07,
+      "loss": 0.2783,
+      "step": 4447
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1528239202657807e-07,
+      "loss": 0.2413,
+      "step": 4448
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.152408637873754e-07,
+      "loss": 0.3112,
+      "step": 4449
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1519933554817276e-07,
+      "loss": 0.2767,
+      "step": 4450
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1515780730897008e-07,
+      "loss": 0.305,
+      "step": 4451
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1511627906976745e-07,
+      "loss": 0.2958,
+      "step": 4452
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1507475083056477e-07,
+      "loss": 0.3306,
+      "step": 4453
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1503322259136213e-07,
+      "loss": 0.3833,
+      "step": 4454
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1499169435215946e-07,
+      "loss": 0.2829,
+      "step": 4455
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.149501661129568e-07,
+      "loss": 0.2905,
+      "step": 4456
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1490863787375414e-07,
+      "loss": 0.2968,
+      "step": 4457
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1486710963455148e-07,
+      "loss": 0.2652,
+      "step": 4458
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1482558139534883e-07,
+      "loss": 0.3527,
+      "step": 4459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.1478405315614617e-07,
+      "loss": 0.3271,
+      "step": 4460
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1474252491694352e-07,
+      "loss": 0.2604,
+      "step": 4461
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1470099667774085e-07,
+      "loss": 0.2827,
+      "step": 4462
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1465946843853819e-07,
+      "loss": 0.2541,
+      "step": 4463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1461794019933554e-07,
+      "loss": 0.2984,
+      "step": 4464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1457641196013288e-07,
+      "loss": 0.289,
+      "step": 4465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1453488372093023e-07,
+      "loss": 0.2991,
+      "step": 4466
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1449335548172756e-07,
+      "loss": 0.2611,
+      "step": 4467
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1445182724252491e-07,
+      "loss": 0.2985,
+      "step": 4468
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1441029900332225e-07,
+      "loss": 0.3293,
+      "step": 4469
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.143687707641196e-07,
+      "loss": 0.2988,
+      "step": 4470
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1432724252491694e-07,
+      "loss": 0.317,
+      "step": 4471
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1428571428571427e-07,
+      "loss": 0.2749,
+      "step": 4472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1424418604651162e-07,
+      "loss": 0.3073,
+      "step": 4473
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1420265780730896e-07,
+      "loss": 0.3023,
+      "step": 4474
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1416112956810631e-07,
+      "loss": 0.2997,
+      "step": 4475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1411960132890365e-07,
+      "loss": 0.2942,
+      "step": 4476
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.14078073089701e-07,
+      "loss": 0.281,
+      "step": 4477
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1403654485049834e-07,
+      "loss": 0.2787,
+      "step": 4478
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1399501661129567e-07,
+      "loss": 0.3295,
+      "step": 4479
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1395348837209302e-07,
+      "loss": 0.2678,
+      "step": 4480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1391196013289036e-07,
+      "loss": 0.2781,
+      "step": 4481
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1387043189368771e-07,
+      "loss": 0.3322,
+      "step": 4482
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1382890365448505e-07,
+      "loss": 0.241,
+      "step": 4483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.137873754152824e-07,
+      "loss": 0.2863,
+      "step": 4484
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1374584717607973e-07,
+      "loss": 0.2915,
+      "step": 4485
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1370431893687707e-07,
+      "loss": 0.3267,
+      "step": 4486
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1366279069767442e-07,
+      "loss": 0.2977,
+      "step": 4487
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1362126245847176e-07,
+      "loss": 0.2793,
+      "step": 4488
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.135797342192691e-07,
+      "loss": 0.2515,
+      "step": 4489
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1353820598006644e-07,
+      "loss": 0.2491,
+      "step": 4490
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1349667774086379e-07,
+      "loss": 0.2937,
+      "step": 4491
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1345514950166113e-07,
+      "loss": 0.2814,
+      "step": 4492
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1341362126245845e-07,
+      "loss": 0.281,
+      "step": 4493
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1337209302325582e-07,
+      "loss": 0.2464,
+      "step": 4494
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.1333056478405314e-07,
+      "loss": 0.2812,
+      "step": 4495
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.132890365448505e-07,
+      "loss": 0.3082,
+      "step": 4496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1324750830564783e-07,
+      "loss": 0.2299,
+      "step": 4497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1320598006644519e-07,
+      "loss": 0.3004,
+      "step": 4498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1316445182724251e-07,
+      "loss": 0.3237,
+      "step": 4499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1312292358803986e-07,
+      "loss": 0.2678,
+      "step": 4500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.130813953488372e-07,
+      "loss": 0.269,
+      "step": 4501
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1303986710963454e-07,
+      "loss": 0.2367,
+      "step": 4502
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1299833887043189e-07,
+      "loss": 0.3112,
+      "step": 4503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1295681063122922e-07,
+      "loss": 0.3116,
+      "step": 4504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1291528239202657e-07,
+      "loss": 0.2702,
+      "step": 4505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1287375415282391e-07,
+      "loss": 0.2594,
+      "step": 4506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1283222591362126e-07,
+      "loss": 0.3112,
+      "step": 4507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.127906976744186e-07,
+      "loss": 0.366,
+      "step": 4508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1274916943521593e-07,
+      "loss": 0.3023,
+      "step": 4509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1270764119601328e-07,
+      "loss": 0.3175,
+      "step": 4510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1266611295681062e-07,
+      "loss": 0.2981,
+      "step": 4511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1262458471760797e-07,
+      "loss": 0.3005,
+      "step": 4512
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1258305647840531e-07,
+      "loss": 0.3041,
+      "step": 4513
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1254152823920266e-07,
+      "loss": 0.3012,
+      "step": 4514
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.125e-07,
+      "loss": 0.2749,
+      "step": 4515
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1245847176079733e-07,
+      "loss": 0.2879,
+      "step": 4516
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1241694352159468e-07,
+      "loss": 0.2697,
+      "step": 4517
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1237541528239202e-07,
+      "loss": 0.3361,
+      "step": 4518
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1233388704318937e-07,
+      "loss": 0.2402,
+      "step": 4519
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.122923588039867e-07,
+      "loss": 0.3191,
+      "step": 4520
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1225083056478406e-07,
+      "loss": 0.2976,
+      "step": 4521
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1220930232558139e-07,
+      "loss": 0.2896,
+      "step": 4522
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1216777408637873e-07,
+      "loss": 0.2984,
+      "step": 4523
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1212624584717608e-07,
+      "loss": 0.2862,
+      "step": 4524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1208471760797342e-07,
+      "loss": 0.3275,
+      "step": 4525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1204318936877077e-07,
+      "loss": 0.2798,
+      "step": 4526
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.120016611295681e-07,
+      "loss": 0.2379,
+      "step": 4527
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1196013289036545e-07,
+      "loss": 0.3001,
+      "step": 4528
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1191860465116279e-07,
+      "loss": 0.2867,
+      "step": 4529
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1187707641196013e-07,
+      "loss": 0.2516,
+      "step": 4530
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1183554817275748e-07,
+      "loss": 0.2628,
+      "step": 4531
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.117940199335548e-07,
+      "loss": 0.2949,
+      "step": 4532
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.1175249169435216e-07,
+      "loss": 0.2753,
+      "step": 4533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1171096345514949e-07,
+      "loss": 0.2922,
+      "step": 4534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1166943521594685e-07,
+      "loss": 0.2523,
+      "step": 4535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1162790697674417e-07,
+      "loss": 0.2836,
+      "step": 4536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1158637873754154e-07,
+      "loss": 0.3094,
+      "step": 4537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1154485049833886e-07,
+      "loss": 0.2818,
+      "step": 4538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.115033222591362e-07,
+      "loss": 0.3066,
+      "step": 4539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1146179401993355e-07,
+      "loss": 0.2476,
+      "step": 4540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1142026578073088e-07,
+      "loss": 0.2541,
+      "step": 4541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1137873754152823e-07,
+      "loss": 0.3069,
+      "step": 4542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1133720930232557e-07,
+      "loss": 0.2859,
+      "step": 4543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1129568106312292e-07,
+      "loss": 0.2822,
+      "step": 4544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1125415282392026e-07,
+      "loss": 0.2617,
+      "step": 4545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.112126245847176e-07,
+      "loss": 0.3549,
+      "step": 4546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1117109634551494e-07,
+      "loss": 0.252,
+      "step": 4547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1112956810631228e-07,
+      "loss": 0.3628,
+      "step": 4548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1108803986710963e-07,
+      "loss": 0.254,
+      "step": 4549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1104651162790697e-07,
+      "loss": 0.2716,
+      "step": 4550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1100498338870432e-07,
+      "loss": 0.292,
+      "step": 4551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1096345514950165e-07,
+      "loss": 0.3226,
+      "step": 4552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1092192691029899e-07,
+      "loss": 0.3663,
+      "step": 4553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1088039867109634e-07,
+      "loss": 0.3353,
+      "step": 4554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1083887043189368e-07,
+      "loss": 0.3304,
+      "step": 4555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1079734219269103e-07,
+      "loss": 0.2814,
+      "step": 4556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1075581395348836e-07,
+      "loss": 0.2802,
+      "step": 4557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1071428571428571e-07,
+      "loss": 0.287,
+      "step": 4558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1067275747508305e-07,
+      "loss": 0.3143,
+      "step": 4559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1063122923588039e-07,
+      "loss": 0.311,
+      "step": 4560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1058970099667774e-07,
+      "loss": 0.3243,
+      "step": 4561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1054817275747508e-07,
+      "loss": 0.2131,
+      "step": 4562
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1050664451827243e-07,
+      "loss": 0.2916,
+      "step": 4563
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1046511627906976e-07,
+      "loss": 0.2752,
+      "step": 4564
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1042358803986711e-07,
+      "loss": 0.2971,
+      "step": 4565
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1038205980066445e-07,
+      "loss": 0.357,
+      "step": 4566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.103405315614618e-07,
+      "loss": 0.3403,
+      "step": 4567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1029900332225914e-07,
+      "loss": 0.3237,
+      "step": 4568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.1025747508305647e-07,
+      "loss": 0.2546,
+      "step": 4569
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1021594684385382e-07,
+      "loss": 0.2649,
+      "step": 4570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1017441860465116e-07,
+      "loss": 0.3672,
+      "step": 4571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1013289036544851e-07,
+      "loss": 0.2722,
+      "step": 4572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1009136212624585e-07,
+      "loss": 0.2676,
+      "step": 4573
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.100498338870432e-07,
+      "loss": 0.3216,
+      "step": 4574
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.1000830564784053e-07,
+      "loss": 0.2959,
+      "step": 4575
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0996677740863786e-07,
+      "loss": 0.244,
+      "step": 4576
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0992524916943522e-07,
+      "loss": 0.3124,
+      "step": 4577
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0988372093023254e-07,
+      "loss": 0.343,
+      "step": 4578
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.098421926910299e-07,
+      "loss": 0.2882,
+      "step": 4579
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0980066445182723e-07,
+      "loss": 0.2274,
+      "step": 4580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0975913621262459e-07,
+      "loss": 0.2753,
+      "step": 4581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0971760797342192e-07,
+      "loss": 0.2969,
+      "step": 4582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0967607973421925e-07,
+      "loss": 0.2652,
+      "step": 4583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.096345514950166e-07,
+      "loss": 0.2962,
+      "step": 4584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0959302325581394e-07,
+      "loss": 0.2952,
+      "step": 4585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0955149501661129e-07,
+      "loss": 0.2707,
+      "step": 4586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0950996677740863e-07,
+      "loss": 0.2984,
+      "step": 4587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0946843853820598e-07,
+      "loss": 0.2521,
+      "step": 4588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0942691029900331e-07,
+      "loss": 0.3894,
+      "step": 4589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0938538205980065e-07,
+      "loss": 0.3033,
+      "step": 4590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.09343853820598e-07,
+      "loss": 0.2547,
+      "step": 4591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0930232558139534e-07,
+      "loss": 0.3156,
+      "step": 4592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0926079734219269e-07,
+      "loss": 0.2625,
+      "step": 4593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0921926910299002e-07,
+      "loss": 0.2724,
+      "step": 4594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0917774086378737e-07,
+      "loss": 0.248,
+      "step": 4595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0913621262458471e-07,
+      "loss": 0.349,
+      "step": 4596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0909468438538206e-07,
+      "loss": 0.2544,
+      "step": 4597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.090531561461794e-07,
+      "loss": 0.3119,
+      "step": 4598
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0901162790697673e-07,
+      "loss": 0.3366,
+      "step": 4599
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0897009966777408e-07,
+      "loss": 0.2712,
+      "step": 4600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0892857142857142e-07,
+      "loss": 0.3081,
+      "step": 4601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0888704318936877e-07,
+      "loss": 0.2665,
+      "step": 4602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0884551495016611e-07,
+      "loss": 0.2805,
+      "step": 4603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.0880398671096346e-07,
+      "loss": 0.2918,
+      "step": 4604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.087624584717608e-07,
+      "loss": 0.296,
+      "step": 4605
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0872093023255813e-07,
+      "loss": 0.3195,
+      "step": 4606
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0867940199335548e-07,
+      "loss": 0.2646,
+      "step": 4607
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0863787375415282e-07,
+      "loss": 0.2722,
+      "step": 4608
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0859634551495017e-07,
+      "loss": 0.3057,
+      "step": 4609
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.085548172757475e-07,
+      "loss": 0.272,
+      "step": 4610
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0851328903654486e-07,
+      "loss": 0.3051,
+      "step": 4611
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0847176079734219e-07,
+      "loss": 0.2818,
+      "step": 4612
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0843023255813953e-07,
+      "loss": 0.3221,
+      "step": 4613
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0838870431893688e-07,
+      "loss": 0.2504,
+      "step": 4614
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.083471760797342e-07,
+      "loss": 0.3176,
+      "step": 4615
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0830564784053157e-07,
+      "loss": 0.3119,
+      "step": 4616
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0826411960132889e-07,
+      "loss": 0.3306,
+      "step": 4617
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0822259136212625e-07,
+      "loss": 0.3179,
+      "step": 4618
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0818106312292358e-07,
+      "loss": 0.3677,
+      "step": 4619
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0813953488372091e-07,
+      "loss": 0.2379,
+      "step": 4620
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0809800664451826e-07,
+      "loss": 0.3007,
+      "step": 4621
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.080564784053156e-07,
+      "loss": 0.2792,
+      "step": 4622
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0801495016611295e-07,
+      "loss": 0.2913,
+      "step": 4623
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0797342192691029e-07,
+      "loss": 0.3404,
+      "step": 4624
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0793189368770764e-07,
+      "loss": 0.2847,
+      "step": 4625
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0789036544850497e-07,
+      "loss": 0.3346,
+      "step": 4626
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0784883720930231e-07,
+      "loss": 0.259,
+      "step": 4627
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0780730897009966e-07,
+      "loss": 0.2723,
+      "step": 4628
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.07765780730897e-07,
+      "loss": 0.3065,
+      "step": 4629
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0772425249169435e-07,
+      "loss": 0.3071,
+      "step": 4630
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0768272425249168e-07,
+      "loss": 0.2684,
+      "step": 4631
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0764119601328903e-07,
+      "loss": 0.2971,
+      "step": 4632
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0759966777408637e-07,
+      "loss": 0.3165,
+      "step": 4633
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0755813953488372e-07,
+      "loss": 0.2889,
+      "step": 4634
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0751661129568106e-07,
+      "loss": 0.3386,
+      "step": 4635
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.074750830564784e-07,
+      "loss": 0.2545,
+      "step": 4636
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0743355481727574e-07,
+      "loss": 0.287,
+      "step": 4637
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0739202657807308e-07,
+      "loss": 0.3174,
+      "step": 4638
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0735049833887043e-07,
+      "loss": 0.3034,
+      "step": 4639
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0730897009966777e-07,
+      "loss": 0.3514,
+      "step": 4640
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.0726744186046512e-07,
+      "loss": 0.2459,
+      "step": 4641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0722591362126245e-07,
+      "loss": 0.256,
+      "step": 4642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0718438538205979e-07,
+      "loss": 0.2681,
+      "step": 4643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0714285714285714e-07,
+      "loss": 0.2738,
+      "step": 4644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0710132890365448e-07,
+      "loss": 0.2817,
+      "step": 4645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0705980066445183e-07,
+      "loss": 0.2638,
+      "step": 4646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0701827242524917e-07,
+      "loss": 0.2526,
+      "step": 4647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0697674418604652e-07,
+      "loss": 0.2375,
+      "step": 4648
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0693521594684385e-07,
+      "loss": 0.3535,
+      "step": 4649
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0689368770764119e-07,
+      "loss": 0.3002,
+      "step": 4650
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0685215946843854e-07,
+      "loss": 0.2832,
+      "step": 4651
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0681063122923588e-07,
+      "loss": 0.3068,
+      "step": 4652
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0676910299003323e-07,
+      "loss": 0.3033,
+      "step": 4653
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0672757475083056e-07,
+      "loss": 0.3269,
+      "step": 4654
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0668604651162791e-07,
+      "loss": 0.295,
+      "step": 4655
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0664451827242525e-07,
+      "loss": 0.2661,
+      "step": 4656
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0660299003322257e-07,
+      "loss": 0.3282,
+      "step": 4657
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0656146179401994e-07,
+      "loss": 0.2568,
+      "step": 4658
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0651993355481726e-07,
+      "loss": 0.2909,
+      "step": 4659
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0647840531561462e-07,
+      "loss": 0.2377,
+      "step": 4660
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0643687707641195e-07,
+      "loss": 0.3268,
+      "step": 4661
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0639534883720931e-07,
+      "loss": 0.3062,
+      "step": 4662
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0635382059800663e-07,
+      "loss": 0.2908,
+      "step": 4663
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0631229235880398e-07,
+      "loss": 0.2645,
+      "step": 4664
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0627076411960132e-07,
+      "loss": 0.3127,
+      "step": 4665
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0622923588039866e-07,
+      "loss": 0.2597,
+      "step": 4666
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0618770764119601e-07,
+      "loss": 0.2446,
+      "step": 4667
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0614617940199334e-07,
+      "loss": 0.2666,
+      "step": 4668
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.061046511627907e-07,
+      "loss": 0.2239,
+      "step": 4669
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0606312292358803e-07,
+      "loss": 0.254,
+      "step": 4670
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0602159468438538e-07,
+      "loss": 0.3035,
+      "step": 4671
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0598006644518272e-07,
+      "loss": 0.2446,
+      "step": 4672
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0593853820598005e-07,
+      "loss": 0.3037,
+      "step": 4673
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.058970099667774e-07,
+      "loss": 0.2649,
+      "step": 4674
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0585548172757474e-07,
+      "loss": 0.2436,
+      "step": 4675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0581395348837209e-07,
+      "loss": 0.3001,
+      "step": 4676
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.0577242524916943e-07,
+      "loss": 0.2982,
+      "step": 4677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0573089700996678e-07,
+      "loss": 0.2868,
+      "step": 4678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0568936877076411e-07,
+      "loss": 0.2938,
+      "step": 4679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0564784053156145e-07,
+      "loss": 0.2647,
+      "step": 4680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.056063122923588e-07,
+      "loss": 0.285,
+      "step": 4681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0556478405315614e-07,
+      "loss": 0.2865,
+      "step": 4682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0552325581395349e-07,
+      "loss": 0.2817,
+      "step": 4683
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0548172757475083e-07,
+      "loss": 0.2593,
+      "step": 4684
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0544019933554818e-07,
+      "loss": 0.2784,
+      "step": 4685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0539867109634551e-07,
+      "loss": 0.2651,
+      "step": 4686
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0535714285714285e-07,
+      "loss": 0.2834,
+      "step": 4687
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.053156146179402e-07,
+      "loss": 0.3148,
+      "step": 4688
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0527408637873754e-07,
+      "loss": 0.3175,
+      "step": 4689
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0523255813953489e-07,
+      "loss": 0.2835,
+      "step": 4690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0519102990033222e-07,
+      "loss": 0.3278,
+      "step": 4691
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0514950166112957e-07,
+      "loss": 0.3389,
+      "step": 4692
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0510797342192691e-07,
+      "loss": 0.3385,
+      "step": 4693
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0506644518272426e-07,
+      "loss": 0.2859,
+      "step": 4694
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.050249169435216e-07,
+      "loss": 0.3564,
+      "step": 4695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0498338870431892e-07,
+      "loss": 0.2882,
+      "step": 4696
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0494186046511628e-07,
+      "loss": 0.2481,
+      "step": 4697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.049003322259136e-07,
+      "loss": 0.2698,
+      "step": 4698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0485880398671097e-07,
+      "loss": 0.2951,
+      "step": 4699
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0481727574750829e-07,
+      "loss": 0.3256,
+      "step": 4700
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0477574750830566e-07,
+      "loss": 0.3306,
+      "step": 4701
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0473421926910298e-07,
+      "loss": 0.2995,
+      "step": 4702
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0469269102990032e-07,
+      "loss": 0.2813,
+      "step": 4703
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0465116279069767e-07,
+      "loss": 0.2673,
+      "step": 4704
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.04609634551495e-07,
+      "loss": 0.3291,
+      "step": 4705
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0456810631229235e-07,
+      "loss": 0.2941,
+      "step": 4706
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0452657807308969e-07,
+      "loss": 0.2664,
+      "step": 4707
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0448504983388704e-07,
+      "loss": 0.2874,
+      "step": 4708
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0444352159468438e-07,
+      "loss": 0.3041,
+      "step": 4709
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0440199335548171e-07,
+      "loss": 0.3358,
+      "step": 4710
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0436046511627906e-07,
+      "loss": 0.2049,
+      "step": 4711
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.043189368770764e-07,
+      "loss": 0.2464,
+      "step": 4712
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.0427740863787375e-07,
+      "loss": 0.2533,
+      "step": 4713
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0423588039867109e-07,
+      "loss": 0.2601,
+      "step": 4714
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0419435215946844e-07,
+      "loss": 0.292,
+      "step": 4715
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0415282392026577e-07,
+      "loss": 0.3453,
+      "step": 4716
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0411129568106311e-07,
+      "loss": 0.2575,
+      "step": 4717
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0406976744186046e-07,
+      "loss": 0.3152,
+      "step": 4718
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.040282392026578e-07,
+      "loss": 0.3403,
+      "step": 4719
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0398671096345515e-07,
+      "loss": 0.2648,
+      "step": 4720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0394518272425248e-07,
+      "loss": 0.2189,
+      "step": 4721
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0390365448504983e-07,
+      "loss": 0.2852,
+      "step": 4722
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0386212624584717e-07,
+      "loss": 0.2505,
+      "step": 4723
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0382059800664451e-07,
+      "loss": 0.2497,
+      "step": 4724
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0377906976744186e-07,
+      "loss": 0.3096,
+      "step": 4725
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.037375415282392e-07,
+      "loss": 0.296,
+      "step": 4726
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0369601328903655e-07,
+      "loss": 0.2966,
+      "step": 4727
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0365448504983388e-07,
+      "loss": 0.3315,
+      "step": 4728
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0361295681063123e-07,
+      "loss": 0.2662,
+      "step": 4729
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0357142857142857e-07,
+      "loss": 0.2533,
+      "step": 4730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0352990033222592e-07,
+      "loss": 0.2928,
+      "step": 4731
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0348837209302326e-07,
+      "loss": 0.2933,
+      "step": 4732
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0344684385382059e-07,
+      "loss": 0.2442,
+      "step": 4733
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0340531561461794e-07,
+      "loss": 0.3496,
+      "step": 4734
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0336378737541528e-07,
+      "loss": 0.2751,
+      "step": 4735
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0332225913621263e-07,
+      "loss": 0.2973,
+      "step": 4736
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0328073089700997e-07,
+      "loss": 0.2894,
+      "step": 4737
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0323920265780732e-07,
+      "loss": 0.3312,
+      "step": 4738
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0319767441860465e-07,
+      "loss": 0.2461,
+      "step": 4739
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0315614617940198e-07,
+      "loss": 0.2515,
+      "step": 4740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0311461794019934e-07,
+      "loss": 0.2273,
+      "step": 4741
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0307308970099666e-07,
+      "loss": 0.2935,
+      "step": 4742
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0303156146179403e-07,
+      "loss": 0.3259,
+      "step": 4743
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0299003322259135e-07,
+      "loss": 0.2772,
+      "step": 4744
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0294850498338871e-07,
+      "loss": 0.2689,
+      "step": 4745
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0290697674418604e-07,
+      "loss": 0.2668,
+      "step": 4746
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0286544850498337e-07,
+      "loss": 0.2509,
+      "step": 4747
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0282392026578072e-07,
+      "loss": 0.2762,
+      "step": 4748
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.0278239202657806e-07,
+      "loss": 0.3147,
+      "step": 4749
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0274086378737541e-07,
+      "loss": 0.2594,
+      "step": 4750
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0269933554817275e-07,
+      "loss": 0.3214,
+      "step": 4751
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.026578073089701e-07,
+      "loss": 0.3261,
+      "step": 4752
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0261627906976743e-07,
+      "loss": 0.2327,
+      "step": 4753
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0257475083056477e-07,
+      "loss": 0.3134,
+      "step": 4754
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0253322259136212e-07,
+      "loss": 0.2964,
+      "step": 4755
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0249169435215946e-07,
+      "loss": 0.3037,
+      "step": 4756
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0245016611295681e-07,
+      "loss": 0.3025,
+      "step": 4757
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0240863787375414e-07,
+      "loss": 0.3013,
+      "step": 4758
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.023671096345515e-07,
+      "loss": 0.2732,
+      "step": 4759
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0232558139534883e-07,
+      "loss": 0.3519,
+      "step": 4760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0228405315614618e-07,
+      "loss": 0.3275,
+      "step": 4761
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0224252491694352e-07,
+      "loss": 0.2797,
+      "step": 4762
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0220099667774085e-07,
+      "loss": 0.2408,
+      "step": 4763
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.021594684385382e-07,
+      "loss": 0.2535,
+      "step": 4764
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0211794019933554e-07,
+      "loss": 0.2801,
+      "step": 4765
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0207641196013289e-07,
+      "loss": 0.2924,
+      "step": 4766
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0203488372093023e-07,
+      "loss": 0.3185,
+      "step": 4767
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0199335548172758e-07,
+      "loss": 0.3073,
+      "step": 4768
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0195182724252492e-07,
+      "loss": 0.2368,
+      "step": 4769
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0191029900332225e-07,
+      "loss": 0.3294,
+      "step": 4770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.018687707641196e-07,
+      "loss": 0.2607,
+      "step": 4771
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0182724252491694e-07,
+      "loss": 0.2704,
+      "step": 4772
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0178571428571429e-07,
+      "loss": 0.2961,
+      "step": 4773
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0174418604651163e-07,
+      "loss": 0.3223,
+      "step": 4774
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0170265780730898e-07,
+      "loss": 0.3167,
+      "step": 4775
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0166112956810631e-07,
+      "loss": 0.3053,
+      "step": 4776
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0161960132890365e-07,
+      "loss": 0.3324,
+      "step": 4777
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.01578073089701e-07,
+      "loss": 0.3302,
+      "step": 4778
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0153654485049832e-07,
+      "loss": 0.3067,
+      "step": 4779
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0149501661129569e-07,
+      "loss": 0.2927,
+      "step": 4780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0145348837209301e-07,
+      "loss": 0.3024,
+      "step": 4781
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0141196013289037e-07,
+      "loss": 0.2448,
+      "step": 4782
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.013704318936877e-07,
+      "loss": 0.3099,
+      "step": 4783
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0132890365448503e-07,
+      "loss": 0.2746,
+      "step": 4784
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.0128737541528238e-07,
+      "loss": 0.2773,
+      "step": 4785
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0124584717607972e-07,
+      "loss": 0.2566,
+      "step": 4786
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0120431893687707e-07,
+      "loss": 0.3213,
+      "step": 4787
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0116279069767441e-07,
+      "loss": 0.3021,
+      "step": 4788
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0112126245847176e-07,
+      "loss": 0.2942,
+      "step": 4789
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.010797342192691e-07,
+      "loss": 0.3186,
+      "step": 4790
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0103820598006644e-07,
+      "loss": 0.2528,
+      "step": 4791
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0099667774086378e-07,
+      "loss": 0.2943,
+      "step": 4792
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0095514950166112e-07,
+      "loss": 0.3351,
+      "step": 4793
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0091362126245847e-07,
+      "loss": 0.335,
+      "step": 4794
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.008720930232558e-07,
+      "loss": 0.3039,
+      "step": 4795
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0083056478405315e-07,
+      "loss": 0.2354,
+      "step": 4796
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0078903654485049e-07,
+      "loss": 0.3309,
+      "step": 4797
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0074750830564784e-07,
+      "loss": 0.2974,
+      "step": 4798
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0070598006644518e-07,
+      "loss": 0.2665,
+      "step": 4799
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0066445182724251e-07,
+      "loss": 0.2312,
+      "step": 4800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0062292358803986e-07,
+      "loss": 0.2594,
+      "step": 4801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.005813953488372e-07,
+      "loss": 0.231,
+      "step": 4802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0053986710963455e-07,
+      "loss": 0.2589,
+      "step": 4803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0049833887043189e-07,
+      "loss": 0.2518,
+      "step": 4804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0045681063122924e-07,
+      "loss": 0.2595,
+      "step": 4805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0041528239202657e-07,
+      "loss": 0.2847,
+      "step": 4806
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0037375415282391e-07,
+      "loss": 0.2768,
+      "step": 4807
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0033222591362126e-07,
+      "loss": 0.2726,
+      "step": 4808
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.002906976744186e-07,
+      "loss": 0.3106,
+      "step": 4809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0024916943521595e-07,
+      "loss": 0.3128,
+      "step": 4810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0020764119601329e-07,
+      "loss": 0.3394,
+      "step": 4811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0016611295681064e-07,
+      "loss": 0.294,
+      "step": 4812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0012458471760797e-07,
+      "loss": 0.2374,
+      "step": 4813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0008305647840531e-07,
+      "loss": 0.2662,
+      "step": 4814
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.0004152823920266e-07,
+      "loss": 0.2704,
+      "step": 4815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1e-07,
+      "loss": 0.307,
+      "step": 4816
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.995847176079735e-08,
+      "loss": 0.3592,
+      "step": 4817
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.991694352159468e-08,
+      "loss": 0.3117,
+      "step": 4818
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.987541528239203e-08,
+      "loss": 0.28,
+      "step": 4819
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.983388704318937e-08,
+      "loss": 0.2766,
+      "step": 4820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.979235880398669e-08,
+      "loss": 0.2742,
+      "step": 4821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.975083056478406e-08,
+      "loss": 0.2596,
+      "step": 4822
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.970930232558138e-08,
+      "loss": 0.3039,
+      "step": 4823
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.966777408637874e-08,
+      "loss": 0.3123,
+      "step": 4824
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.962624584717607e-08,
+      "loss": 0.2817,
+      "step": 4825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.958471760797343e-08,
+      "loss": 0.2739,
+      "step": 4826
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.954318936877075e-08,
+      "loss": 0.3023,
+      "step": 4827
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.95016611295681e-08,
+      "loss": 0.2708,
+      "step": 4828
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.946013289036544e-08,
+      "loss": 0.2577,
+      "step": 4829
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.941860465116278e-08,
+      "loss": 0.2352,
+      "step": 4830
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.937707641196013e-08,
+      "loss": 0.2644,
+      "step": 4831
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.933554817275746e-08,
+      "loss": 0.2661,
+      "step": 4832
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.929401993355481e-08,
+      "loss": 0.2789,
+      "step": 4833
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.925249169435215e-08,
+      "loss": 0.26,
+      "step": 4834
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.92109634551495e-08,
+      "loss": 0.2955,
+      "step": 4835
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.916943521594684e-08,
+      "loss": 0.2975,
+      "step": 4836
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.912790697674417e-08,
+      "loss": 0.3136,
+      "step": 4837
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.908637873754152e-08,
+      "loss": 0.3319,
+      "step": 4838
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.904485049833886e-08,
+      "loss": 0.2471,
+      "step": 4839
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.900332225913621e-08,
+      "loss": 0.2425,
+      "step": 4840
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.896179401993355e-08,
+      "loss": 0.3071,
+      "step": 4841
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.89202657807309e-08,
+      "loss": 0.2697,
+      "step": 4842
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.887873754152823e-08,
+      "loss": 0.3164,
+      "step": 4843
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.883720930232557e-08,
+      "loss": 0.2965,
+      "step": 4844
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.879568106312292e-08,
+      "loss": 0.2935,
+      "step": 4845
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.875415282392026e-08,
+      "loss": 0.2729,
+      "step": 4846
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.871262458471761e-08,
+      "loss": 0.3048,
+      "step": 4847
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.867109634551494e-08,
+      "loss": 0.3264,
+      "step": 4848
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.86295681063123e-08,
+      "loss": 0.3031,
+      "step": 4849
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.858803986710963e-08,
+      "loss": 0.3081,
+      "step": 4850
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.854651162790697e-08,
+      "loss": 0.3197,
+      "step": 4851
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.850498338870432e-08,
+      "loss": 0.3012,
+      "step": 4852
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.846345514950166e-08,
+      "loss": 0.2742,
+      "step": 4853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.8421926910299e-08,
+      "loss": 0.2732,
+      "step": 4854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.838039867109634e-08,
+      "loss": 0.2534,
+      "step": 4855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.833887043189369e-08,
+      "loss": 0.2753,
+      "step": 4856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.829734219269103e-08,
+      "loss": 0.2596,
+      "step": 4857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.825581395348838e-08,
+      "loss": 0.3117,
+      "step": 4858
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.821428571428572e-08,
+      "loss": 0.2806,
+      "step": 4859
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.817275747508304e-08,
+      "loss": 0.2551,
+      "step": 4860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.81312292358804e-08,
+      "loss": 0.3428,
+      "step": 4861
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.808970099667773e-08,
+      "loss": 0.3139,
+      "step": 4862
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.804817275747509e-08,
+      "loss": 0.315,
+      "step": 4863
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.800664451827241e-08,
+      "loss": 0.3239,
+      "step": 4864
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.796511627906978e-08,
+      "loss": 0.2894,
+      "step": 4865
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.79235880398671e-08,
+      "loss": 0.2689,
+      "step": 4866
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.788205980066444e-08,
+      "loss": 0.29,
+      "step": 4867
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.784053156146179e-08,
+      "loss": 0.346,
+      "step": 4868
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.779900332225912e-08,
+      "loss": 0.2796,
+      "step": 4869
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.775747508305647e-08,
+      "loss": 0.269,
+      "step": 4870
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.771594684385381e-08,
+      "loss": 0.2667,
+      "step": 4871
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.767441860465116e-08,
+      "loss": 0.2605,
+      "step": 4872
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.76328903654485e-08,
+      "loss": 0.3236,
+      "step": 4873
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.759136212624583e-08,
+      "loss": 0.2871,
+      "step": 4874
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.754983388704318e-08,
+      "loss": 0.366,
+      "step": 4875
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.750830564784052e-08,
+      "loss": 0.2163,
+      "step": 4876
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.746677740863787e-08,
+      "loss": 0.2909,
+      "step": 4877
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.742524916943521e-08,
+      "loss": 0.3368,
+      "step": 4878
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.738372093023256e-08,
+      "loss": 0.3079,
+      "step": 4879
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.73421926910299e-08,
+      "loss": 0.2156,
+      "step": 4880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.730066445182723e-08,
+      "loss": 0.3023,
+      "step": 4881
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.725913621262458e-08,
+      "loss": 0.2301,
+      "step": 4882
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.721760797342192e-08,
+      "loss": 0.2816,
+      "step": 4883
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.717607973421927e-08,
+      "loss": 0.2882,
+      "step": 4884
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.71345514950166e-08,
+      "loss": 0.2654,
+      "step": 4885
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.709302325581395e-08,
+      "loss": 0.264,
+      "step": 4886
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.705149501661129e-08,
+      "loss": 0.3675,
+      "step": 4887
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.700996677740864e-08,
+      "loss": 0.2812,
+      "step": 4888
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.696843853820598e-08,
+      "loss": 0.3216,
+      "step": 4889
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.692691029900331e-08,
+      "loss": 0.244,
+      "step": 4890
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.688538205980066e-08,
+      "loss": 0.2791,
+      "step": 4891
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.6843853820598e-08,
+      "loss": 0.2798,
+      "step": 4892
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.680232558139535e-08,
+      "loss": 0.262,
+      "step": 4893
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.676079734219269e-08,
+      "loss": 0.3083,
+      "step": 4894
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.671926910299004e-08,
+      "loss": 0.3632,
+      "step": 4895
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.667774086378738e-08,
+      "loss": 0.2773,
+      "step": 4896
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.663621262458471e-08,
+      "loss": 0.3573,
+      "step": 4897
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.659468438538206e-08,
+      "loss": 0.2368,
+      "step": 4898
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.65531561461794e-08,
+      "loss": 0.2633,
+      "step": 4899
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.651162790697675e-08,
+      "loss": 0.2988,
+      "step": 4900
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.647009966777409e-08,
+      "loss": 0.2501,
+      "step": 4901
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.642857142857144e-08,
+      "loss": 0.3363,
+      "step": 4902
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.638704318936877e-08,
+      "loss": 0.359,
+      "step": 4903
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.63455149501661e-08,
+      "loss": 0.2512,
+      "step": 4904
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.630398671096346e-08,
+      "loss": 0.3248,
+      "step": 4905
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.626245847176078e-08,
+      "loss": 0.2708,
+      "step": 4906
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.622093023255815e-08,
+      "loss": 0.2679,
+      "step": 4907
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.617940199335547e-08,
+      "loss": 0.2547,
+      "step": 4908
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.613787375415283e-08,
+      "loss": 0.2837,
+      "step": 4909
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.609634551495016e-08,
+      "loss": 0.35,
+      "step": 4910
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.605481727574749e-08,
+      "loss": 0.3765,
+      "step": 4911
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.601328903654484e-08,
+      "loss": 0.289,
+      "step": 4912
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.597176079734218e-08,
+      "loss": 0.2844,
+      "step": 4913
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.593023255813953e-08,
+      "loss": 0.349,
+      "step": 4914
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.588870431893687e-08,
+      "loss": 0.2783,
+      "step": 4915
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.584717607973422e-08,
+      "loss": 0.3451,
+      "step": 4916
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.580564784053155e-08,
+      "loss": 0.3169,
+      "step": 4917
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.576411960132889e-08,
+      "loss": 0.2355,
+      "step": 4918
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.572259136212624e-08,
+      "loss": 0.3101,
+      "step": 4919
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.568106312292358e-08,
+      "loss": 0.3037,
+      "step": 4920
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.563953488372093e-08,
+      "loss": 0.3307,
+      "step": 4921
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.559800664451826e-08,
+      "loss": 0.2846,
+      "step": 4922
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.555647840531561e-08,
+      "loss": 0.311,
+      "step": 4923
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.551495016611295e-08,
+      "loss": 0.2821,
+      "step": 4924
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.54734219269103e-08,
+      "loss": 0.2873,
+      "step": 4925
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.543189368770764e-08,
+      "loss": 0.2631,
+      "step": 4926
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.539036544850497e-08,
+      "loss": 0.2607,
+      "step": 4927
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.534883720930232e-08,
+      "loss": 0.3414,
+      "step": 4928
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.530730897009966e-08,
+      "loss": 0.3027,
+      "step": 4929
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.526578073089701e-08,
+      "loss": 0.3421,
+      "step": 4930
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.522425249169435e-08,
+      "loss": 0.326,
+      "step": 4931
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.51827242524917e-08,
+      "loss": 0.3185,
+      "step": 4932
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.514119601328903e-08,
+      "loss": 0.3085,
+      "step": 4933
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.509966777408637e-08,
+      "loss": 0.2477,
+      "step": 4934
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.505813953488372e-08,
+      "loss": 0.2612,
+      "step": 4935
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.501661129568106e-08,
+      "loss": 0.3482,
+      "step": 4936
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.497508305647841e-08,
+      "loss": 0.2969,
+      "step": 4937
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.493355481727575e-08,
+      "loss": 0.2768,
+      "step": 4938
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.48920265780731e-08,
+      "loss": 0.2532,
+      "step": 4939
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.485049833887043e-08,
+      "loss": 0.3432,
+      "step": 4940
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.480897009966777e-08,
+      "loss": 0.3075,
+      "step": 4941
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.476744186046512e-08,
+      "loss": 0.2961,
+      "step": 4942
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.472591362126244e-08,
+      "loss": 0.2192,
+      "step": 4943
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.46843853820598e-08,
+      "loss": 0.258,
+      "step": 4944
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.464285714285713e-08,
+      "loss": 0.3345,
+      "step": 4945
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.460132890365449e-08,
+      "loss": 0.3075,
+      "step": 4946
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.455980066445182e-08,
+      "loss": 0.3151,
+      "step": 4947
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.451827242524915e-08,
+      "loss": 0.346,
+      "step": 4948
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.44767441860465e-08,
+      "loss": 0.2373,
+      "step": 4949
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.443521594684384e-08,
+      "loss": 0.2773,
+      "step": 4950
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.439368770764119e-08,
+      "loss": 0.26,
+      "step": 4951
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.435215946843853e-08,
+      "loss": 0.3207,
+      "step": 4952
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.431063122923588e-08,
+      "loss": 0.3016,
+      "step": 4953
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.426910299003321e-08,
+      "loss": 0.2636,
+      "step": 4954
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.422757475083056e-08,
+      "loss": 0.3346,
+      "step": 4955
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.41860465116279e-08,
+      "loss": 0.3197,
+      "step": 4956
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.414451827242524e-08,
+      "loss": 0.2498,
+      "step": 4957
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.410299003322259e-08,
+      "loss": 0.3085,
+      "step": 4958
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.406146179401992e-08,
+      "loss": 0.2929,
+      "step": 4959
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.401993355481727e-08,
+      "loss": 0.3252,
+      "step": 4960
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.397840531561461e-08,
+      "loss": 0.2914,
+      "step": 4961
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.393687707641196e-08,
+      "loss": 0.3121,
+      "step": 4962
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.38953488372093e-08,
+      "loss": 0.295,
+      "step": 4963
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.385382059800663e-08,
+      "loss": 0.2853,
+      "step": 4964
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.381229235880398e-08,
+      "loss": 0.2547,
+      "step": 4965
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.377076411960132e-08,
+      "loss": 0.2962,
+      "step": 4966
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.372923588039867e-08,
+      "loss": 0.3109,
+      "step": 4967
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.368770764119601e-08,
+      "loss": 0.365,
+      "step": 4968
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.364617940199336e-08,
+      "loss": 0.3276,
+      "step": 4969
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.36046511627907e-08,
+      "loss": 0.2912,
+      "step": 4970
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.356312292358803e-08,
+      "loss": 0.2356,
+      "step": 4971
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.352159468438538e-08,
+      "loss": 0.3027,
+      "step": 4972
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.348006644518272e-08,
+      "loss": 0.331,
+      "step": 4973
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.343853820598007e-08,
+      "loss": 0.314,
+      "step": 4974
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.33970099667774e-08,
+      "loss": 0.3073,
+      "step": 4975
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.335548172757475e-08,
+      "loss": 0.2497,
+      "step": 4976
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.331395348837209e-08,
+      "loss": 0.3075,
+      "step": 4977
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.327242524916943e-08,
+      "loss": 0.2787,
+      "step": 4978
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.323089700996678e-08,
+      "loss": 0.2691,
+      "step": 4979
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.318936877076412e-08,
+      "loss": 0.2872,
+      "step": 4980
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.314784053156147e-08,
+      "loss": 0.3325,
+      "step": 4981
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.31063122923588e-08,
+      "loss": 0.2866,
+      "step": 4982
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.306478405315615e-08,
+      "loss": 0.3483,
+      "step": 4983
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.302325581395349e-08,
+      "loss": 0.3013,
+      "step": 4984
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.298172757475084e-08,
+      "loss": 0.2843,
+      "step": 4985
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.294019933554818e-08,
+      "loss": 0.3598,
+      "step": 4986
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.28986710963455e-08,
+      "loss": 0.2605,
+      "step": 4987
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.285714285714286e-08,
+      "loss": 0.2344,
+      "step": 4988
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.281561461794019e-08,
+      "loss": 0.2745,
+      "step": 4989
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.277408637873755e-08,
+      "loss": 0.3008,
+      "step": 4990
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.273255813953487e-08,
+      "loss": 0.296,
+      "step": 4991
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.269102990033222e-08,
+      "loss": 0.296,
+      "step": 4992
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.264950166112956e-08,
+      "loss": 0.2367,
+      "step": 4993
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.26079734219269e-08,
+      "loss": 0.2441,
+      "step": 4994
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.256644518272425e-08,
+      "loss": 0.237,
+      "step": 4995
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.252491694352158e-08,
+      "loss": 0.2245,
+      "step": 4996
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.248338870431893e-08,
+      "loss": 0.3849,
+      "step": 4997
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.244186046511627e-08,
+      "loss": 0.273,
+      "step": 4998
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.240033222591362e-08,
+      "loss": 0.311,
+      "step": 4999
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.235880398671096e-08,
+      "loss": 0.2843,
+      "step": 5000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.23172757475083e-08,
+      "loss": 0.2612,
+      "step": 5001
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.227574750830564e-08,
+      "loss": 0.2849,
+      "step": 5002
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.223421926910298e-08,
+      "loss": 0.3658,
+      "step": 5003
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.219269102990033e-08,
+      "loss": 0.309,
+      "step": 5004
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.215116279069767e-08,
+      "loss": 0.2981,
+      "step": 5005
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.210963455149502e-08,
+      "loss": 0.3704,
+      "step": 5006
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.206810631229235e-08,
+      "loss": 0.2843,
+      "step": 5007
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.202657807308969e-08,
+      "loss": 0.233,
+      "step": 5008
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.198504983388704e-08,
+      "loss": 0.2659,
+      "step": 5009
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.194352159468438e-08,
+      "loss": 0.2704,
+      "step": 5010
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.190199335548173e-08,
+      "loss": 0.2683,
+      "step": 5011
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.186046511627906e-08,
+      "loss": 0.266,
+      "step": 5012
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.181893687707641e-08,
+      "loss": 0.2767,
+      "step": 5013
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.177740863787375e-08,
+      "loss": 0.2196,
+      "step": 5014
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.17358803986711e-08,
+      "loss": 0.2981,
+      "step": 5015
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.169435215946844e-08,
+      "loss": 0.3023,
+      "step": 5016
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.165282392026577e-08,
+      "loss": 0.2757,
+      "step": 5017
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.161129568106312e-08,
+      "loss": 0.3245,
+      "step": 5018
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.156976744186046e-08,
+      "loss": 0.2747,
+      "step": 5019
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.152823920265781e-08,
+      "loss": 0.3558,
+      "step": 5020
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.148671096345515e-08,
+      "loss": 0.2932,
+      "step": 5021
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.14451827242525e-08,
+      "loss": 0.3294,
+      "step": 5022
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.140365448504984e-08,
+      "loss": 0.2701,
+      "step": 5023
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.136212624584716e-08,
+      "loss": 0.3126,
+      "step": 5024
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.132059800664452e-08,
+      "loss": 0.2855,
+      "step": 5025
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.127906976744185e-08,
+      "loss": 0.27,
+      "step": 5026
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.123754152823921e-08,
+      "loss": 0.2788,
+      "step": 5027
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.119601328903653e-08,
+      "loss": 0.282,
+      "step": 5028
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.11544850498339e-08,
+      "loss": 0.2818,
+      "step": 5029
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.111295681063122e-08,
+      "loss": 0.325,
+      "step": 5030
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.107142857142856e-08,
+      "loss": 0.2394,
+      "step": 5031
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.10299003322259e-08,
+      "loss": 0.2667,
+      "step": 5032
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.098837209302324e-08,
+      "loss": 0.3257,
+      "step": 5033
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.094684385382059e-08,
+      "loss": 0.3097,
+      "step": 5034
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.090531561461793e-08,
+      "loss": 0.3124,
+      "step": 5035
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.086378737541528e-08,
+      "loss": 0.3549,
+      "step": 5036
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.082225913621262e-08,
+      "loss": 0.2893,
+      "step": 5037
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.078073089700995e-08,
+      "loss": 0.3792,
+      "step": 5038
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.07392026578073e-08,
+      "loss": 0.3114,
+      "step": 5039
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.069767441860464e-08,
+      "loss": 0.2572,
+      "step": 5040
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.065614617940199e-08,
+      "loss": 0.2666,
+      "step": 5041
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.061461794019933e-08,
+      "loss": 0.2719,
+      "step": 5042
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.057308970099668e-08,
+      "loss": 0.2841,
+      "step": 5043
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.053156146179401e-08,
+      "loss": 0.2896,
+      "step": 5044
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.049003322259135e-08,
+      "loss": 0.2647,
+      "step": 5045
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.04485049833887e-08,
+      "loss": 0.3358,
+      "step": 5046
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.040697674418604e-08,
+      "loss": 0.2975,
+      "step": 5047
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.036544850498339e-08,
+      "loss": 0.2566,
+      "step": 5048
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.032392026578072e-08,
+      "loss": 0.2633,
+      "step": 5049
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.028239202657807e-08,
+      "loss": 0.3468,
+      "step": 5050
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.024086378737541e-08,
+      "loss": 0.2739,
+      "step": 5051
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.019933554817276e-08,
+      "loss": 0.3053,
+      "step": 5052
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.01578073089701e-08,
+      "loss": 0.324,
+      "step": 5053
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.011627906976743e-08,
+      "loss": 0.2206,
+      "step": 5054
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.007475083056478e-08,
+      "loss": 0.3331,
+      "step": 5055
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.003322259136212e-08,
+      "loss": 0.3184,
+      "step": 5056
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.999169435215947e-08,
+      "loss": 0.2485,
+      "step": 5057
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.995016611295681e-08,
+      "loss": 0.2469,
+      "step": 5058
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.990863787375416e-08,
+      "loss": 0.3804,
+      "step": 5059
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.98671096345515e-08,
+      "loss": 0.3018,
+      "step": 5060
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.982558139534883e-08,
+      "loss": 0.2996,
+      "step": 5061
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.978405315614618e-08,
+      "loss": 0.2793,
+      "step": 5062
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.974252491694352e-08,
+      "loss": 0.3266,
+      "step": 5063
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.970099667774087e-08,
+      "loss": 0.3188,
+      "step": 5064
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.96594684385382e-08,
+      "loss": 0.3078,
+      "step": 5065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.961794019933556e-08,
+      "loss": 0.2881,
+      "step": 5066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.957641196013289e-08,
+      "loss": 0.3387,
+      "step": 5067
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.953488372093022e-08,
+      "loss": 0.2643,
+      "step": 5068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.949335548172758e-08,
+      "loss": 0.3278,
+      "step": 5069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.94518272425249e-08,
+      "loss": 0.2995,
+      "step": 5070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.941029900332227e-08,
+      "loss": 0.2649,
+      "step": 5071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.936877076411959e-08,
+      "loss": 0.2664,
+      "step": 5072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.932724252491695e-08,
+      "loss": 0.3154,
+      "step": 5073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.928571428571428e-08,
+      "loss": 0.2065,
+      "step": 5074
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.924418604651161e-08,
+      "loss": 0.3303,
+      "step": 5075
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.920265780730896e-08,
+      "loss": 0.2423,
+      "step": 5076
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.91611295681063e-08,
+      "loss": 0.297,
+      "step": 5077
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.911960132890365e-08,
+      "loss": 0.2486,
+      "step": 5078
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.907807308970099e-08,
+      "loss": 0.2913,
+      "step": 5079
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.903654485049834e-08,
+      "loss": 0.3258,
+      "step": 5080
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.899501661129567e-08,
+      "loss": 0.2953,
+      "step": 5081
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.895348837209302e-08,
+      "loss": 0.274,
+      "step": 5082
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.891196013289036e-08,
+      "loss": 0.2484,
+      "step": 5083
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.88704318936877e-08,
+      "loss": 0.3483,
+      "step": 5084
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.882890365448505e-08,
+      "loss": 0.2626,
+      "step": 5085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.878737541528238e-08,
+      "loss": 0.3156,
+      "step": 5086
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.874584717607973e-08,
+      "loss": 0.3114,
+      "step": 5087
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.870431893687707e-08,
+      "loss": 0.2675,
+      "step": 5088
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.866279069767442e-08,
+      "loss": 0.274,
+      "step": 5089
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.862126245847176e-08,
+      "loss": 0.3217,
+      "step": 5090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.85797342192691e-08,
+      "loss": 0.2853,
+      "step": 5091
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.853820598006644e-08,
+      "loss": 0.347,
+      "step": 5092
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.849667774086378e-08,
+      "loss": 0.2712,
+      "step": 5093
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.845514950166113e-08,
+      "loss": 0.2584,
+      "step": 5094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.841362126245847e-08,
+      "loss": 0.3008,
+      "step": 5095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.837209302325582e-08,
+      "loss": 0.2808,
+      "step": 5096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.833056478405315e-08,
+      "loss": 0.2798,
+      "step": 5097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.828903654485049e-08,
+      "loss": 0.2686,
+      "step": 5098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.824750830564784e-08,
+      "loss": 0.2593,
+      "step": 5099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.820598006644518e-08,
+      "loss": 0.3355,
+      "step": 5100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.816445182724253e-08,
+      "loss": 0.2981,
+      "step": 5101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.812292358803986e-08,
+      "loss": 0.2936,
+      "step": 5102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.808139534883721e-08,
+      "loss": 0.2601,
+      "step": 5103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.803986710963455e-08,
+      "loss": 0.3258,
+      "step": 5104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.799833887043189e-08,
+      "loss": 0.2539,
+      "step": 5105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.795681063122924e-08,
+      "loss": 0.2833,
+      "step": 5106
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.791528239202656e-08,
+      "loss": 0.3214,
+      "step": 5107
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.787375415282393e-08,
+      "loss": 0.2813,
+      "step": 5108
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.783222591362125e-08,
+      "loss": 0.3506,
+      "step": 5109
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.779069767441861e-08,
+      "loss": 0.2098,
+      "step": 5110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.774916943521594e-08,
+      "loss": 0.2364,
+      "step": 5111
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.77076411960133e-08,
+      "loss": 0.2848,
+      "step": 5112
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.766611295681062e-08,
+      "loss": 0.3034,
+      "step": 5113
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.762458471760796e-08,
+      "loss": 0.3037,
+      "step": 5114
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.758305647840531e-08,
+      "loss": 0.3325,
+      "step": 5115
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.754152823920265e-08,
+      "loss": 0.339,
+      "step": 5116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.75e-08,
+      "loss": 0.274,
+      "step": 5117
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.745847176079733e-08,
+      "loss": 0.2829,
+      "step": 5118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.741694352159468e-08,
+      "loss": 0.3076,
+      "step": 5119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.737541528239202e-08,
+      "loss": 0.2627,
+      "step": 5120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.733388704318936e-08,
+      "loss": 0.2798,
+      "step": 5121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.72923588039867e-08,
+      "loss": 0.2909,
+      "step": 5122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.725083056478404e-08,
+      "loss": 0.3107,
+      "step": 5123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.720930232558139e-08,
+      "loss": 0.2894,
+      "step": 5124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.716777408637873e-08,
+      "loss": 0.2833,
+      "step": 5125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.712624584717608e-08,
+      "loss": 0.3046,
+      "step": 5126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.708471760797342e-08,
+      "loss": 0.3101,
+      "step": 5127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.704318936877075e-08,
+      "loss": 0.3167,
+      "step": 5128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.70016611295681e-08,
+      "loss": 0.2954,
+      "step": 5129
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.696013289036544e-08,
+      "loss": 0.3306,
+      "step": 5130
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.691860465116279e-08,
+      "loss": 0.3357,
+      "step": 5131
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.687707641196013e-08,
+      "loss": 0.3729,
+      "step": 5132
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.683554817275748e-08,
+      "loss": 0.3077,
+      "step": 5133
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.679401993355481e-08,
+      "loss": 0.3092,
+      "step": 5134
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.675249169435215e-08,
+      "loss": 0.2328,
+      "step": 5135
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.67109634551495e-08,
+      "loss": 0.2993,
+      "step": 5136
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.666943521594684e-08,
+      "loss": 0.2708,
+      "step": 5137
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.662790697674419e-08,
+      "loss": 0.3532,
+      "step": 5138
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.658637873754152e-08,
+      "loss": 0.3215,
+      "step": 5139
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.654485049833887e-08,
+      "loss": 0.2932,
+      "step": 5140
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.650332225913621e-08,
+      "loss": 0.3166,
+      "step": 5141
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.646179401993355e-08,
+      "loss": 0.3291,
+      "step": 5142
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.64202657807309e-08,
+      "loss": 0.2859,
+      "step": 5143
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.637873754152823e-08,
+      "loss": 0.312,
+      "step": 5144
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.633720930232558e-08,
+      "loss": 0.2592,
+      "step": 5145
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.629568106312292e-08,
+      "loss": 0.3451,
+      "step": 5146
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.625415282392027e-08,
+      "loss": 0.3296,
+      "step": 5147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.621262458471761e-08,
+      "loss": 0.3731,
+      "step": 5148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.617109634551496e-08,
+      "loss": 0.2821,
+      "step": 5149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.61295681063123e-08,
+      "loss": 0.3045,
+      "step": 5150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.608803986710962e-08,
+      "loss": 0.3435,
+      "step": 5151
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.604651162790698e-08,
+      "loss": 0.3183,
+      "step": 5152
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.60049833887043e-08,
+      "loss": 0.2824,
+      "step": 5153
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.596345514950167e-08,
+      "loss": 0.2919,
+      "step": 5154
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.592192691029899e-08,
+      "loss": 0.2268,
+      "step": 5155
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.588039867109634e-08,
+      "loss": 0.2696,
+      "step": 5156
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.583887043189368e-08,
+      "loss": 0.2146,
+      "step": 5157
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.579734219269102e-08,
+      "loss": 0.2707,
+      "step": 5158
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.575581395348837e-08,
+      "loss": 0.3024,
+      "step": 5159
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.57142857142857e-08,
+      "loss": 0.3033,
+      "step": 5160
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.567275747508305e-08,
+      "loss": 0.2891,
+      "step": 5161
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.563122923588039e-08,
+      "loss": 0.2798,
+      "step": 5162
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.558970099667774e-08,
+      "loss": 0.2925,
+      "step": 5163
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.554817275747508e-08,
+      "loss": 0.2672,
+      "step": 5164
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.550664451827241e-08,
+      "loss": 0.2884,
+      "step": 5165
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.546511627906976e-08,
+      "loss": 0.2627,
+      "step": 5166
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.54235880398671e-08,
+      "loss": 0.3531,
+      "step": 5167
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.538205980066445e-08,
+      "loss": 0.2739,
+      "step": 5168
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.534053156146179e-08,
+      "loss": 0.2709,
+      "step": 5169
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.529900332225914e-08,
+      "loss": 0.3365,
+      "step": 5170
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.525747508305647e-08,
+      "loss": 0.2769,
+      "step": 5171
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.521594684385381e-08,
+      "loss": 0.3386,
+      "step": 5172
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.517441860465116e-08,
+      "loss": 0.2567,
+      "step": 5173
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.51328903654485e-08,
+      "loss": 0.2981,
+      "step": 5174
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.509136212624585e-08,
+      "loss": 0.2436,
+      "step": 5175
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.504983388704318e-08,
+      "loss": 0.3115,
+      "step": 5176
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.500830564784053e-08,
+      "loss": 0.2534,
+      "step": 5177
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.496677740863787e-08,
+      "loss": 0.2558,
+      "step": 5178
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.492524916943522e-08,
+      "loss": 0.3539,
+      "step": 5179
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.488372093023256e-08,
+      "loss": 0.3172,
+      "step": 5180
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.48421926910299e-08,
+      "loss": 0.2525,
+      "step": 5181
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.480066445182724e-08,
+      "loss": 0.289,
+      "step": 5182
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.475913621262458e-08,
+      "loss": 0.2987,
+      "step": 5183
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.471760797342193e-08,
+      "loss": 0.26,
+      "step": 5184
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.467607973421927e-08,
+      "loss": 0.2604,
+      "step": 5185
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.463455149501662e-08,
+      "loss": 0.2916,
+      "step": 5186
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.459302325581395e-08,
+      "loss": 0.2411,
+      "step": 5187
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.455149501661128e-08,
+      "loss": 0.2921,
+      "step": 5188
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.450996677740864e-08,
+      "loss": 0.3477,
+      "step": 5189
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.446843853820597e-08,
+      "loss": 0.2507,
+      "step": 5190
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.442691029900333e-08,
+      "loss": 0.3905,
+      "step": 5191
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.438538205980065e-08,
+      "loss": 0.2736,
+      "step": 5192
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.434385382059802e-08,
+      "loss": 0.2734,
+      "step": 5193
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.430232558139534e-08,
+      "loss": 0.312,
+      "step": 5194
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.426079734219268e-08,
+      "loss": 0.3389,
+      "step": 5195
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.421926910299003e-08,
+      "loss": 0.3387,
+      "step": 5196
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.417774086378736e-08,
+      "loss": 0.2502,
+      "step": 5197
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.413621262458471e-08,
+      "loss": 0.258,
+      "step": 5198
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.409468438538205e-08,
+      "loss": 0.4199,
+      "step": 5199
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.40531561461794e-08,
+      "loss": 0.2756,
+      "step": 5200
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.401162790697674e-08,
+      "loss": 0.2313,
+      "step": 5201
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.397009966777407e-08,
+      "loss": 0.2883,
+      "step": 5202
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.392857142857142e-08,
+      "loss": 0.3099,
+      "step": 5203
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.388704318936876e-08,
+      "loss": 0.3238,
+      "step": 5204
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.384551495016611e-08,
+      "loss": 0.3106,
+      "step": 5205
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.380398671096345e-08,
+      "loss": 0.272,
+      "step": 5206
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.37624584717608e-08,
+      "loss": 0.2358,
+      "step": 5207
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.372093023255813e-08,
+      "loss": 0.2866,
+      "step": 5208
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.367940199335548e-08,
+      "loss": 0.2969,
+      "step": 5209
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.363787375415282e-08,
+      "loss": 0.2302,
+      "step": 5210
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.359634551495016e-08,
+      "loss": 0.2233,
+      "step": 5211
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.355481727574751e-08,
+      "loss": 0.3185,
+      "step": 5212
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.351328903654484e-08,
+      "loss": 0.3099,
+      "step": 5213
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.34717607973422e-08,
+      "loss": 0.2721,
+      "step": 5214
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.343023255813953e-08,
+      "loss": 0.2997,
+      "step": 5215
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.338870431893688e-08,
+      "loss": 0.3397,
+      "step": 5216
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.334717607973422e-08,
+      "loss": 0.28,
+      "step": 5217
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.330564784053155e-08,
+      "loss": 0.3211,
+      "step": 5218
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.32641196013289e-08,
+      "loss": 0.2604,
+      "step": 5219
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.322259136212624e-08,
+      "loss": 0.3021,
+      "step": 5220
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.318106312292359e-08,
+      "loss": 0.258,
+      "step": 5221
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.313953488372093e-08,
+      "loss": 0.3101,
+      "step": 5222
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.309800664451828e-08,
+      "loss": 0.3268,
+      "step": 5223
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.305647840531561e-08,
+      "loss": 0.26,
+      "step": 5224
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.301495016611295e-08,
+      "loss": 0.2667,
+      "step": 5225
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.29734219269103e-08,
+      "loss": 0.3,
+      "step": 5226
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.293189368770764e-08,
+      "loss": 0.2582,
+      "step": 5227
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.289036544850499e-08,
+      "loss": 0.2895,
+      "step": 5228
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.284883720930232e-08,
+      "loss": 0.2409,
+      "step": 5229
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.280730897009967e-08,
+      "loss": 0.3135,
+      "step": 5230
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.276578073089701e-08,
+      "loss": 0.3506,
+      "step": 5231
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.272425249169434e-08,
+      "loss": 0.2797,
+      "step": 5232
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.26827242524917e-08,
+      "loss": 0.3147,
+      "step": 5233
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.264119601328902e-08,
+      "loss": 0.2977,
+      "step": 5234
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.259966777408639e-08,
+      "loss": 0.3371,
+      "step": 5235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.255813953488371e-08,
+      "loss": 0.2422,
+      "step": 5236
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.251661129568107e-08,
+      "loss": 0.2728,
+      "step": 5237
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.24750830564784e-08,
+      "loss": 0.251,
+      "step": 5238
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.243355481727573e-08,
+      "loss": 0.2839,
+      "step": 5239
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.239202657807308e-08,
+      "loss": 0.3395,
+      "step": 5240
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.235049833887042e-08,
+      "loss": 0.2946,
+      "step": 5241
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.230897009966777e-08,
+      "loss": 0.2985,
+      "step": 5242
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.22674418604651e-08,
+      "loss": 0.336,
+      "step": 5243
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.222591362126246e-08,
+      "loss": 0.3181,
+      "step": 5244
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.218438538205979e-08,
+      "loss": 0.2914,
+      "step": 5245
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.214285714285714e-08,
+      "loss": 0.2879,
+      "step": 5246
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.210132890365448e-08,
+      "loss": 0.3447,
+      "step": 5247
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.205980066445182e-08,
+      "loss": 0.3032,
+      "step": 5248
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.201827242524917e-08,
+      "loss": 0.2577,
+      "step": 5249
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.19767441860465e-08,
+      "loss": 0.2648,
+      "step": 5250
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.193521594684385e-08,
+      "loss": 0.3029,
+      "step": 5251
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.189368770764119e-08,
+      "loss": 0.3141,
+      "step": 5252
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.185215946843854e-08,
+      "loss": 0.2997,
+      "step": 5253
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.181063122923588e-08,
+      "loss": 0.2348,
+      "step": 5254
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.176910299003321e-08,
+      "loss": 0.249,
+      "step": 5255
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.172757475083056e-08,
+      "loss": 0.2852,
+      "step": 5256
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.16860465116279e-08,
+      "loss": 0.2869,
+      "step": 5257
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.164451827242525e-08,
+      "loss": 0.2813,
+      "step": 5258
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.160299003322259e-08,
+      "loss": 0.2708,
+      "step": 5259
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.156146179401994e-08,
+      "loss": 0.2718,
+      "step": 5260
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.151993355481727e-08,
+      "loss": 0.2835,
+      "step": 5261
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.147840531561461e-08,
+      "loss": 0.321,
+      "step": 5262
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.143687707641196e-08,
+      "loss": 0.3594,
+      "step": 5263
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.13953488372093e-08,
+      "loss": 0.2567,
+      "step": 5264
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.135382059800665e-08,
+      "loss": 0.327,
+      "step": 5265
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.131229235880398e-08,
+      "loss": 0.2292,
+      "step": 5266
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.127076411960133e-08,
+      "loss": 0.2571,
+      "step": 5267
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.122923588039867e-08,
+      "loss": 0.2889,
+      "step": 5268
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.118770764119601e-08,
+      "loss": 0.2975,
+      "step": 5269
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.114617940199336e-08,
+      "loss": 0.2918,
+      "step": 5270
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.110465116279068e-08,
+      "loss": 0.3368,
+      "step": 5271
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.106312292358804e-08,
+      "loss": 0.2586,
+      "step": 5272
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.102159468438537e-08,
+      "loss": 0.2774,
+      "step": 5273
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.098006644518273e-08,
+      "loss": 0.3333,
+      "step": 5274
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.093853820598006e-08,
+      "loss": 0.2389,
+      "step": 5275
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.089700996677742e-08,
+      "loss": 0.3229,
+      "step": 5276
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.085548172757474e-08,
+      "loss": 0.2621,
+      "step": 5277
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.081395348837208e-08,
+      "loss": 0.2937,
+      "step": 5278
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.077242524916943e-08,
+      "loss": 0.2747,
+      "step": 5279
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.073089700996677e-08,
+      "loss": 0.2509,
+      "step": 5280
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.068936877076412e-08,
+      "loss": 0.2791,
+      "step": 5281
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.064784053156145e-08,
+      "loss": 0.3269,
+      "step": 5282
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.06063122923588e-08,
+      "loss": 0.3132,
+      "step": 5283
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.056478405315614e-08,
+      "loss": 0.3019,
+      "step": 5284
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.052325581395348e-08,
+      "loss": 0.3202,
+      "step": 5285
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.048172757475083e-08,
+      "loss": 0.3022,
+      "step": 5286
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.044019933554816e-08,
+      "loss": 0.2479,
+      "step": 5287
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.039867109634551e-08,
+      "loss": 0.277,
+      "step": 5288
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.035714285714285e-08,
+      "loss": 0.322,
+      "step": 5289
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.03156146179402e-08,
+      "loss": 0.2797,
+      "step": 5290
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.027408637873754e-08,
+      "loss": 0.2951,
+      "step": 5291
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.023255813953487e-08,
+      "loss": 0.2369,
+      "step": 5292
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.019102990033222e-08,
+      "loss": 0.283,
+      "step": 5293
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.014950166112956e-08,
+      "loss": 0.2792,
+      "step": 5294
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.010797342192691e-08,
+      "loss": 0.28,
+      "step": 5295
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.006644518272425e-08,
+      "loss": 0.2856,
+      "step": 5296
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.00249169435216e-08,
+      "loss": 0.2582,
+      "step": 5297
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.998338870431893e-08,
+      "loss": 0.3136,
+      "step": 5298
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.994186046511627e-08,
+      "loss": 0.2771,
+      "step": 5299
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.990033222591362e-08,
+      "loss": 0.2603,
+      "step": 5300
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.985880398671096e-08,
+      "loss": 0.33,
+      "step": 5301
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.981727574750831e-08,
+      "loss": 0.2744,
+      "step": 5302
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.977574750830564e-08,
+      "loss": 0.3229,
+      "step": 5303
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.9734219269103e-08,
+      "loss": 0.2393,
+      "step": 5304
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.969269102990033e-08,
+      "loss": 0.2714,
+      "step": 5305
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.965116279069768e-08,
+      "loss": 0.2981,
+      "step": 5306
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.960963455149502e-08,
+      "loss": 0.3106,
+      "step": 5307
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.956810631229235e-08,
+      "loss": 0.2868,
+      "step": 5308
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.95265780730897e-08,
+      "loss": 0.3291,
+      "step": 5309
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.948504983388704e-08,
+      "loss": 0.2892,
+      "step": 5310
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.944352159468439e-08,
+      "loss": 0.3114,
+      "step": 5311
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.940199335548173e-08,
+      "loss": 0.2711,
+      "step": 5312
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.936046511627908e-08,
+      "loss": 0.2733,
+      "step": 5313
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.931893687707641e-08,
+      "loss": 0.2664,
+      "step": 5314
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.927740863787374e-08,
+      "loss": 0.3101,
+      "step": 5315
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.92358803986711e-08,
+      "loss": 0.2903,
+      "step": 5316
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.919435215946843e-08,
+      "loss": 0.2883,
+      "step": 5317
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.915282392026579e-08,
+      "loss": 0.3135,
+      "step": 5318
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.911129568106311e-08,
+      "loss": 0.2853,
+      "step": 5319
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.906976744186046e-08,
+      "loss": 0.2334,
+      "step": 5320
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.90282392026578e-08,
+      "loss": 0.2591,
+      "step": 5321
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.898671096345514e-08,
+      "loss": 0.2333,
+      "step": 5322
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.894518272425249e-08,
+      "loss": 0.2638,
+      "step": 5323
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.890365448504982e-08,
+      "loss": 0.3038,
+      "step": 5324
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.886212624584717e-08,
+      "loss": 0.3156,
+      "step": 5325
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.882059800664451e-08,
+      "loss": 0.3078,
+      "step": 5326
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.877906976744186e-08,
+      "loss": 0.2488,
+      "step": 5327
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.87375415282392e-08,
+      "loss": 0.2644,
+      "step": 5328
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.869601328903653e-08,
+      "loss": 0.3309,
+      "step": 5329
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.865448504983388e-08,
+      "loss": 0.3105,
+      "step": 5330
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.861295681063122e-08,
+      "loss": 0.3024,
+      "step": 5331
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.857142857142857e-08,
+      "loss": 0.2862,
+      "step": 5332
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.85299003322259e-08,
+      "loss": 0.3277,
+      "step": 5333
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.848837209302326e-08,
+      "loss": 0.2655,
+      "step": 5334
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.844684385382059e-08,
+      "loss": 0.2615,
+      "step": 5335
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.840531561461793e-08,
+      "loss": 0.263,
+      "step": 5336
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.836378737541528e-08,
+      "loss": 0.2776,
+      "step": 5337
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.832225913621262e-08,
+      "loss": 0.2397,
+      "step": 5338
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.828073089700997e-08,
+      "loss": 0.2602,
+      "step": 5339
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.82392026578073e-08,
+      "loss": 0.2629,
+      "step": 5340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.819767441860465e-08,
+      "loss": 0.2816,
+      "step": 5341
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.815614617940199e-08,
+      "loss": 0.3008,
+      "step": 5342
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.811461794019934e-08,
+      "loss": 0.2741,
+      "step": 5343
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.807308970099668e-08,
+      "loss": 0.3131,
+      "step": 5344
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.803156146179401e-08,
+      "loss": 0.2197,
+      "step": 5345
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.799003322259136e-08,
+      "loss": 0.296,
+      "step": 5346
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.79485049833887e-08,
+      "loss": 0.3046,
+      "step": 5347
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.790697674418605e-08,
+      "loss": 0.2827,
+      "step": 5348
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.786544850498339e-08,
+      "loss": 0.3112,
+      "step": 5349
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.782392026578074e-08,
+      "loss": 0.2457,
+      "step": 5350
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.778239202657807e-08,
+      "loss": 0.3265,
+      "step": 5351
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.77408637873754e-08,
+      "loss": 0.2502,
+      "step": 5352
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.769933554817276e-08,
+      "loss": 0.3229,
+      "step": 5353
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.765780730897008e-08,
+      "loss": 0.3039,
+      "step": 5354
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.761627906976745e-08,
+      "loss": 0.3298,
+      "step": 5355
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.757475083056477e-08,
+      "loss": 0.318,
+      "step": 5356
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.753322259136214e-08,
+      "loss": 0.2849,
+      "step": 5357
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.749169435215946e-08,
+      "loss": 0.2729,
+      "step": 5358
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.74501661129568e-08,
+      "loss": 0.3361,
+      "step": 5359
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.740863787375415e-08,
+      "loss": 0.2901,
+      "step": 5360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.736710963455148e-08,
+      "loss": 0.3341,
+      "step": 5361
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.732558139534883e-08,
+      "loss": 0.2841,
+      "step": 5362
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.728405315614617e-08,
+      "loss": 0.2701,
+      "step": 5363
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.724252491694352e-08,
+      "loss": 0.2514,
+      "step": 5364
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.720099667774086e-08,
+      "loss": 0.2816,
+      "step": 5365
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.715946843853819e-08,
+      "loss": 0.2773,
+      "step": 5366
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.711794019933554e-08,
+      "loss": 0.2359,
+      "step": 5367
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.707641196013288e-08,
+      "loss": 0.3037,
+      "step": 5368
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.703488372093023e-08,
+      "loss": 0.3311,
+      "step": 5369
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.699335548172757e-08,
+      "loss": 0.2162,
+      "step": 5370
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.695182724252492e-08,
+      "loss": 0.277,
+      "step": 5371
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.691029900332225e-08,
+      "loss": 0.2623,
+      "step": 5372
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.68687707641196e-08,
+      "loss": 0.3224,
+      "step": 5373
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.682724252491694e-08,
+      "loss": 0.2965,
+      "step": 5374
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.678571428571428e-08,
+      "loss": 0.3589,
+      "step": 5375
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.674418604651163e-08,
+      "loss": 0.354,
+      "step": 5376
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.670265780730896e-08,
+      "loss": 0.2892,
+      "step": 5377
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.666112956810631e-08,
+      "loss": 0.332,
+      "step": 5378
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.661960132890365e-08,
+      "loss": 0.3241,
+      "step": 5379
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.6578073089701e-08,
+      "loss": 0.3428,
+      "step": 5380
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.653654485049834e-08,
+      "loss": 0.2669,
+      "step": 5381
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.649501661129567e-08,
+      "loss": 0.3253,
+      "step": 5382
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.645348837209302e-08,
+      "loss": 0.3009,
+      "step": 5383
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.641196013289036e-08,
+      "loss": 0.3303,
+      "step": 5384
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.637043189368771e-08,
+      "loss": 0.289,
+      "step": 5385
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.632890365448505e-08,
+      "loss": 0.2964,
+      "step": 5386
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.62873754152824e-08,
+      "loss": 0.279,
+      "step": 5387
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.624584717607973e-08,
+      "loss": 0.2917,
+      "step": 5388
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.620431893687707e-08,
+      "loss": 0.2713,
+      "step": 5389
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.616279069767442e-08,
+      "loss": 0.2611,
+      "step": 5390
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.612126245847176e-08,
+      "loss": 0.241,
+      "step": 5391
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.607973421926911e-08,
+      "loss": 0.2583,
+      "step": 5392
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.603820598006644e-08,
+      "loss": 0.2936,
+      "step": 5393
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.59966777408638e-08,
+      "loss": 0.3221,
+      "step": 5394
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.595514950166113e-08,
+      "loss": 0.2774,
+      "step": 5395
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.591362126245845e-08,
+      "loss": 0.3192,
+      "step": 5396
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.587209302325582e-08,
+      "loss": 0.2159,
+      "step": 5397
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.583056478405314e-08,
+      "loss": 0.2967,
+      "step": 5398
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.57890365448505e-08,
+      "loss": 0.3246,
+      "step": 5399
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.574750830564783e-08,
+      "loss": 0.2715,
+      "step": 5400
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.570598006644519e-08,
+      "loss": 0.262,
+      "step": 5401
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.566445182724252e-08,
+      "loss": 0.2762,
+      "step": 5402
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.562292358803987e-08,
+      "loss": 0.2746,
+      "step": 5403
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.55813953488372e-08,
+      "loss": 0.349,
+      "step": 5404
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.553986710963454e-08,
+      "loss": 0.3233,
+      "step": 5405
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.549833887043189e-08,
+      "loss": 0.3282,
+      "step": 5406
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.545681063122923e-08,
+      "loss": 0.3098,
+      "step": 5407
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.541528239202658e-08,
+      "loss": 0.3006,
+      "step": 5408
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.537375415282391e-08,
+      "loss": 0.2804,
+      "step": 5409
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.533222591362126e-08,
+      "loss": 0.3361,
+      "step": 5410
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.52906976744186e-08,
+      "loss": 0.2685,
+      "step": 5411
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.524916943521594e-08,
+      "loss": 0.2498,
+      "step": 5412
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.520764119601329e-08,
+      "loss": 0.3067,
+      "step": 5413
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.516611295681062e-08,
+      "loss": 0.2868,
+      "step": 5414
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.512458471760797e-08,
+      "loss": 0.2588,
+      "step": 5415
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.508305647840531e-08,
+      "loss": 0.2973,
+      "step": 5416
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.504152823920266e-08,
+      "loss": 0.3304,
+      "step": 5417
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.5e-08,
+      "loss": 0.2262,
+      "step": 5418
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.495847176079733e-08,
+      "loss": 0.3248,
+      "step": 5419
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.491694352159468e-08,
+      "loss": 0.322,
+      "step": 5420
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.487541528239202e-08,
+      "loss": 0.2925,
+      "step": 5421
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.483388704318936e-08,
+      "loss": 0.2459,
+      "step": 5422
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.479235880398671e-08,
+      "loss": 0.3168,
+      "step": 5423
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.475083056478404e-08,
+      "loss": 0.2263,
+      "step": 5424
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.47093023255814e-08,
+      "loss": 0.2785,
+      "step": 5425
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.466777408637873e-08,
+      "loss": 0.2355,
+      "step": 5426
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.462624584717608e-08,
+      "loss": 0.2712,
+      "step": 5427
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.458471760797342e-08,
+      "loss": 0.3143,
+      "step": 5428
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.454318936877077e-08,
+      "loss": 0.3276,
+      "step": 5429
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.45016611295681e-08,
+      "loss": 0.2738,
+      "step": 5430
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.446013289036544e-08,
+      "loss": 0.2794,
+      "step": 5431
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.441860465116279e-08,
+      "loss": 0.2608,
+      "step": 5432
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.437707641196013e-08,
+      "loss": 0.2869,
+      "step": 5433
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.433554817275748e-08,
+      "loss": 0.2442,
+      "step": 5434
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.429401993355481e-08,
+      "loss": 0.3022,
+      "step": 5435
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.425249169435216e-08,
+      "loss": 0.2719,
+      "step": 5436
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.421096345514949e-08,
+      "loss": 0.2749,
+      "step": 5437
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.416943521594684e-08,
+      "loss": 0.2665,
+      "step": 5438
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.412790697674417e-08,
+      "loss": 0.2803,
+      "step": 5439
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.408637873754153e-08,
+      "loss": 0.2999,
+      "step": 5440
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.404485049833886e-08,
+      "loss": 0.3657,
+      "step": 5441
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.400332225913621e-08,
+      "loss": 0.3202,
+      "step": 5442
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.396179401993355e-08,
+      "loss": 0.2864,
+      "step": 5443
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.39202657807309e-08,
+      "loss": 0.2899,
+      "step": 5444
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.387873754152824e-08,
+      "loss": 0.2676,
+      "step": 5445
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.383720930232557e-08,
+      "loss": 0.2827,
+      "step": 5446
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.379568106312292e-08,
+      "loss": 0.334,
+      "step": 5447
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.375415282392026e-08,
+      "loss": 0.3309,
+      "step": 5448
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.371262458471761e-08,
+      "loss": 0.2921,
+      "step": 5449
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.367109634551495e-08,
+      "loss": 0.2154,
+      "step": 5450
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.36295681063123e-08,
+      "loss": 0.2878,
+      "step": 5451
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.358803986710963e-08,
+      "loss": 0.3139,
+      "step": 5452
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.354651162790697e-08,
+      "loss": 0.2972,
+      "step": 5453
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.350498338870432e-08,
+      "loss": 0.3157,
+      "step": 5454
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.346345514950166e-08,
+      "loss": 0.255,
+      "step": 5455
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.3421926910299e-08,
+      "loss": 0.3697,
+      "step": 5456
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.338039867109634e-08,
+      "loss": 0.3646,
+      "step": 5457
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.333887043189369e-08,
+      "loss": 0.2762,
+      "step": 5458
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.329734219269103e-08,
+      "loss": 0.2849,
+      "step": 5459
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.325581395348837e-08,
+      "loss": 0.2526,
+      "step": 5460
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.32142857142857e-08,
+      "loss": 0.2785,
+      "step": 5461
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.317275747508305e-08,
+      "loss": 0.2653,
+      "step": 5462
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.313122923588039e-08,
+      "loss": 0.2877,
+      "step": 5463
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.308970099667774e-08,
+      "loss": 0.2549,
+      "step": 5464
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.304817275747508e-08,
+      "loss": 0.2898,
+      "step": 5465
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.300664451827243e-08,
+      "loss": 0.2898,
+      "step": 5466
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.296511627906976e-08,
+      "loss": 0.27,
+      "step": 5467
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.29235880398671e-08,
+      "loss": 0.2487,
+      "step": 5468
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.288205980066445e-08,
+      "loss": 0.268,
+      "step": 5469
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.284053156146179e-08,
+      "loss": 0.3036,
+      "step": 5470
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.279900332225914e-08,
+      "loss": 0.2924,
+      "step": 5471
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.275747508305647e-08,
+      "loss": 0.3276,
+      "step": 5472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.271594684385382e-08,
+      "loss": 0.2671,
+      "step": 5473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.267441860465116e-08,
+      "loss": 0.3075,
+      "step": 5474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.26328903654485e-08,
+      "loss": 0.2677,
+      "step": 5475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.259136212624585e-08,
+      "loss": 0.3067,
+      "step": 5476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.254983388704318e-08,
+      "loss": 0.3433,
+      "step": 5477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.250830564784053e-08,
+      "loss": 0.2951,
+      "step": 5478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.246677740863787e-08,
+      "loss": 0.2931,
+      "step": 5479
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.242524916943522e-08,
+      "loss": 0.285,
+      "step": 5480
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.238372093023256e-08,
+      "loss": 0.3123,
+      "step": 5481
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.23421926910299e-08,
+      "loss": 0.2402,
+      "step": 5482
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.230066445182723e-08,
+      "loss": 0.254,
+      "step": 5483
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.225913621262458e-08,
+      "loss": 0.3604,
+      "step": 5484
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.221760797342192e-08,
+      "loss": 0.2785,
+      "step": 5485
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.217607973421927e-08,
+      "loss": 0.268,
+      "step": 5486
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.21345514950166e-08,
+      "loss": 0.2735,
+      "step": 5487
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.209302325581396e-08,
+      "loss": 0.3031,
+      "step": 5488
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.205149501661129e-08,
+      "loss": 0.3095,
+      "step": 5489
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.200996677740863e-08,
+      "loss": 0.3477,
+      "step": 5490
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.196843853820598e-08,
+      "loss": 0.3159,
+      "step": 5491
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.192691029900332e-08,
+      "loss": 0.2533,
+      "step": 5492
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.188538205980067e-08,
+      "loss": 0.3359,
+      "step": 5493
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.1843853820598e-08,
+      "loss": 0.3498,
+      "step": 5494
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.180232558139535e-08,
+      "loss": 0.231,
+      "step": 5495
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.176079734219269e-08,
+      "loss": 0.2891,
+      "step": 5496
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.171926910299003e-08,
+      "loss": 0.312,
+      "step": 5497
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.167774086378738e-08,
+      "loss": 0.3276,
+      "step": 5498
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.163621262458471e-08,
+      "loss": 0.3085,
+      "step": 5499
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.159468438538206e-08,
+      "loss": 0.3573,
+      "step": 5500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.15531561461794e-08,
+      "loss": 0.2833,
+      "step": 5501
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.151162790697674e-08,
+      "loss": 0.2261,
+      "step": 5502
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.147009966777409e-08,
+      "loss": 0.2906,
+      "step": 5503
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.142857142857142e-08,
+      "loss": 0.2741,
+      "step": 5504
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.138704318936876e-08,
+      "loss": 0.2766,
+      "step": 5505
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.134551495016611e-08,
+      "loss": 0.2683,
+      "step": 5506
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.130398671096345e-08,
+      "loss": 0.2521,
+      "step": 5507
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.12624584717608e-08,
+      "loss": 0.2673,
+      "step": 5508
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.122093023255813e-08,
+      "loss": 0.3271,
+      "step": 5509
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.117940199335548e-08,
+      "loss": 0.2921,
+      "step": 5510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.113787375415282e-08,
+      "loss": 0.3025,
+      "step": 5511
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.109634551495016e-08,
+      "loss": 0.3043,
+      "step": 5512
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.105481727574751e-08,
+      "loss": 0.3146,
+      "step": 5513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.101328903654484e-08,
+      "loss": 0.2755,
+      "step": 5514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.09717607973422e-08,
+      "loss": 0.2931,
+      "step": 5515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.093023255813953e-08,
+      "loss": 0.3277,
+      "step": 5516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.088870431893688e-08,
+      "loss": 0.2782,
+      "step": 5517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.084717607973422e-08,
+      "loss": 0.2488,
+      "step": 5518
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.080564784053155e-08,
+      "loss": 0.2662,
+      "step": 5519
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.076411960132889e-08,
+      "loss": 0.3182,
+      "step": 5520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.072259136212624e-08,
+      "loss": 0.2377,
+      "step": 5521
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.068106312292358e-08,
+      "loss": 0.3151,
+      "step": 5522
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.063953488372093e-08,
+      "loss": 0.3367,
+      "step": 5523
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.059800664451827e-08,
+      "loss": 0.3092,
+      "step": 5524
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.055647840531562e-08,
+      "loss": 0.2882,
+      "step": 5525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.051495016611295e-08,
+      "loss": 0.2834,
+      "step": 5526
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.047342192691029e-08,
+      "loss": 0.2619,
+      "step": 5527
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.043189368770764e-08,
+      "loss": 0.2783,
+      "step": 5528
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.039036544850498e-08,
+      "loss": 0.3083,
+      "step": 5529
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.034883720930233e-08,
+      "loss": 0.3486,
+      "step": 5530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.030730897009966e-08,
+      "loss": 0.3395,
+      "step": 5531
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.026578073089701e-08,
+      "loss": 0.2737,
+      "step": 5532
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.022425249169435e-08,
+      "loss": 0.2988,
+      "step": 5533
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.018272425249169e-08,
+      "loss": 0.3219,
+      "step": 5534
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.014119601328904e-08,
+      "loss": 0.3095,
+      "step": 5535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.009966777408637e-08,
+      "loss": 0.2572,
+      "step": 5536
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.005813953488372e-08,
+      "loss": 0.343,
+      "step": 5537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.001661129568106e-08,
+      "loss": 0.2861,
+      "step": 5538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.997508305647841e-08,
+      "loss": 0.3104,
+      "step": 5539
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.993355481727575e-08,
+      "loss": 0.2191,
+      "step": 5540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.98920265780731e-08,
+      "loss": 0.3032,
+      "step": 5541
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.985049833887042e-08,
+      "loss": 0.2954,
+      "step": 5542
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.980897009966777e-08,
+      "loss": 0.2468,
+      "step": 5543
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 6.976744186046511e-08,
+      "loss": 0.2745,
+      "step": 5544
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.972591362126246e-08,
+      "loss": 0.2823,
+      "step": 5545
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.96843853820598e-08,
+      "loss": 0.2858,
+      "step": 5546
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.964285714285714e-08,
+      "loss": 0.2828,
+      "step": 5547
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.960132890365448e-08,
+      "loss": 0.265,
+      "step": 5548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.955980066445182e-08,
+      "loss": 0.3568,
+      "step": 5549
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.951827242524917e-08,
+      "loss": 0.2755,
+      "step": 5550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.94767441860465e-08,
+      "loss": 0.2939,
+      "step": 5551
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.943521594684385e-08,
+      "loss": 0.2618,
+      "step": 5552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.939368770764119e-08,
+      "loss": 0.3263,
+      "step": 5553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.935215946843854e-08,
+      "loss": 0.2744,
+      "step": 5554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.931063122923588e-08,
+      "loss": 0.3197,
+      "step": 5555
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.926910299003323e-08,
+      "loss": 0.3161,
+      "step": 5556
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.922757475083056e-08,
+      "loss": 0.311,
+      "step": 5557
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.91860465116279e-08,
+      "loss": 0.311,
+      "step": 5558
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.914451827242525e-08,
+      "loss": 0.2475,
+      "step": 5559
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.910299003322259e-08,
+      "loss": 0.256,
+      "step": 5560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.906146179401994e-08,
+      "loss": 0.2537,
+      "step": 5561
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.901993355481727e-08,
+      "loss": 0.3197,
+      "step": 5562
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.897840531561462e-08,
+      "loss": 0.2736,
+      "step": 5563
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.893687707641195e-08,
+      "loss": 0.2691,
+      "step": 5564
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.88953488372093e-08,
+      "loss": 0.242,
+      "step": 5565
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.885382059800664e-08,
+      "loss": 0.2221,
+      "step": 5566
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.881229235880399e-08,
+      "loss": 0.2607,
+      "step": 5567
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.877076411960132e-08,
+      "loss": 0.3058,
+      "step": 5568
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.872923588039867e-08,
+      "loss": 0.272,
+      "step": 5569
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.868770764119601e-08,
+      "loss": 0.251,
+      "step": 5570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.864617940199335e-08,
+      "loss": 0.2944,
+      "step": 5571
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.86046511627907e-08,
+      "loss": 0.3088,
+      "step": 5572
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.856312292358803e-08,
+      "loss": 0.332,
+      "step": 5573
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.852159468438538e-08,
+      "loss": 0.2634,
+      "step": 5574
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.848006644518272e-08,
+      "loss": 0.3021,
+      "step": 5575
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.843853820598007e-08,
+      "loss": 0.3493,
+      "step": 5576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.83970099667774e-08,
+      "loss": 0.3093,
+      "step": 5577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.835548172757476e-08,
+      "loss": 0.2718,
+      "step": 5578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.831395348837209e-08,
+      "loss": 0.2852,
+      "step": 5579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 6.827242524916943e-08,
+      "loss": 0.3477,
+      "step": 5580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.823089700996678e-08,
+      "loss": 0.3005,
+      "step": 5581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.818936877076412e-08,
+      "loss": 0.295,
+      "step": 5582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.814784053156145e-08,
+      "loss": 0.3173,
+      "step": 5583
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.81063122923588e-08,
+      "loss": 0.3235,
+      "step": 5584
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.806478405315614e-08,
+      "loss": 0.2272,
+      "step": 5585
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.802325581395348e-08,
+      "loss": 0.3405,
+      "step": 5586
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.798172757475083e-08,
+      "loss": 0.3456,
+      "step": 5587
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.794019933554816e-08,
+      "loss": 0.2625,
+      "step": 5588
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.789867109634551e-08,
+      "loss": 0.2844,
+      "step": 5589
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.785714285714285e-08,
+      "loss": 0.2813,
+      "step": 5590
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.78156146179402e-08,
+      "loss": 0.3075,
+      "step": 5591
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.777408637873754e-08,
+      "loss": 0.3277,
+      "step": 5592
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.773255813953489e-08,
+      "loss": 0.2793,
+      "step": 5593
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.769102990033222e-08,
+      "loss": 0.3186,
+      "step": 5594
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.764950166112956e-08,
+      "loss": 0.3325,
+      "step": 5595
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.760797342192691e-08,
+      "loss": 0.2607,
+      "step": 5596
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.756644518272425e-08,
+      "loss": 0.3102,
+      "step": 5597
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.75249169435216e-08,
+      "loss": 0.2441,
+      "step": 5598
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.748338870431893e-08,
+      "loss": 0.2986,
+      "step": 5599
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.744186046511628e-08,
+      "loss": 0.2634,
+      "step": 5600
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.740033222591361e-08,
+      "loss": 0.2383,
+      "step": 5601
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.735880398671096e-08,
+      "loss": 0.2731,
+      "step": 5602
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.73172757475083e-08,
+      "loss": 0.3052,
+      "step": 5603
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.727574750830564e-08,
+      "loss": 0.2675,
+      "step": 5604
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.723421926910298e-08,
+      "loss": 0.3309,
+      "step": 5605
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.719269102990033e-08,
+      "loss": 0.2613,
+      "step": 5606
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.715116279069767e-08,
+      "loss": 0.2664,
+      "step": 5607
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.710963455149502e-08,
+      "loss": 0.2808,
+      "step": 5608
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.706810631229236e-08,
+      "loss": 0.3091,
+      "step": 5609
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.702657807308969e-08,
+      "loss": 0.258,
+      "step": 5610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.698504983388704e-08,
+      "loss": 0.3572,
+      "step": 5611
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.694352159468438e-08,
+      "loss": 0.3083,
+      "step": 5612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.690199335548173e-08,
+      "loss": 0.2412,
+      "step": 5613
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.686046511627907e-08,
+      "loss": 0.2823,
+      "step": 5614
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.681893687707642e-08,
+      "loss": 0.2703,
+      "step": 5615
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 6.677740863787375e-08,
+      "loss": 0.3231,
+      "step": 5616
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.673588039867109e-08,
+      "loss": 0.2928,
+      "step": 5617
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.669435215946844e-08,
+      "loss": 0.2576,
+      "step": 5618
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.665282392026578e-08,
+      "loss": 0.2276,
+      "step": 5619
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.661129568106313e-08,
+      "loss": 0.2908,
+      "step": 5620
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.656976744186046e-08,
+      "loss": 0.292,
+      "step": 5621
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.652823920265781e-08,
+      "loss": 0.2501,
+      "step": 5622
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.648671096345515e-08,
+      "loss": 0.3699,
+      "step": 5623
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.644518272425249e-08,
+      "loss": 0.2908,
+      "step": 5624
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.640365448504982e-08,
+      "loss": 0.296,
+      "step": 5625
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.636212624584717e-08,
+      "loss": 0.2365,
+      "step": 5626
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.632059800664451e-08,
+      "loss": 0.2533,
+      "step": 5627
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.627906976744186e-08,
+      "loss": 0.2943,
+      "step": 5628
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.62375415282392e-08,
+      "loss": 0.314,
+      "step": 5629
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.619601328903655e-08,
+      "loss": 0.2652,
+      "step": 5630
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.615448504983388e-08,
+      "loss": 0.2692,
+      "step": 5631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.611295681063122e-08,
+      "loss": 0.2492,
+      "step": 5632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.607142857142857e-08,
+      "loss": 0.2695,
+      "step": 5633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.602990033222591e-08,
+      "loss": 0.2813,
+      "step": 5634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.598837209302326e-08,
+      "loss": 0.3684,
+      "step": 5635
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.59468438538206e-08,
+      "loss": 0.2633,
+      "step": 5636
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.590531561461794e-08,
+      "loss": 0.2794,
+      "step": 5637
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.586378737541528e-08,
+      "loss": 0.2788,
+      "step": 5638
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.582225913621262e-08,
+      "loss": 0.288,
+      "step": 5639
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.578073089700997e-08,
+      "loss": 0.3077,
+      "step": 5640
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.57392026578073e-08,
+      "loss": 0.3383,
+      "step": 5641
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.569767441860465e-08,
+      "loss": 0.2508,
+      "step": 5642
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.565614617940199e-08,
+      "loss": 0.3233,
+      "step": 5643
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.561461794019934e-08,
+      "loss": 0.3164,
+      "step": 5644
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.557308970099668e-08,
+      "loss": 0.3026,
+      "step": 5645
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.553156146179401e-08,
+      "loss": 0.3391,
+      "step": 5646
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.549003322259135e-08,
+      "loss": 0.2843,
+      "step": 5647
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.54485049833887e-08,
+      "loss": 0.3067,
+      "step": 5648
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.540697674418604e-08,
+      "loss": 0.3266,
+      "step": 5649
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.536544850498339e-08,
+      "loss": 0.3033,
+      "step": 5650
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.532392026578073e-08,
+      "loss": 0.3188,
+      "step": 5651
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 6.528239202657808e-08,
+      "loss": 0.2651,
+      "step": 5652
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.524086378737541e-08,
+      "loss": 0.3203,
+      "step": 5653
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.519933554817275e-08,
+      "loss": 0.3097,
+      "step": 5654
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.51578073089701e-08,
+      "loss": 0.2553,
+      "step": 5655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.511627906976744e-08,
+      "loss": 0.2878,
+      "step": 5656
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.507475083056479e-08,
+      "loss": 0.3139,
+      "step": 5657
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.503322259136212e-08,
+      "loss": 0.3083,
+      "step": 5658
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.499169435215947e-08,
+      "loss": 0.3178,
+      "step": 5659
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.495016611295681e-08,
+      "loss": 0.3215,
+      "step": 5660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.490863787375415e-08,
+      "loss": 0.3263,
+      "step": 5661
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.48671096345515e-08,
+      "loss": 0.2437,
+      "step": 5662
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.482558139534883e-08,
+      "loss": 0.2927,
+      "step": 5663
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.478405315614618e-08,
+      "loss": 0.2244,
+      "step": 5664
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.474252491694352e-08,
+      "loss": 0.3121,
+      "step": 5665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.470099667774086e-08,
+      "loss": 0.292,
+      "step": 5666
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.46594684385382e-08,
+      "loss": 0.2691,
+      "step": 5667
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.461794019933554e-08,
+      "loss": 0.3096,
+      "step": 5668
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.457641196013288e-08,
+      "loss": 0.2746,
+      "step": 5669
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.453488372093023e-08,
+      "loss": 0.2869,
+      "step": 5670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.449335548172757e-08,
+      "loss": 0.2473,
+      "step": 5671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.445182724252492e-08,
+      "loss": 0.319,
+      "step": 5672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.441029900332225e-08,
+      "loss": 0.2434,
+      "step": 5673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.43687707641196e-08,
+      "loss": 0.2535,
+      "step": 5674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.432724252491694e-08,
+      "loss": 0.366,
+      "step": 5675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.428571428571428e-08,
+      "loss": 0.2427,
+      "step": 5676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.424418604651163e-08,
+      "loss": 0.2389,
+      "step": 5677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.420265780730896e-08,
+      "loss": 0.2434,
+      "step": 5678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.416112956810631e-08,
+      "loss": 0.3419,
+      "step": 5679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.411960132890365e-08,
+      "loss": 0.2906,
+      "step": 5680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.4078073089701e-08,
+      "loss": 0.2804,
+      "step": 5681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.403654485049834e-08,
+      "loss": 0.2575,
+      "step": 5682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.399501661129567e-08,
+      "loss": 0.2732,
+      "step": 5683
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.395348837209301e-08,
+      "loss": 0.2502,
+      "step": 5684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.391196013289036e-08,
+      "loss": 0.2741,
+      "step": 5685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.38704318936877e-08,
+      "loss": 0.2826,
+      "step": 5686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.382890365448505e-08,
+      "loss": 0.2698,
+      "step": 5687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 6.378737541528238e-08,
+      "loss": 0.3368,
+      "step": 5688
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.374584717607973e-08,
+      "loss": 0.2975,
+      "step": 5689
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.370431893687707e-08,
+      "loss": 0.307,
+      "step": 5690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.366279069767441e-08,
+      "loss": 0.2509,
+      "step": 5691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.362126245847176e-08,
+      "loss": 0.2792,
+      "step": 5692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.35797342192691e-08,
+      "loss": 0.3202,
+      "step": 5693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.353820598006645e-08,
+      "loss": 0.3062,
+      "step": 5694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.349667774086378e-08,
+      "loss": 0.2679,
+      "step": 5695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.345514950166113e-08,
+      "loss": 0.2991,
+      "step": 5696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.341362126245847e-08,
+      "loss": 0.262,
+      "step": 5697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.33720930232558e-08,
+      "loss": 0.3216,
+      "step": 5698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.333056478405316e-08,
+      "loss": 0.2589,
+      "step": 5699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.328903654485049e-08,
+      "loss": 0.3012,
+      "step": 5700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.324750830564784e-08,
+      "loss": 0.3689,
+      "step": 5701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.320598006644518e-08,
+      "loss": 0.2671,
+      "step": 5702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.316445182724253e-08,
+      "loss": 0.2949,
+      "step": 5703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.312292358803987e-08,
+      "loss": 0.2545,
+      "step": 5704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.308139534883722e-08,
+      "loss": 0.2743,
+      "step": 5705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.303986710963454e-08,
+      "loss": 0.2659,
+      "step": 5706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.299833887043189e-08,
+      "loss": 0.2493,
+      "step": 5707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.295681063122923e-08,
+      "loss": 0.2903,
+      "step": 5708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.291528239202658e-08,
+      "loss": 0.3334,
+      "step": 5709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.287375415282391e-08,
+      "loss": 0.287,
+      "step": 5710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.283222591362126e-08,
+      "loss": 0.3805,
+      "step": 5711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.27906976744186e-08,
+      "loss": 0.2719,
+      "step": 5712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.274916943521594e-08,
+      "loss": 0.3245,
+      "step": 5713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.270764119601329e-08,
+      "loss": 0.3183,
+      "step": 5714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.266611295681062e-08,
+      "loss": 0.2847,
+      "step": 5715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.262458471760797e-08,
+      "loss": 0.2398,
+      "step": 5716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.258305647840531e-08,
+      "loss": 0.2813,
+      "step": 5717
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.254152823920266e-08,
+      "loss": 0.2882,
+      "step": 5718
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.25e-08,
+      "loss": 0.2663,
+      "step": 5719
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.245847176079735e-08,
+      "loss": 0.2797,
+      "step": 5720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.241694352159468e-08,
+      "loss": 0.2797,
+      "step": 5721
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.237541528239202e-08,
+      "loss": 0.3105,
+      "step": 5722
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.233388704318937e-08,
+      "loss": 0.2618,
+      "step": 5723
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.229235880398671e-08,
+      "loss": 0.3114,
+      "step": 5724
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 6.225083056478406e-08,
+      "loss": 0.3472,
+      "step": 5725
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.22093023255814e-08,
+      "loss": 0.291,
+      "step": 5726
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.216777408637874e-08,
+      "loss": 0.2666,
+      "step": 5727
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.212624584717607e-08,
+      "loss": 0.2747,
+      "step": 5728
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.208471760797342e-08,
+      "loss": 0.2845,
+      "step": 5729
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.204318936877075e-08,
+      "loss": 0.2953,
+      "step": 5730
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.20016611295681e-08,
+      "loss": 0.2756,
+      "step": 5731
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.196013289036544e-08,
+      "loss": 0.2246,
+      "step": 5732
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.191860465116279e-08,
+      "loss": 0.2989,
+      "step": 5733
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.187707641196013e-08,
+      "loss": 0.2955,
+      "step": 5734
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.183554817275748e-08,
+      "loss": 0.3026,
+      "step": 5735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.179401993355482e-08,
+      "loss": 0.2787,
+      "step": 5736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.175249169435215e-08,
+      "loss": 0.2876,
+      "step": 5737
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.17109634551495e-08,
+      "loss": 0.3412,
+      "step": 5738
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.166943521594684e-08,
+      "loss": 0.2591,
+      "step": 5739
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.162790697674419e-08,
+      "loss": 0.3022,
+      "step": 5740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.158637873754153e-08,
+      "loss": 0.272,
+      "step": 5741
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.154485049833888e-08,
+      "loss": 0.2611,
+      "step": 5742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.150332225913621e-08,
+      "loss": 0.2643,
+      "step": 5743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.146179401993355e-08,
+      "loss": 0.3082,
+      "step": 5744
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.14202657807309e-08,
+      "loss": 0.2826,
+      "step": 5745
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.137873754152824e-08,
+      "loss": 0.318,
+      "step": 5746
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.133720930232557e-08,
+      "loss": 0.2904,
+      "step": 5747
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.129568106312292e-08,
+      "loss": 0.2759,
+      "step": 5748
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.125415282392026e-08,
+      "loss": 0.2625,
+      "step": 5749
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.121262458471761e-08,
+      "loss": 0.3047,
+      "step": 5750
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.117109634551495e-08,
+      "loss": 0.3261,
+      "step": 5751
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.112956810631228e-08,
+      "loss": 0.3057,
+      "step": 5752
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.108803986710963e-08,
+      "loss": 0.2511,
+      "step": 5753
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.104651162790697e-08,
+      "loss": 0.3079,
+      "step": 5754
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.100498338870432e-08,
+      "loss": 0.2877,
+      "step": 5755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.096345514950166e-08,
+      "loss": 0.2705,
+      "step": 5756
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.092192691029901e-08,
+      "loss": 0.2211,
+      "step": 5757
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.088039867109634e-08,
+      "loss": 0.2709,
+      "step": 5758
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.083887043189368e-08,
+      "loss": 0.2753,
+      "step": 5759
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.079734219269103e-08,
+      "loss": 0.2852,
+      "step": 5760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 6.075581395348837e-08,
+      "loss": 0.2838,
+      "step": 5761
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.071428571428572e-08,
+      "loss": 0.2913,
+      "step": 5762
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.067275747508305e-08,
+      "loss": 0.3484,
+      "step": 5763
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.06312292358804e-08,
+      "loss": 0.2808,
+      "step": 5764
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.058970099667774e-08,
+      "loss": 0.2375,
+      "step": 5765
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.054817275747508e-08,
+      "loss": 0.3671,
+      "step": 5766
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.050664451827241e-08,
+      "loss": 0.3167,
+      "step": 5767
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.046511627906976e-08,
+      "loss": 0.3102,
+      "step": 5768
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.04235880398671e-08,
+      "loss": 0.2597,
+      "step": 5769
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.038205980066445e-08,
+      "loss": 0.3281,
+      "step": 5770
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.034053156146179e-08,
+      "loss": 0.2429,
+      "step": 5771
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.029900332225914e-08,
+      "loss": 0.2124,
+      "step": 5772
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.025747508305647e-08,
+      "loss": 0.2935,
+      "step": 5773
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.021594684385381e-08,
+      "loss": 0.3033,
+      "step": 5774
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.017441860465116e-08,
+      "loss": 0.2897,
+      "step": 5775
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.01328903654485e-08,
+      "loss": 0.2597,
+      "step": 5776
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.009136212624585e-08,
+      "loss": 0.2602,
+      "step": 5777
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.004983388704319e-08,
+      "loss": 0.2891,
+      "step": 5778
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 6.000830564784054e-08,
+      "loss": 0.2582,
+      "step": 5779
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.996677740863787e-08,
+      "loss": 0.2789,
+      "step": 5780
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.992524916943521e-08,
+      "loss": 0.3105,
+      "step": 5781
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.988372093023256e-08,
+      "loss": 0.2747,
+      "step": 5782
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.98421926910299e-08,
+      "loss": 0.2438,
+      "step": 5783
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.980066445182725e-08,
+      "loss": 0.2813,
+      "step": 5784
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.975913621262458e-08,
+      "loss": 0.2739,
+      "step": 5785
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.971760797342193e-08,
+      "loss": 0.2612,
+      "step": 5786
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.967607973421927e-08,
+      "loss": 0.2733,
+      "step": 5787
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.96345514950166e-08,
+      "loss": 0.3154,
+      "step": 5788
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.959302325581395e-08,
+      "loss": 0.2913,
+      "step": 5789
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.955149501661129e-08,
+      "loss": 0.2781,
+      "step": 5790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9509966777408636e-08,
+      "loss": 0.2521,
+      "step": 5791
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.946843853820598e-08,
+      "loss": 0.2422,
+      "step": 5792
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.942691029900332e-08,
+      "loss": 0.3428,
+      "step": 5793
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9385382059800666e-08,
+      "loss": 0.3196,
+      "step": 5794
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9343853820598e-08,
+      "loss": 0.3347,
+      "step": 5795
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.930232558139534e-08,
+      "loss": 0.3005,
+      "step": 5796
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.9260797342192683e-08,
+      "loss": 0.2896,
+      "step": 5797
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.921926910299003e-08,
+      "loss": 0.3173,
+      "step": 5798
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.917774086378737e-08,
+      "loss": 0.3579,
+      "step": 5799
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.9136212624584714e-08,
+      "loss": 0.3029,
+      "step": 5800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.909468438538206e-08,
+      "loss": 0.2109,
+      "step": 5801
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.90531561461794e-08,
+      "loss": 0.2271,
+      "step": 5802
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.901162790697674e-08,
+      "loss": 0.294,
+      "step": 5803
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.897009966777408e-08,
+      "loss": 0.3193,
+      "step": 5804
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8928571428571424e-08,
+      "loss": 0.31,
+      "step": 5805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.888704318936877e-08,
+      "loss": 0.2932,
+      "step": 5806
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.884551495016611e-08,
+      "loss": 0.2687,
+      "step": 5807
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8803986710963454e-08,
+      "loss": 0.3316,
+      "step": 5808
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.87624584717608e-08,
+      "loss": 0.2767,
+      "step": 5809
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8720930232558134e-08,
+      "loss": 0.2736,
+      "step": 5810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.867940199335548e-08,
+      "loss": 0.2619,
+      "step": 5811
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.863787375415282e-08,
+      "loss": 0.2645,
+      "step": 5812
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8596345514950165e-08,
+      "loss": 0.3074,
+      "step": 5813
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.855481727574751e-08,
+      "loss": 0.3088,
+      "step": 5814
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.851328903654485e-08,
+      "loss": 0.3502,
+      "step": 5815
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8471760797342195e-08,
+      "loss": 0.3046,
+      "step": 5816
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.843023255813954e-08,
+      "loss": 0.2558,
+      "step": 5817
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.838870431893687e-08,
+      "loss": 0.315,
+      "step": 5818
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.834717607973421e-08,
+      "loss": 0.2656,
+      "step": 5819
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8305647840531555e-08,
+      "loss": 0.2752,
+      "step": 5820
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.82641196013289e-08,
+      "loss": 0.2724,
+      "step": 5821
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.822259136212624e-08,
+      "loss": 0.3452,
+      "step": 5822
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8181063122923585e-08,
+      "loss": 0.273,
+      "step": 5823
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.813953488372093e-08,
+      "loss": 0.2849,
+      "step": 5824
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8098006644518266e-08,
+      "loss": 0.2095,
+      "step": 5825
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.805647840531561e-08,
+      "loss": 0.2621,
+      "step": 5826
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.801495016611295e-08,
+      "loss": 0.3071,
+      "step": 5827
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.7973421926910296e-08,
+      "loss": 0.2413,
+      "step": 5828
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.793189368770764e-08,
+      "loss": 0.3031,
+      "step": 5829
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.789036544850498e-08,
+      "loss": 0.2644,
+      "step": 5830
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.7848837209302326e-08,
+      "loss": 0.2768,
+      "step": 5831
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.780730897009967e-08,
+      "loss": 0.2828,
+      "step": 5832
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.7765780730897006e-08,
+      "loss": 0.3466,
+      "step": 5833
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.772425249169435e-08,
+      "loss": 0.2977,
+      "step": 5834
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.768272425249169e-08,
+      "loss": 0.2574,
+      "step": 5835
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7641196013289036e-08,
+      "loss": 0.2834,
+      "step": 5836
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.759966777408638e-08,
+      "loss": 0.2418,
+      "step": 5837
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.755813953488372e-08,
+      "loss": 0.2487,
+      "step": 5838
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7516611295681067e-08,
+      "loss": 0.2775,
+      "step": 5839
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.74750830564784e-08,
+      "loss": 0.3219,
+      "step": 5840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.743355481727574e-08,
+      "loss": 0.3176,
+      "step": 5841
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7392026578073084e-08,
+      "loss": 0.3092,
+      "step": 5842
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.735049833887043e-08,
+      "loss": 0.3389,
+      "step": 5843
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.730897009966777e-08,
+      "loss": 0.3354,
+      "step": 5844
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7267441860465114e-08,
+      "loss": 0.2843,
+      "step": 5845
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.722591362126246e-08,
+      "loss": 0.2659,
+      "step": 5846
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.71843853820598e-08,
+      "loss": 0.2859,
+      "step": 5847
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.714285714285714e-08,
+      "loss": 0.2755,
+      "step": 5848
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.710132890365448e-08,
+      "loss": 0.2664,
+      "step": 5849
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.7059800664451824e-08,
+      "loss": 0.2668,
+      "step": 5850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.701827242524917e-08,
+      "loss": 0.2813,
+      "step": 5851
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.697674418604651e-08,
+      "loss": 0.2714,
+      "step": 5852
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6935215946843854e-08,
+      "loss": 0.2637,
+      "step": 5853
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.68936877076412e-08,
+      "loss": 0.3114,
+      "step": 5854
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6852159468438535e-08,
+      "loss": 0.2598,
+      "step": 5855
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.681063122923588e-08,
+      "loss": 0.2804,
+      "step": 5856
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.676910299003322e-08,
+      "loss": 0.2652,
+      "step": 5857
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6727574750830565e-08,
+      "loss": 0.2402,
+      "step": 5858
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.668604651162791e-08,
+      "loss": 0.2493,
+      "step": 5859
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.664451827242525e-08,
+      "loss": 0.2652,
+      "step": 5860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6602990033222595e-08,
+      "loss": 0.2467,
+      "step": 5861
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.656146179401993e-08,
+      "loss": 0.2855,
+      "step": 5862
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.651993355481727e-08,
+      "loss": 0.2808,
+      "step": 5863
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.647840531561461e-08,
+      "loss": 0.2452,
+      "step": 5864
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6436877076411955e-08,
+      "loss": 0.2692,
+      "step": 5865
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.63953488372093e-08,
+      "loss": 0.2885,
+      "step": 5866
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.635382059800664e-08,
+      "loss": 0.2713,
+      "step": 5867
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.6312292358803986e-08,
+      "loss": 0.2378,
+      "step": 5868
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.627076411960133e-08,
+      "loss": 0.2905,
+      "step": 5869
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.6229235880398666e-08,
+      "loss": 0.3072,
+      "step": 5870
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.618770764119601e-08,
+      "loss": 0.3114,
+      "step": 5871
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.614617940199335e-08,
+      "loss": 0.2715,
+      "step": 5872
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.6104651162790696e-08,
+      "loss": 0.2885,
+      "step": 5873
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.606312292358804e-08,
+      "loss": 0.2966,
+      "step": 5874
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.602159468438538e-08,
+      "loss": 0.3238,
+      "step": 5875
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5980066445182726e-08,
+      "loss": 0.2923,
+      "step": 5876
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.593853820598006e-08,
+      "loss": 0.26,
+      "step": 5877
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.58970099667774e-08,
+      "loss": 0.3235,
+      "step": 5878
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.585548172757474e-08,
+      "loss": 0.2755,
+      "step": 5879
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5813953488372087e-08,
+      "loss": 0.2856,
+      "step": 5880
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.577242524916943e-08,
+      "loss": 0.3273,
+      "step": 5881
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5730897009966773e-08,
+      "loss": 0.3078,
+      "step": 5882
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.568936877076412e-08,
+      "loss": 0.2998,
+      "step": 5883
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.564784053156146e-08,
+      "loss": 0.3326,
+      "step": 5884
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.56063122923588e-08,
+      "loss": 0.3196,
+      "step": 5885
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.556478405315614e-08,
+      "loss": 0.2532,
+      "step": 5886
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5523255813953484e-08,
+      "loss": 0.261,
+      "step": 5887
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.548172757475083e-08,
+      "loss": 0.2818,
+      "step": 5888
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.544019933554817e-08,
+      "loss": 0.2527,
+      "step": 5889
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5398671096345514e-08,
+      "loss": 0.2236,
+      "step": 5890
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.535714285714286e-08,
+      "loss": 0.275,
+      "step": 5891
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5315614617940194e-08,
+      "loss": 0.3345,
+      "step": 5892
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.527408637873754e-08,
+      "loss": 0.2509,
+      "step": 5893
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.523255813953488e-08,
+      "loss": 0.2555,
+      "step": 5894
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5191029900332224e-08,
+      "loss": 0.2867,
+      "step": 5895
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.514950166112957e-08,
+      "loss": 0.3168,
+      "step": 5896
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.510797342192691e-08,
+      "loss": 0.3121,
+      "step": 5897
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.5066445182724255e-08,
+      "loss": 0.3131,
+      "step": 5898
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.50249169435216e-08,
+      "loss": 0.3014,
+      "step": 5899
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.498338870431893e-08,
+      "loss": 0.2588,
+      "step": 5900
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.494186046511627e-08,
+      "loss": 0.312,
+      "step": 5901
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4900332225913615e-08,
+      "loss": 0.2597,
+      "step": 5902
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.485880398671096e-08,
+      "loss": 0.2422,
+      "step": 5903
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.48172757475083e-08,
+      "loss": 0.3455,
+      "step": 5904
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.4775747508305645e-08,
+      "loss": 0.2607,
+      "step": 5905
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.473421926910299e-08,
+      "loss": 0.2099,
+      "step": 5906
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4692691029900325e-08,
+      "loss": 0.2926,
+      "step": 5907
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.465116279069767e-08,
+      "loss": 0.3099,
+      "step": 5908
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.460963455149501e-08,
+      "loss": 0.2903,
+      "step": 5909
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4568106312292356e-08,
+      "loss": 0.312,
+      "step": 5910
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.45265780730897e-08,
+      "loss": 0.2529,
+      "step": 5911
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.448504983388704e-08,
+      "loss": 0.2931,
+      "step": 5912
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4443521594684386e-08,
+      "loss": 0.2687,
+      "step": 5913
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.440199335548173e-08,
+      "loss": 0.2741,
+      "step": 5914
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4360465116279066e-08,
+      "loss": 0.2192,
+      "step": 5915
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.431893687707641e-08,
+      "loss": 0.2855,
+      "step": 5916
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.427740863787375e-08,
+      "loss": 0.3155,
+      "step": 5917
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4235880398671096e-08,
+      "loss": 0.3003,
+      "step": 5918
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.419435215946844e-08,
+      "loss": 0.2421,
+      "step": 5919
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.415282392026578e-08,
+      "loss": 0.3377,
+      "step": 5920
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4111295681063126e-08,
+      "loss": 0.2981,
+      "step": 5921
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.4069767441860457e-08,
+      "loss": 0.3637,
+      "step": 5922
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.40282392026578e-08,
+      "loss": 0.3115,
+      "step": 5923
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3986710963455143e-08,
+      "loss": 0.3097,
+      "step": 5924
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.394518272425249e-08,
+      "loss": 0.3055,
+      "step": 5925
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.390365448504983e-08,
+      "loss": 0.3467,
+      "step": 5926
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3862126245847174e-08,
+      "loss": 0.207,
+      "step": 5927
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.382059800664452e-08,
+      "loss": 0.2547,
+      "step": 5928
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.377906976744186e-08,
+      "loss": 0.2761,
+      "step": 5929
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.37375415282392e-08,
+      "loss": 0.281,
+      "step": 5930
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.369601328903654e-08,
+      "loss": 0.3183,
+      "step": 5931
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3654485049833884e-08,
+      "loss": 0.2951,
+      "step": 5932
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.361295681063123e-08,
+      "loss": 0.2939,
+      "step": 5933
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.357142857142857e-08,
+      "loss": 0.2945,
+      "step": 5934
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3529900332225914e-08,
+      "loss": 0.3241,
+      "step": 5935
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.348837209302326e-08,
+      "loss": 0.2832,
+      "step": 5936
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3446843853820594e-08,
+      "loss": 0.3336,
+      "step": 5937
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.340531561461794e-08,
+      "loss": 0.3174,
+      "step": 5938
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.336378737541528e-08,
+      "loss": 0.2575,
+      "step": 5939
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.3322259136212625e-08,
+      "loss": 0.2655,
+      "step": 5940
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.328073089700997e-08,
+      "loss": 0.2734,
+      "step": 5941
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.323920265780731e-08,
+      "loss": 0.2696,
+      "step": 5942
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.3197674418604655e-08,
+      "loss": 0.277,
+      "step": 5943
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.315614617940199e-08,
+      "loss": 0.3349,
+      "step": 5944
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.311461794019933e-08,
+      "loss": 0.2345,
+      "step": 5945
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.307308970099667e-08,
+      "loss": 0.3312,
+      "step": 5946
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.3031561461794015e-08,
+      "loss": 0.2495,
+      "step": 5947
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.299003322259136e-08,
+      "loss": 0.2816,
+      "step": 5948
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.29485049833887e-08,
+      "loss": 0.3725,
+      "step": 5949
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2906976744186046e-08,
+      "loss": 0.2638,
+      "step": 5950
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.286544850498339e-08,
+      "loss": 0.2772,
+      "step": 5951
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2823920265780726e-08,
+      "loss": 0.2796,
+      "step": 5952
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.278239202657807e-08,
+      "loss": 0.2492,
+      "step": 5953
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.274086378737541e-08,
+      "loss": 0.2759,
+      "step": 5954
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2699335548172756e-08,
+      "loss": 0.3076,
+      "step": 5955
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.26578073089701e-08,
+      "loss": 0.3286,
+      "step": 5956
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.261627906976744e-08,
+      "loss": 0.327,
+      "step": 5957
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2574750830564786e-08,
+      "loss": 0.3031,
+      "step": 5958
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.253322259136213e-08,
+      "loss": 0.3071,
+      "step": 5959
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.249169435215946e-08,
+      "loss": 0.281,
+      "step": 5960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.24501661129568e-08,
+      "loss": 0.2626,
+      "step": 5961
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2408637873754146e-08,
+      "loss": 0.3151,
+      "step": 5962
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.236710963455149e-08,
+      "loss": 0.2714,
+      "step": 5963
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.232558139534883e-08,
+      "loss": 0.3084,
+      "step": 5964
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.228405315614618e-08,
+      "loss": 0.317,
+      "step": 5965
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.224252491694352e-08,
+      "loss": 0.3495,
+      "step": 5966
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.220099667774086e-08,
+      "loss": 0.308,
+      "step": 5967
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.21594684385382e-08,
+      "loss": 0.3056,
+      "step": 5968
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.2117940199335544e-08,
+      "loss": 0.3317,
+      "step": 5969
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.207641196013289e-08,
+      "loss": 0.3159,
+      "step": 5970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.203488372093023e-08,
+      "loss": 0.2947,
+      "step": 5971
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1993355481727574e-08,
+      "loss": 0.3094,
+      "step": 5972
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.195182724252492e-08,
+      "loss": 0.2278,
+      "step": 5973
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1910299003322254e-08,
+      "loss": 0.268,
+      "step": 5974
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.18687707641196e-08,
+      "loss": 0.3708,
+      "step": 5975
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.182724252491694e-08,
+      "loss": 0.326,
+      "step": 5976
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.1785714285714284e-08,
+      "loss": 0.3345,
+      "step": 5977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.174418604651163e-08,
+      "loss": 0.272,
+      "step": 5978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.170265780730897e-08,
+      "loss": 0.2863,
+      "step": 5979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1661129568106315e-08,
+      "loss": 0.3056,
+      "step": 5980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.161960132890366e-08,
+      "loss": 0.3218,
+      "step": 5981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.157807308970099e-08,
+      "loss": 0.2823,
+      "step": 5982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.153654485049833e-08,
+      "loss": 0.3012,
+      "step": 5983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1495016611295675e-08,
+      "loss": 0.263,
+      "step": 5984
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.145348837209302e-08,
+      "loss": 0.2815,
+      "step": 5985
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.141196013289036e-08,
+      "loss": 0.2676,
+      "step": 5986
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1370431893687705e-08,
+      "loss": 0.2897,
+      "step": 5987
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.132890365448505e-08,
+      "loss": 0.2867,
+      "step": 5988
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1287375415282385e-08,
+      "loss": 0.2702,
+      "step": 5989
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.124584717607973e-08,
+      "loss": 0.3349,
+      "step": 5990
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.120431893687707e-08,
+      "loss": 0.2833,
+      "step": 5991
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1162790697674416e-08,
+      "loss": 0.2755,
+      "step": 5992
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.112126245847176e-08,
+      "loss": 0.3024,
+      "step": 5993
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.10797342192691e-08,
+      "loss": 0.267,
+      "step": 5994
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.1038205980066446e-08,
+      "loss": 0.2479,
+      "step": 5995
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.099667774086379e-08,
+      "loss": 0.2938,
+      "step": 5996
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0955149501661126e-08,
+      "loss": 0.2784,
+      "step": 5997
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.091362126245847e-08,
+      "loss": 0.2588,
+      "step": 5998
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.087209302325581e-08,
+      "loss": 0.2731,
+      "step": 5999
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0830564784053156e-08,
+      "loss": 0.3217,
+      "step": 6000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.07890365448505e-08,
+      "loss": 0.2459,
+      "step": 6001
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.074750830564784e-08,
+      "loss": 0.2916,
+      "step": 6002
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0705980066445186e-08,
+      "loss": 0.3046,
+      "step": 6003
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0664451827242517e-08,
+      "loss": 0.3315,
+      "step": 6004
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.062292358803986e-08,
+      "loss": 0.28,
+      "step": 6005
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0581395348837203e-08,
+      "loss": 0.3527,
+      "step": 6006
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.053986710963455e-08,
+      "loss": 0.3158,
+      "step": 6007
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.049833887043189e-08,
+      "loss": 0.2874,
+      "step": 6008
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.0456810631229234e-08,
+      "loss": 0.31,
+      "step": 6009
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.041528239202658e-08,
+      "loss": 0.254,
+      "step": 6010
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.037375415282392e-08,
+      "loss": 0.2786,
+      "step": 6011
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.033222591362126e-08,
+      "loss": 0.2958,
+      "step": 6012
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.02906976744186e-08,
+      "loss": 0.2901,
+      "step": 6013
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.0249169435215944e-08,
+      "loss": 0.2959,
+      "step": 6014
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.020764119601329e-08,
+      "loss": 0.2883,
+      "step": 6015
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.016611295681063e-08,
+      "loss": 0.253,
+      "step": 6016
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.0124584717607974e-08,
+      "loss": 0.2856,
+      "step": 6017
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.008305647840532e-08,
+      "loss": 0.2325,
+      "step": 6018
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.0041528239202654e-08,
+      "loss": 0.2788,
+      "step": 6019
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5e-08,
+      "loss": 0.2909,
+      "step": 6020
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.995847176079734e-08,
+      "loss": 0.2892,
+      "step": 6021
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9916943521594685e-08,
+      "loss": 0.3085,
+      "step": 6022
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.987541528239203e-08,
+      "loss": 0.2662,
+      "step": 6023
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.983388704318937e-08,
+      "loss": 0.3039,
+      "step": 6024
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9792358803986715e-08,
+      "loss": 0.3198,
+      "step": 6025
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.975083056478405e-08,
+      "loss": 0.3149,
+      "step": 6026
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.970930232558139e-08,
+      "loss": 0.2861,
+      "step": 6027
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.966777408637873e-08,
+      "loss": 0.2548,
+      "step": 6028
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9626245847176075e-08,
+      "loss": 0.2499,
+      "step": 6029
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.958471760797342e-08,
+      "loss": 0.2586,
+      "step": 6030
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.954318936877076e-08,
+      "loss": 0.2795,
+      "step": 6031
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9501661129568105e-08,
+      "loss": 0.2709,
+      "step": 6032
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.946013289036545e-08,
+      "loss": 0.2575,
+      "step": 6033
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9418604651162786e-08,
+      "loss": 0.2949,
+      "step": 6034
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.937707641196013e-08,
+      "loss": 0.2842,
+      "step": 6035
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.933554817275747e-08,
+      "loss": 0.2913,
+      "step": 6036
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9294019933554816e-08,
+      "loss": 0.2689,
+      "step": 6037
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.925249169435216e-08,
+      "loss": 0.2746,
+      "step": 6038
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.92109634551495e-08,
+      "loss": 0.3472,
+      "step": 6039
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9169435215946846e-08,
+      "loss": 0.2472,
+      "step": 6040
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.912790697674419e-08,
+      "loss": 0.3179,
+      "step": 6041
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.908637873754152e-08,
+      "loss": 0.3206,
+      "step": 6042
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.904485049833886e-08,
+      "loss": 0.2784,
+      "step": 6043
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.9003322259136206e-08,
+      "loss": 0.2993,
+      "step": 6044
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.896179401993355e-08,
+      "loss": 0.3166,
+      "step": 6045
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.892026578073089e-08,
+      "loss": 0.2567,
+      "step": 6046
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.8878737541528237e-08,
+      "loss": 0.2935,
+      "step": 6047
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.883720930232558e-08,
+      "loss": 0.2987,
+      "step": 6048
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.879568106312292e-08,
+      "loss": 0.2696,
+      "step": 6049
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.875415282392026e-08,
+      "loss": 0.2668,
+      "step": 6050
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8712624584717604e-08,
+      "loss": 0.3185,
+      "step": 6051
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.867109634551495e-08,
+      "loss": 0.3003,
+      "step": 6052
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.862956810631229e-08,
+      "loss": 0.2542,
+      "step": 6053
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8588039867109634e-08,
+      "loss": 0.2934,
+      "step": 6054
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.854651162790698e-08,
+      "loss": 0.287,
+      "step": 6055
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.850498338870432e-08,
+      "loss": 0.3013,
+      "step": 6056
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.846345514950166e-08,
+      "loss": 0.3096,
+      "step": 6057
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8421926910299e-08,
+      "loss": 0.2794,
+      "step": 6058
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8380398671096344e-08,
+      "loss": 0.2661,
+      "step": 6059
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.833887043189369e-08,
+      "loss": 0.3286,
+      "step": 6060
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.829734219269103e-08,
+      "loss": 0.3201,
+      "step": 6061
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8255813953488374e-08,
+      "loss": 0.3647,
+      "step": 6062
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.821428571428572e-08,
+      "loss": 0.319,
+      "step": 6063
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.817275747508305e-08,
+      "loss": 0.3166,
+      "step": 6064
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.813122923588039e-08,
+      "loss": 0.2943,
+      "step": 6065
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.8089700996677735e-08,
+      "loss": 0.2536,
+      "step": 6066
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.804817275747508e-08,
+      "loss": 0.225,
+      "step": 6067
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.800664451827242e-08,
+      "loss": 0.3399,
+      "step": 6068
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7965116279069765e-08,
+      "loss": 0.2669,
+      "step": 6069
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.792358803986711e-08,
+      "loss": 0.2966,
+      "step": 6070
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7882059800664445e-08,
+      "loss": 0.3521,
+      "step": 6071
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.784053156146179e-08,
+      "loss": 0.2582,
+      "step": 6072
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.779900332225913e-08,
+      "loss": 0.2842,
+      "step": 6073
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7757475083056475e-08,
+      "loss": 0.2572,
+      "step": 6074
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.771594684385382e-08,
+      "loss": 0.2757,
+      "step": 6075
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.767441860465116e-08,
+      "loss": 0.2982,
+      "step": 6076
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7632890365448506e-08,
+      "loss": 0.2736,
+      "step": 6077
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.759136212624585e-08,
+      "loss": 0.3063,
+      "step": 6078
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7549833887043186e-08,
+      "loss": 0.2395,
+      "step": 6079
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.750830564784053e-08,
+      "loss": 0.3076,
+      "step": 6080
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.746677740863787e-08,
+      "loss": 0.259,
+      "step": 6081
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7425249169435216e-08,
+      "loss": 0.2279,
+      "step": 6082
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.738372093023256e-08,
+      "loss": 0.3252,
+      "step": 6083
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.73421926910299e-08,
+      "loss": 0.2893,
+      "step": 6084
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7300664451827246e-08,
+      "loss": 0.2553,
+      "step": 6085
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.7259136212624576e-08,
+      "loss": 0.3303,
+      "step": 6086
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.721760797342192e-08,
+      "loss": 0.246,
+      "step": 6087
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.717607973421926e-08,
+      "loss": 0.2845,
+      "step": 6088
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7134551495016607e-08,
+      "loss": 0.2549,
+      "step": 6089
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.709302325581395e-08,
+      "loss": 0.2665,
+      "step": 6090
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.7051495016611293e-08,
+      "loss": 0.2778,
+      "step": 6091
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.700996677740864e-08,
+      "loss": 0.3094,
+      "step": 6092
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.696843853820598e-08,
+      "loss": 0.2739,
+      "step": 6093
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.692691029900332e-08,
+      "loss": 0.2349,
+      "step": 6094
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.688538205980066e-08,
+      "loss": 0.2825,
+      "step": 6095
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6843853820598004e-08,
+      "loss": 0.2626,
+      "step": 6096
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.680232558139535e-08,
+      "loss": 0.2583,
+      "step": 6097
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.676079734219269e-08,
+      "loss": 0.2673,
+      "step": 6098
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6719269102990034e-08,
+      "loss": 0.3064,
+      "step": 6099
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.667774086378738e-08,
+      "loss": 0.2654,
+      "step": 6100
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6636212624584714e-08,
+      "loss": 0.2752,
+      "step": 6101
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.659468438538206e-08,
+      "loss": 0.3272,
+      "step": 6102
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.65531561461794e-08,
+      "loss": 0.2914,
+      "step": 6103
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6511627906976744e-08,
+      "loss": 0.2946,
+      "step": 6104
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.647009966777409e-08,
+      "loss": 0.254,
+      "step": 6105
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.642857142857143e-08,
+      "loss": 0.3292,
+      "step": 6106
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6387043189368775e-08,
+      "loss": 0.2245,
+      "step": 6107
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.634551495016611e-08,
+      "loss": 0.3404,
+      "step": 6108
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.630398671096345e-08,
+      "loss": 0.3373,
+      "step": 6109
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.626245847176079e-08,
+      "loss": 0.2733,
+      "step": 6110
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6220930232558135e-08,
+      "loss": 0.2885,
+      "step": 6111
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.617940199335548e-08,
+      "loss": 0.2664,
+      "step": 6112
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.613787375415282e-08,
+      "loss": 0.2701,
+      "step": 6113
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6096345514950165e-08,
+      "loss": 0.3638,
+      "step": 6114
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.605481727574751e-08,
+      "loss": 0.2416,
+      "step": 6115
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.6013289036544845e-08,
+      "loss": 0.3278,
+      "step": 6116
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.597176079734219e-08,
+      "loss": 0.3748,
+      "step": 6117
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.593023255813953e-08,
+      "loss": 0.3012,
+      "step": 6118
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.5888704318936876e-08,
+      "loss": 0.3074,
+      "step": 6119
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.584717607973422e-08,
+      "loss": 0.3059,
+      "step": 6120
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.580564784053156e-08,
+      "loss": 0.2947,
+      "step": 6121
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.5764119601328906e-08,
+      "loss": 0.3117,
+      "step": 6122
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.572259136212625e-08,
+      "loss": 0.2658,
+      "step": 6123
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.568106312292358e-08,
+      "loss": 0.3279,
+      "step": 6124
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.563953488372092e-08,
+      "loss": 0.3648,
+      "step": 6125
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5598006644518266e-08,
+      "loss": 0.2394,
+      "step": 6126
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.555647840531561e-08,
+      "loss": 0.2923,
+      "step": 6127
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.551495016611295e-08,
+      "loss": 0.2588,
+      "step": 6128
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5473421926910296e-08,
+      "loss": 0.2544,
+      "step": 6129
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.543189368770764e-08,
+      "loss": 0.293,
+      "step": 6130
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5390365448504977e-08,
+      "loss": 0.2768,
+      "step": 6131
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.534883720930232e-08,
+      "loss": 0.3129,
+      "step": 6132
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.534883720930232e-08,
+      "loss": 0.2623,
+      "step": 6133
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5307308970099663e-08,
+      "loss": 0.2324,
+      "step": 6134
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.526578073089701e-08,
+      "loss": 0.214,
+      "step": 6135
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.522425249169435e-08,
+      "loss": 0.2688,
+      "step": 6136
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.5182724252491694e-08,
+      "loss": 0.2594,
+      "step": 6137
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.514119601328904e-08,
+      "loss": 0.2851,
+      "step": 6138
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.509966777408638e-08,
+      "loss": 0.3369,
+      "step": 6139
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.505813953488372e-08,
+      "loss": 0.3241,
+      "step": 6140
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.501661129568106e-08,
+      "loss": 0.2714,
+      "step": 6141
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4975083056478404e-08,
+      "loss": 0.2943,
+      "step": 6142
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.493355481727575e-08,
+      "loss": 0.3244,
+      "step": 6143
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.489202657807309e-08,
+      "loss": 0.2735,
+      "step": 6144
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4850498338870434e-08,
+      "loss": 0.3514,
+      "step": 6145
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.480897009966778e-08,
+      "loss": 0.3125,
+      "step": 6146
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.476744186046511e-08,
+      "loss": 0.2741,
+      "step": 6147
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.472591362126245e-08,
+      "loss": 0.3697,
+      "step": 6148
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4684385382059795e-08,
+      "loss": 0.3011,
+      "step": 6149
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.464285714285714e-08,
+      "loss": 0.2925,
+      "step": 6150
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.460132890365448e-08,
+      "loss": 0.2903,
+      "step": 6151
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4559800664451825e-08,
+      "loss": 0.2866,
+      "step": 6152
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.451827242524917e-08,
+      "loss": 0.2547,
+      "step": 6153
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.447674418604651e-08,
+      "loss": 0.2398,
+      "step": 6154
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.443521594684385e-08,
+      "loss": 0.2773,
+      "step": 6155
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.439368770764119e-08,
+      "loss": 0.3034,
+      "step": 6156
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.4352159468438535e-08,
+      "loss": 0.3026,
+      "step": 6157
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.431063122923588e-08,
+      "loss": 0.2338,
+      "step": 6158
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.426910299003322e-08,
+      "loss": 0.2998,
+      "step": 6159
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4227574750830565e-08,
+      "loss": 0.3058,
+      "step": 6160
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.418604651162791e-08,
+      "loss": 0.3392,
+      "step": 6161
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4144518272425246e-08,
+      "loss": 0.3005,
+      "step": 6162
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.410299003322259e-08,
+      "loss": 0.2779,
+      "step": 6163
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.406146179401993e-08,
+      "loss": 0.2675,
+      "step": 6164
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.4019933554817276e-08,
+      "loss": 0.2919,
+      "step": 6165
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.397840531561462e-08,
+      "loss": 0.2677,
+      "step": 6166
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.393687707641196e-08,
+      "loss": 0.285,
+      "step": 6167
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3895348837209306e-08,
+      "loss": 0.3192,
+      "step": 6168
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.385382059800665e-08,
+      "loss": 0.2986,
+      "step": 6169
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.381229235880398e-08,
+      "loss": 0.2686,
+      "step": 6170
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.377076411960132e-08,
+      "loss": 0.2962,
+      "step": 6171
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3729235880398666e-08,
+      "loss": 0.3601,
+      "step": 6172
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.368770764119601e-08,
+      "loss": 0.2136,
+      "step": 6173
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.364617940199335e-08,
+      "loss": 0.3294,
+      "step": 6174
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3604651162790697e-08,
+      "loss": 0.3112,
+      "step": 6175
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.356312292358804e-08,
+      "loss": 0.2397,
+      "step": 6176
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.352159468438538e-08,
+      "loss": 0.2936,
+      "step": 6177
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.348006644518272e-08,
+      "loss": 0.2732,
+      "step": 6178
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3438538205980064e-08,
+      "loss": 0.2793,
+      "step": 6179
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.339700996677741e-08,
+      "loss": 0.266,
+      "step": 6180
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.335548172757475e-08,
+      "loss": 0.2544,
+      "step": 6181
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3313953488372094e-08,
+      "loss": 0.3128,
+      "step": 6182
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.327242524916944e-08,
+      "loss": 0.2434,
+      "step": 6183
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3230897009966774e-08,
+      "loss": 0.2911,
+      "step": 6184
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.318936877076412e-08,
+      "loss": 0.2487,
+      "step": 6185
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.314784053156146e-08,
+      "loss": 0.2569,
+      "step": 6186
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3106312292358804e-08,
+      "loss": 0.2636,
+      "step": 6187
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.306478405315615e-08,
+      "loss": 0.2758,
+      "step": 6188
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.302325581395349e-08,
+      "loss": 0.2496,
+      "step": 6189
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2981727574750834e-08,
+      "loss": 0.2369,
+      "step": 6190
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.294019933554817e-08,
+      "loss": 0.2921,
+      "step": 6191
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.289867109634551e-08,
+      "loss": 0.2395,
+      "step": 6192
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.285714285714285e-08,
+      "loss": 0.3486,
+      "step": 6193
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2815614617940195e-08,
+      "loss": 0.3324,
+      "step": 6194
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.277408637873754e-08,
+      "loss": 0.2805,
+      "step": 6195
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.273255813953488e-08,
+      "loss": 0.305,
+      "step": 6196
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2691029900332225e-08,
+      "loss": 0.3143,
+      "step": 6197
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.264950166112957e-08,
+      "loss": 0.3084,
+      "step": 6198
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2607973421926905e-08,
+      "loss": 0.2515,
+      "step": 6199
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.256644518272425e-08,
+      "loss": 0.2697,
+      "step": 6200
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.252491694352159e-08,
+      "loss": 0.2987,
+      "step": 6201
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2483388704318935e-08,
+      "loss": 0.2938,
+      "step": 6202
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.244186046511628e-08,
+      "loss": 0.2487,
+      "step": 6203
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.240033222591362e-08,
+      "loss": 0.2377,
+      "step": 6204
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2358803986710966e-08,
+      "loss": 0.2933,
+      "step": 6205
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.231727574750831e-08,
+      "loss": 0.2771,
+      "step": 6206
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.227574750830564e-08,
+      "loss": 0.2597,
+      "step": 6207
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.223421926910298e-08,
+      "loss": 0.3353,
+      "step": 6208
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2192691029900326e-08,
+      "loss": 0.2877,
+      "step": 6209
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.215116279069767e-08,
+      "loss": 0.3723,
+      "step": 6210
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.210963455149501e-08,
+      "loss": 0.2847,
+      "step": 6211
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2068106312292356e-08,
+      "loss": 0.3084,
+      "step": 6212
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.20265780730897e-08,
+      "loss": 0.3026,
+      "step": 6213
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1985049833887036e-08,
+      "loss": 0.2839,
+      "step": 6214
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.194352159468438e-08,
+      "loss": 0.2724,
+      "step": 6215
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.190199335548172e-08,
+      "loss": 0.2886,
+      "step": 6216
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1860465116279067e-08,
+      "loss": 0.3212,
+      "step": 6217
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.181893687707641e-08,
+      "loss": 0.2965,
+      "step": 6218
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1777408637873753e-08,
+      "loss": 0.3199,
+      "step": 6219
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.17358803986711e-08,
+      "loss": 0.3487,
+      "step": 6220
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.169435215946844e-08,
+      "loss": 0.2696,
+      "step": 6221
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.165282392026578e-08,
+      "loss": 0.369,
+      "step": 6222
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.161129568106312e-08,
+      "loss": 0.2542,
+      "step": 6223
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1569767441860464e-08,
+      "loss": 0.2826,
+      "step": 6224
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.152823920265781e-08,
+      "loss": 0.3799,
+      "step": 6225
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.148671096345515e-08,
+      "loss": 0.3367,
+      "step": 6226
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.1445182724252494e-08,
+      "loss": 0.3269,
+      "step": 6227
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.140365448504984e-08,
+      "loss": 0.3231,
+      "step": 6228
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.136212624584717e-08,
+      "loss": 0.2939,
+      "step": 6229
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.132059800664451e-08,
+      "loss": 0.279,
+      "step": 6230
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1279069767441854e-08,
+      "loss": 0.3032,
+      "step": 6231
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.12375415282392e-08,
+      "loss": 0.2752,
+      "step": 6232
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.119601328903654e-08,
+      "loss": 0.2737,
+      "step": 6233
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.1154485049833885e-08,
+      "loss": 0.2845,
+      "step": 6234
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.111295681063123e-08,
+      "loss": 0.3181,
+      "step": 6235
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.107142857142857e-08,
+      "loss": 0.2904,
+      "step": 6236
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.102990033222591e-08,
+      "loss": 0.2866,
+      "step": 6237
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.098837209302325e-08,
+      "loss": 0.2503,
+      "step": 6238
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0946843853820595e-08,
+      "loss": 0.3169,
+      "step": 6239
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.090531561461794e-08,
+      "loss": 0.3306,
+      "step": 6240
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.086378737541528e-08,
+      "loss": 0.2692,
+      "step": 6241
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0822259136212625e-08,
+      "loss": 0.2992,
+      "step": 6242
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.078073089700997e-08,
+      "loss": 0.2687,
+      "step": 6243
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0739202657807305e-08,
+      "loss": 0.2901,
+      "step": 6244
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.069767441860465e-08,
+      "loss": 0.3112,
+      "step": 6245
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.065614617940199e-08,
+      "loss": 0.304,
+      "step": 6246
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0614617940199336e-08,
+      "loss": 0.2195,
+      "step": 6247
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.057308970099668e-08,
+      "loss": 0.3361,
+      "step": 6248
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.053156146179402e-08,
+      "loss": 0.3264,
+      "step": 6249
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0490033222591366e-08,
+      "loss": 0.258,
+      "step": 6250
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.044850498338871e-08,
+      "loss": 0.2838,
+      "step": 6251
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.040697674418604e-08,
+      "loss": 0.2779,
+      "step": 6252
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.036544850498338e-08,
+      "loss": 0.2984,
+      "step": 6253
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0323920265780726e-08,
+      "loss": 0.2938,
+      "step": 6254
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.028239202657807e-08,
+      "loss": 0.2656,
+      "step": 6255
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.024086378737541e-08,
+      "loss": 0.3497,
+      "step": 6256
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0199335548172756e-08,
+      "loss": 0.3432,
+      "step": 6257
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.01578073089701e-08,
+      "loss": 0.25,
+      "step": 6258
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0116279069767437e-08,
+      "loss": 0.3159,
+      "step": 6259
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.007475083056478e-08,
+      "loss": 0.31,
+      "step": 6260
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.0033222591362123e-08,
+      "loss": 0.2839,
+      "step": 6261
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.999169435215947e-08,
+      "loss": 0.307,
+      "step": 6262
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.995016611295681e-08,
+      "loss": 0.2873,
+      "step": 6263
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.9908637873754154e-08,
+      "loss": 0.3189,
+      "step": 6264
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.98671096345515e-08,
+      "loss": 0.2489,
+      "step": 6265
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 3.982558139534884e-08,
+      "loss": 0.2867,
+      "step": 6266
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.978405315614618e-08,
+      "loss": 0.2336,
+      "step": 6267
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.974252491694352e-08,
+      "loss": 0.2809,
+      "step": 6268
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9700996677740864e-08,
+      "loss": 0.3879,
+      "step": 6269
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.965946843853821e-08,
+      "loss": 0.3151,
+      "step": 6270
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.961794019933555e-08,
+      "loss": 0.3039,
+      "step": 6271
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9576411960132894e-08,
+      "loss": 0.3165,
+      "step": 6272
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.953488372093023e-08,
+      "loss": 0.2492,
+      "step": 6273
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.949335548172757e-08,
+      "loss": 0.2907,
+      "step": 6274
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.945182724252491e-08,
+      "loss": 0.3092,
+      "step": 6275
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9410299003322255e-08,
+      "loss": 0.2906,
+      "step": 6276
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.93687707641196e-08,
+      "loss": 0.2777,
+      "step": 6277
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.932724252491694e-08,
+      "loss": 0.335,
+      "step": 6278
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9285714285714285e-08,
+      "loss": 0.3329,
+      "step": 6279
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.924418604651163e-08,
+      "loss": 0.2674,
+      "step": 6280
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9202657807308965e-08,
+      "loss": 0.3007,
+      "step": 6281
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.916112956810631e-08,
+      "loss": 0.3326,
+      "step": 6282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.911960132890365e-08,
+      "loss": 0.2934,
+      "step": 6283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.9078073089700995e-08,
+      "loss": 0.3191,
+      "step": 6284
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.903654485049834e-08,
+      "loss": 0.2301,
+      "step": 6285
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.899501661129568e-08,
+      "loss": 0.3227,
+      "step": 6286
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8953488372093025e-08,
+      "loss": 0.2695,
+      "step": 6287
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.891196013289037e-08,
+      "loss": 0.3268,
+      "step": 6288
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.88704318936877e-08,
+      "loss": 0.2763,
+      "step": 6289
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.882890365448504e-08,
+      "loss": 0.3143,
+      "step": 6290
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8787375415282386e-08,
+      "loss": 0.2952,
+      "step": 6291
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.874584717607973e-08,
+      "loss": 0.2937,
+      "step": 6292
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.870431893687707e-08,
+      "loss": 0.2382,
+      "step": 6293
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8662790697674416e-08,
+      "loss": 0.2726,
+      "step": 6294
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.862126245847176e-08,
+      "loss": 0.3278,
+      "step": 6295
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8579734219269096e-08,
+      "loss": 0.2996,
+      "step": 6296
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.853820598006644e-08,
+      "loss": 0.3497,
+      "step": 6297
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.849667774086378e-08,
+      "loss": 0.2973,
+      "step": 6298
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.8455149501661126e-08,
+      "loss": 0.2676,
+      "step": 6299
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.841362126245847e-08,
+      "loss": 0.3054,
+      "step": 6300
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.837209302325581e-08,
+      "loss": 0.2847,
+      "step": 6301
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 3.833056478405316e-08,
+      "loss": 0.2923,
+      "step": 6302
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.82890365448505e-08,
+      "loss": 0.2536,
+      "step": 6303
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.824750830564784e-08,
+      "loss": 0.266,
+      "step": 6304
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.820598006644518e-08,
+      "loss": 0.3103,
+      "step": 6305
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.8164451827242524e-08,
+      "loss": 0.2521,
+      "step": 6306
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.812292358803987e-08,
+      "loss": 0.3872,
+      "step": 6307
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.808139534883721e-08,
+      "loss": 0.2955,
+      "step": 6308
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.8039867109634554e-08,
+      "loss": 0.3864,
+      "step": 6309
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.79983388704319e-08,
+      "loss": 0.3479,
+      "step": 6310
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.795681063122923e-08,
+      "loss": 0.3382,
+      "step": 6311
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.791528239202657e-08,
+      "loss": 0.3305,
+      "step": 6312
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7873754152823914e-08,
+      "loss": 0.2863,
+      "step": 6313
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.783222591362126e-08,
+      "loss": 0.2928,
+      "step": 6314
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.77906976744186e-08,
+      "loss": 0.2778,
+      "step": 6315
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7749169435215944e-08,
+      "loss": 0.2421,
+      "step": 6316
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.770764119601329e-08,
+      "loss": 0.2805,
+      "step": 6317
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.766611295681063e-08,
+      "loss": 0.2444,
+      "step": 6318
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.762458471760797e-08,
+      "loss": 0.2993,
+      "step": 6319
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.758305647840531e-08,
+      "loss": 0.3527,
+      "step": 6320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7541528239202655e-08,
+      "loss": 0.3642,
+      "step": 6321
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.75e-08,
+      "loss": 0.3013,
+      "step": 6322
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.745847176079734e-08,
+      "loss": 0.2823,
+      "step": 6323
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.741694352159468e-08,
+      "loss": 0.2921,
+      "step": 6324
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.737541528239202e-08,
+      "loss": 0.3019,
+      "step": 6325
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7333887043189365e-08,
+      "loss": 0.3119,
+      "step": 6326
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.729235880398671e-08,
+      "loss": 0.2217,
+      "step": 6327
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.725083056478405e-08,
+      "loss": 0.3638,
+      "step": 6328
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7209302325581396e-08,
+      "loss": 0.2146,
+      "step": 6329
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.716777408637874e-08,
+      "loss": 0.2561,
+      "step": 6330
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.712624584717608e-08,
+      "loss": 0.2659,
+      "step": 6331
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.708471760797342e-08,
+      "loss": 0.285,
+      "step": 6332
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.704318936877076e-08,
+      "loss": 0.2878,
+      "step": 6333
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.7001661129568106e-08,
+      "loss": 0.29,
+      "step": 6334
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.696013289036545e-08,
+      "loss": 0.2643,
+      "step": 6335
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6918604651162786e-08,
+      "loss": 0.2185,
+      "step": 6336
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.687707641196013e-08,
+      "loss": 0.3426,
+      "step": 6337
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.683554817275747e-08,
+      "loss": 0.3263,
+      "step": 6338
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 3.6794019933554816e-08,
+      "loss": 0.3123,
+      "step": 6339
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.675249169435216e-08,
+      "loss": 0.2992,
+      "step": 6340
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.67109634551495e-08,
+      "loss": 0.2644,
+      "step": 6341
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6669435215946847e-08,
+      "loss": 0.2847,
+      "step": 6342
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6627906976744183e-08,
+      "loss": 0.2906,
+      "step": 6343
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.658637873754153e-08,
+      "loss": 0.265,
+      "step": 6344
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.654485049833887e-08,
+      "loss": 0.2955,
+      "step": 6345
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6503322259136214e-08,
+      "loss": 0.3275,
+      "step": 6346
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.646179401993355e-08,
+      "loss": 0.3092,
+      "step": 6347
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6420265780730894e-08,
+      "loss": 0.2554,
+      "step": 6348
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.637873754152824e-08,
+      "loss": 0.3394,
+      "step": 6349
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.633720930232558e-08,
+      "loss": 0.3015,
+      "step": 6350
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6295681063122924e-08,
+      "loss": 0.2787,
+      "step": 6351
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.625415282392027e-08,
+      "loss": 0.2924,
+      "step": 6352
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.621262458471761e-08,
+      "loss": 0.2878,
+      "step": 6353
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.617109634551495e-08,
+      "loss": 0.2919,
+      "step": 6354
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.612956810631229e-08,
+      "loss": 0.2959,
+      "step": 6355
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6088039867109634e-08,
+      "loss": 0.2416,
+      "step": 6356
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.604651162790698e-08,
+      "loss": 0.2654,
+      "step": 6357
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.6004983388704315e-08,
+      "loss": 0.2326,
+      "step": 6358
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.596345514950166e-08,
+      "loss": 0.2688,
+      "step": 6359
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5921926910299e-08,
+      "loss": 0.2331,
+      "step": 6360
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5880398671096345e-08,
+      "loss": 0.2605,
+      "step": 6361
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.583887043189369e-08,
+      "loss": 0.316,
+      "step": 6362
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.579734219269103e-08,
+      "loss": 0.2632,
+      "step": 6363
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.575581395348837e-08,
+      "loss": 0.294,
+      "step": 6364
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.571428571428571e-08,
+      "loss": 0.2763,
+      "step": 6365
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5672757475083055e-08,
+      "loss": 0.3237,
+      "step": 6366
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.56312292358804e-08,
+      "loss": 0.2435,
+      "step": 6367
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.558970099667774e-08,
+      "loss": 0.2455,
+      "step": 6368
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.554817275747508e-08,
+      "loss": 0.2853,
+      "step": 6369
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.550664451827242e-08,
+      "loss": 0.2595,
+      "step": 6370
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5465116279069766e-08,
+      "loss": 0.3162,
+      "step": 6371
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.542358803986711e-08,
+      "loss": 0.2749,
+      "step": 6372
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.5382059800664446e-08,
+      "loss": 0.2722,
+      "step": 6373
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.534053156146179e-08,
+      "loss": 0.346,
+      "step": 6374
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 3.529900332225913e-08,
+      "loss": 0.3421,
+      "step": 6375
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5257475083056476e-08,
+      "loss": 0.271,
+      "step": 6376
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.521594684385382e-08,
+      "loss": 0.2693,
+      "step": 6377
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.517441860465116e-08,
+      "loss": 0.3364,
+      "step": 6378
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5132890365448506e-08,
+      "loss": 0.2824,
+      "step": 6379
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.509136212624584e-08,
+      "loss": 0.2439,
+      "step": 6380
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.5049833887043186e-08,
+      "loss": 0.2563,
+      "step": 6381
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.500830564784053e-08,
+      "loss": 0.3354,
+      "step": 6382
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.496677740863787e-08,
+      "loss": 0.2998,
+      "step": 6383
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.492524916943521e-08,
+      "loss": 0.2294,
+      "step": 6384
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4883720930232553e-08,
+      "loss": 0.3354,
+      "step": 6385
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.48421926910299e-08,
+      "loss": 0.3199,
+      "step": 6386
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.480066445182724e-08,
+      "loss": 0.3322,
+      "step": 6387
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4759136212624584e-08,
+      "loss": 0.2838,
+      "step": 6388
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.471760797342193e-08,
+      "loss": 0.3115,
+      "step": 6389
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.467607973421927e-08,
+      "loss": 0.3196,
+      "step": 6390
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4634551495016614e-08,
+      "loss": 0.2313,
+      "step": 6391
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.459302325581395e-08,
+      "loss": 0.2546,
+      "step": 6392
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4551495016611294e-08,
+      "loss": 0.2855,
+      "step": 6393
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.450996677740864e-08,
+      "loss": 0.2841,
+      "step": 6394
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4468438538205974e-08,
+      "loss": 0.3247,
+      "step": 6395
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.442691029900332e-08,
+      "loss": 0.2464,
+      "step": 6396
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.438538205980066e-08,
+      "loss": 0.2692,
+      "step": 6397
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4343853820598004e-08,
+      "loss": 0.2972,
+      "step": 6398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.430232558139535e-08,
+      "loss": 0.2615,
+      "step": 6399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.426079734219269e-08,
+      "loss": 0.2885,
+      "step": 6400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4219269102990035e-08,
+      "loss": 0.3008,
+      "step": 6401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.417774086378738e-08,
+      "loss": 0.3214,
+      "step": 6402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.4136212624584715e-08,
+      "loss": 0.3322,
+      "step": 6403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.409468438538206e-08,
+      "loss": 0.3101,
+      "step": 6404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.40531561461794e-08,
+      "loss": 0.2838,
+      "step": 6405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.401162790697674e-08,
+      "loss": 0.2949,
+      "step": 6406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.397009966777408e-08,
+      "loss": 0.3058,
+      "step": 6407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3928571428571425e-08,
+      "loss": 0.289,
+      "step": 6408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.388704318936877e-08,
+      "loss": 0.3561,
+      "step": 6409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.384551495016611e-08,
+      "loss": 0.2781,
+      "step": 6410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 3.3803986710963455e-08,
+      "loss": 0.2766,
+      "step": 6411
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.37624584717608e-08,
+      "loss": 0.2991,
+      "step": 6412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.372093023255814e-08,
+      "loss": 0.2967,
+      "step": 6413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.367940199335548e-08,
+      "loss": 0.2941,
+      "step": 6414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.363787375415282e-08,
+      "loss": 0.3607,
+      "step": 6415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3596345514950166e-08,
+      "loss": 0.2417,
+      "step": 6416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.355481727574751e-08,
+      "loss": 0.3719,
+      "step": 6417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3513289036544846e-08,
+      "loss": 0.2692,
+      "step": 6418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.347176079734219e-08,
+      "loss": 0.3309,
+      "step": 6419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.343023255813953e-08,
+      "loss": 0.2866,
+      "step": 6420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3388704318936876e-08,
+      "loss": 0.2348,
+      "step": 6421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.334717607973422e-08,
+      "loss": 0.2085,
+      "step": 6422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.330564784053156e-08,
+      "loss": 0.3461,
+      "step": 6423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3264119601328906e-08,
+      "loss": 0.2249,
+      "step": 6424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.322259136212624e-08,
+      "loss": 0.2345,
+      "step": 6425
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3181063122923587e-08,
+      "loss": 0.312,
+      "step": 6426
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.313953488372093e-08,
+      "loss": 0.2989,
+      "step": 6427
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3098006644518273e-08,
+      "loss": 0.3001,
+      "step": 6428
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.305647840531561e-08,
+      "loss": 0.2933,
+      "step": 6429
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.3014950166112954e-08,
+      "loss": 0.3083,
+      "step": 6430
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.29734219269103e-08,
+      "loss": 0.2364,
+      "step": 6431
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.293189368770764e-08,
+      "loss": 0.2799,
+      "step": 6432
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2890365448504984e-08,
+      "loss": 0.3368,
+      "step": 6433
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.284883720930233e-08,
+      "loss": 0.261,
+      "step": 6434
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.280730897009967e-08,
+      "loss": 0.2889,
+      "step": 6435
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.276578073089701e-08,
+      "loss": 0.2556,
+      "step": 6436
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.272425249169435e-08,
+      "loss": 0.3041,
+      "step": 6437
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2682724252491694e-08,
+      "loss": 0.4015,
+      "step": 6438
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.264119601328904e-08,
+      "loss": 0.2729,
+      "step": 6439
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2599667774086374e-08,
+      "loss": 0.2545,
+      "step": 6440
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.255813953488372e-08,
+      "loss": 0.2585,
+      "step": 6441
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.251661129568106e-08,
+      "loss": 0.2365,
+      "step": 6442
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.2475083056478405e-08,
+      "loss": 0.3172,
+      "step": 6443
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.243355481727575e-08,
+      "loss": 0.301,
+      "step": 6444
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.239202657807309e-08,
+      "loss": 0.3547,
+      "step": 6445
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.235049833887043e-08,
+      "loss": 0.2925,
+      "step": 6446
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 3.230897009966777e-08,
+      "loss": 0.2877,
+      "step": 6447
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2267441860465115e-08,
+      "loss": 0.2655,
+      "step": 6448
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.222591362126246e-08,
+      "loss": 0.2791,
+      "step": 6449
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.21843853820598e-08,
+      "loss": 0.2552,
+      "step": 6450
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.214285714285714e-08,
+      "loss": 0.297,
+      "step": 6451
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.210132890365448e-08,
+      "loss": 0.2998,
+      "step": 6452
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.2059800664451825e-08,
+      "loss": 0.2852,
+      "step": 6453
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.201827242524917e-08,
+      "loss": 0.2609,
+      "step": 6454
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1976744186046506e-08,
+      "loss": 0.2847,
+      "step": 6455
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.193521594684385e-08,
+      "loss": 0.2628,
+      "step": 6456
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.189368770764119e-08,
+      "loss": 0.3238,
+      "step": 6457
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1852159468438536e-08,
+      "loss": 0.2497,
+      "step": 6458
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.181063122923588e-08,
+      "loss": 0.2512,
+      "step": 6459
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.176910299003322e-08,
+      "loss": 0.3571,
+      "step": 6460
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1727574750830566e-08,
+      "loss": 0.306,
+      "step": 6461
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.16860465116279e-08,
+      "loss": 0.2464,
+      "step": 6462
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1644518272425246e-08,
+      "loss": 0.2998,
+      "step": 6463
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.160299003322259e-08,
+      "loss": 0.2594,
+      "step": 6464
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.156146179401993e-08,
+      "loss": 0.2646,
+      "step": 6465
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.151993355481727e-08,
+      "loss": 0.29,
+      "step": 6466
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.147840531561461e-08,
+      "loss": 0.3219,
+      "step": 6467
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1436877076411957e-08,
+      "loss": 0.2713,
+      "step": 6468
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.13953488372093e-08,
+      "loss": 0.2982,
+      "step": 6469
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1353820598006643e-08,
+      "loss": 0.2703,
+      "step": 6470
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.131229235880399e-08,
+      "loss": 0.294,
+      "step": 6471
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.127076411960133e-08,
+      "loss": 0.3496,
+      "step": 6472
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1229235880398674e-08,
+      "loss": 0.2794,
+      "step": 6473
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.118770764119601e-08,
+      "loss": 0.2604,
+      "step": 6474
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1146179401993354e-08,
+      "loss": 0.2309,
+      "step": 6475
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.11046511627907e-08,
+      "loss": 0.3343,
+      "step": 6476
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.1063122923588034e-08,
+      "loss": 0.3547,
+      "step": 6477
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.102159468438538e-08,
+      "loss": 0.2268,
+      "step": 6478
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.098006644518272e-08,
+      "loss": 0.3553,
+      "step": 6479
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0938538205980064e-08,
+      "loss": 0.2987,
+      "step": 6480
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.089700996677741e-08,
+      "loss": 0.3082,
+      "step": 6481
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.085548172757475e-08,
+      "loss": 0.2598,
+      "step": 6482
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 3.0813953488372094e-08,
+      "loss": 0.2585,
+      "step": 6483
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.077242524916944e-08,
+      "loss": 0.3101,
+      "step": 6484
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0730897009966775e-08,
+      "loss": 0.2959,
+      "step": 6485
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.068936877076412e-08,
+      "loss": 0.2542,
+      "step": 6486
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.064784053156146e-08,
+      "loss": 0.2803,
+      "step": 6487
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0606312292358805e-08,
+      "loss": 0.3639,
+      "step": 6488
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.056478405315614e-08,
+      "loss": 0.2571,
+      "step": 6489
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0523255813953485e-08,
+      "loss": 0.321,
+      "step": 6490
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.048172757475083e-08,
+      "loss": 0.3124,
+      "step": 6491
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.044019933554817e-08,
+      "loss": 0.2473,
+      "step": 6492
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0398671096345515e-08,
+      "loss": 0.2913,
+      "step": 6493
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.035714285714286e-08,
+      "loss": 0.2856,
+      "step": 6494
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.03156146179402e-08,
+      "loss": 0.2698,
+      "step": 6495
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.027408637873754e-08,
+      "loss": 0.3371,
+      "step": 6496
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.023255813953488e-08,
+      "loss": 0.2896,
+      "step": 6497
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0191029900332226e-08,
+      "loss": 0.3573,
+      "step": 6498
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.014950166112957e-08,
+      "loss": 0.2837,
+      "step": 6499
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.0107973421926906e-08,
+      "loss": 0.3642,
+      "step": 6500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.006644518272425e-08,
+      "loss": 0.3261,
+      "step": 6501
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.002491694352159e-08,
+      "loss": 0.2043,
+      "step": 6502
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9983388704318936e-08,
+      "loss": 0.2837,
+      "step": 6503
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.994186046511628e-08,
+      "loss": 0.3247,
+      "step": 6504
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.990033222591362e-08,
+      "loss": 0.3326,
+      "step": 6505
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9858803986710966e-08,
+      "loss": 0.2834,
+      "step": 6506
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.98172757475083e-08,
+      "loss": 0.3135,
+      "step": 6507
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9775747508305646e-08,
+      "loss": 0.2324,
+      "step": 6508
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.973421926910299e-08,
+      "loss": 0.2479,
+      "step": 6509
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9692691029900333e-08,
+      "loss": 0.2759,
+      "step": 6510
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.965116279069767e-08,
+      "loss": 0.2973,
+      "step": 6511
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9609634551495013e-08,
+      "loss": 0.2475,
+      "step": 6512
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9568106312292357e-08,
+      "loss": 0.2852,
+      "step": 6513
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.95265780730897e-08,
+      "loss": 0.2762,
+      "step": 6514
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.948504983388704e-08,
+      "loss": 0.2747,
+      "step": 6515
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9443521594684384e-08,
+      "loss": 0.3025,
+      "step": 6516
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9401993355481727e-08,
+      "loss": 0.3076,
+      "step": 6517
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.9360465116279067e-08,
+      "loss": 0.2667,
+      "step": 6518
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.931893687707641e-08,
+      "loss": 0.2663,
+      "step": 6519
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9277408637873754e-08,
+      "loss": 0.2848,
+      "step": 6520
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9235880398671097e-08,
+      "loss": 0.2718,
+      "step": 6521
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9194352159468434e-08,
+      "loss": 0.2757,
+      "step": 6522
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9152823920265778e-08,
+      "loss": 0.2642,
+      "step": 6523
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.911129568106312e-08,
+      "loss": 0.294,
+      "step": 6524
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9069767441860464e-08,
+      "loss": 0.2924,
+      "step": 6525
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.9028239202657805e-08,
+      "loss": 0.3072,
+      "step": 6526
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8986710963455148e-08,
+      "loss": 0.2931,
+      "step": 6527
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.894518272425249e-08,
+      "loss": 0.2493,
+      "step": 6528
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8903654485049835e-08,
+      "loss": 0.3979,
+      "step": 6529
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8862126245847175e-08,
+      "loss": 0.3143,
+      "step": 6530
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8820598006644518e-08,
+      "loss": 0.3236,
+      "step": 6531
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.877906976744186e-08,
+      "loss": 0.3017,
+      "step": 6532
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.87375415282392e-08,
+      "loss": 0.2786,
+      "step": 6533
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8696013289036542e-08,
+      "loss": 0.2462,
+      "step": 6534
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8654485049833885e-08,
+      "loss": 0.2667,
+      "step": 6535
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.861295681063123e-08,
+      "loss": 0.2596,
+      "step": 6536
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.857142857142857e-08,
+      "loss": 0.3635,
+      "step": 6537
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8529900332225912e-08,
+      "loss": 0.2892,
+      "step": 6538
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8488372093023256e-08,
+      "loss": 0.2978,
+      "step": 6539
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.84468438538206e-08,
+      "loss": 0.3147,
+      "step": 6540
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.840531561461794e-08,
+      "loss": 0.2787,
+      "step": 6541
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8363787375415282e-08,
+      "loss": 0.2883,
+      "step": 6542
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8322259136212626e-08,
+      "loss": 0.3188,
+      "step": 6543
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8280730897009966e-08,
+      "loss": 0.3583,
+      "step": 6544
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8239202657807306e-08,
+      "loss": 0.3159,
+      "step": 6545
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.819767441860465e-08,
+      "loss": 0.2873,
+      "step": 6546
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8156146179401993e-08,
+      "loss": 0.2935,
+      "step": 6547
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8114617940199333e-08,
+      "loss": 0.318,
+      "step": 6548
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.8073089700996676e-08,
+      "loss": 0.3373,
+      "step": 6549
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.803156146179402e-08,
+      "loss": 0.2374,
+      "step": 6550
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7990033222591363e-08,
+      "loss": 0.3182,
+      "step": 6551
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.79485049833887e-08,
+      "loss": 0.2877,
+      "step": 6552
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7906976744186043e-08,
+      "loss": 0.3019,
+      "step": 6553
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.7865448504983387e-08,
+      "loss": 0.2396,
+      "step": 6554
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.782392026578073e-08,
+      "loss": 0.2546,
+      "step": 6555
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.778239202657807e-08,
+      "loss": 0.2907,
+      "step": 6556
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7740863787375414e-08,
+      "loss": 0.3423,
+      "step": 6557
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7699335548172757e-08,
+      "loss": 0.266,
+      "step": 6558
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7657807308970097e-08,
+      "loss": 0.2627,
+      "step": 6559
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.761627906976744e-08,
+      "loss": 0.2286,
+      "step": 6560
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7574750830564784e-08,
+      "loss": 0.3448,
+      "step": 6561
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7533222591362127e-08,
+      "loss": 0.2406,
+      "step": 6562
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7491694352159464e-08,
+      "loss": 0.296,
+      "step": 6563
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7450166112956808e-08,
+      "loss": 0.2434,
+      "step": 6564
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.740863787375415e-08,
+      "loss": 0.3272,
+      "step": 6565
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7367109634551494e-08,
+      "loss": 0.2595,
+      "step": 6566
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7325581395348834e-08,
+      "loss": 0.3237,
+      "step": 6567
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7284053156146178e-08,
+      "loss": 0.2532,
+      "step": 6568
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.724252491694352e-08,
+      "loss": 0.3119,
+      "step": 6569
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7200996677740865e-08,
+      "loss": 0.2559,
+      "step": 6570
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7159468438538205e-08,
+      "loss": 0.3105,
+      "step": 6571
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7117940199335548e-08,
+      "loss": 0.3281,
+      "step": 6572
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.707641196013289e-08,
+      "loss": 0.267,
+      "step": 6573
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.7034883720930228e-08,
+      "loss": 0.3506,
+      "step": 6574
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6993355481727572e-08,
+      "loss": 0.2915,
+      "step": 6575
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6951827242524915e-08,
+      "loss": 0.3038,
+      "step": 6576
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.691029900332226e-08,
+      "loss": 0.2573,
+      "step": 6577
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.68687707641196e-08,
+      "loss": 0.2863,
+      "step": 6578
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6827242524916942e-08,
+      "loss": 0.2978,
+      "step": 6579
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6785714285714285e-08,
+      "loss": 0.2643,
+      "step": 6580
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.674418604651163e-08,
+      "loss": 0.2781,
+      "step": 6581
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.670265780730897e-08,
+      "loss": 0.2886,
+      "step": 6582
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6661129568106312e-08,
+      "loss": 0.3162,
+      "step": 6583
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6619601328903656e-08,
+      "loss": 0.2654,
+      "step": 6584
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6578073089700996e-08,
+      "loss": 0.3304,
+      "step": 6585
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6536544850498336e-08,
+      "loss": 0.2832,
+      "step": 6586
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.649501661129568e-08,
+      "loss": 0.3159,
+      "step": 6587
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6453488372093023e-08,
+      "loss": 0.362,
+      "step": 6588
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6411960132890363e-08,
+      "loss": 0.2601,
+      "step": 6589
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.6370431893687706e-08,
+      "loss": 0.3123,
+      "step": 6590
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.632890365448505e-08,
+      "loss": 0.3558,
+      "step": 6591
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6287375415282393e-08,
+      "loss": 0.2599,
+      "step": 6592
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.624584717607973e-08,
+      "loss": 0.3379,
+      "step": 6593
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6204318936877073e-08,
+      "loss": 0.2515,
+      "step": 6594
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6162790697674417e-08,
+      "loss": 0.3138,
+      "step": 6595
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.612126245847176e-08,
+      "loss": 0.2782,
+      "step": 6596
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.60797342192691e-08,
+      "loss": 0.2697,
+      "step": 6597
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.6038205980066444e-08,
+      "loss": 0.2985,
+      "step": 6598
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5996677740863787e-08,
+      "loss": 0.3113,
+      "step": 6599
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5955149501661127e-08,
+      "loss": 0.2828,
+      "step": 6600
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.591362126245847e-08,
+      "loss": 0.3355,
+      "step": 6601
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5872093023255814e-08,
+      "loss": 0.2921,
+      "step": 6602
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5830564784053157e-08,
+      "loss": 0.2649,
+      "step": 6603
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5789036544850494e-08,
+      "loss": 0.2769,
+      "step": 6604
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5747508305647837e-08,
+      "loss": 0.3107,
+      "step": 6605
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.570598006644518e-08,
+      "loss": 0.2772,
+      "step": 6606
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5664451827242524e-08,
+      "loss": 0.2781,
+      "step": 6607
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5622923588039864e-08,
+      "loss": 0.3114,
+      "step": 6608
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5581395348837208e-08,
+      "loss": 0.2919,
+      "step": 6609
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.553986710963455e-08,
+      "loss": 0.2497,
+      "step": 6610
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5498338870431895e-08,
+      "loss": 0.2651,
+      "step": 6611
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5456810631229235e-08,
+      "loss": 0.3929,
+      "step": 6612
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5415282392026578e-08,
+      "loss": 0.2986,
+      "step": 6613
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.537375415282392e-08,
+      "loss": 0.3114,
+      "step": 6614
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5332225913621258e-08,
+      "loss": 0.295,
+      "step": 6615
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5290697674418602e-08,
+      "loss": 0.3045,
+      "step": 6616
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5249169435215945e-08,
+      "loss": 0.2808,
+      "step": 6617
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.520764119601329e-08,
+      "loss": 0.2802,
+      "step": 6618
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.516611295681063e-08,
+      "loss": 0.2913,
+      "step": 6619
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5124584717607972e-08,
+      "loss": 0.3047,
+      "step": 6620
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5083056478405315e-08,
+      "loss": 0.3068,
+      "step": 6621
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.504152823920266e-08,
+      "loss": 0.2769,
+      "step": 6622
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.5e-08,
+      "loss": 0.2535,
+      "step": 6623
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4958471760797342e-08,
+      "loss": 0.2501,
+      "step": 6624
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4916943521594686e-08,
+      "loss": 0.2309,
+      "step": 6625
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4875415282392026e-08,
+      "loss": 0.2913,
+      "step": 6626
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4833887043189366e-08,
+      "loss": 0.2993,
+      "step": 6627
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.479235880398671e-08,
+      "loss": 0.2508,
+      "step": 6628
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4750830564784053e-08,
+      "loss": 0.3681,
+      "step": 6629
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4709302325581393e-08,
+      "loss": 0.312,
+      "step": 6630
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4667774086378736e-08,
+      "loss": 0.3059,
+      "step": 6631
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.462624584717608e-08,
+      "loss": 0.3274,
+      "step": 6632
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4584717607973423e-08,
+      "loss": 0.2958,
+      "step": 6633
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.454318936877076e-08,
+      "loss": 0.2587,
+      "step": 6634
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4501661129568103e-08,
+      "loss": 0.3414,
+      "step": 6635
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4460132890365447e-08,
+      "loss": 0.3189,
+      "step": 6636
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.441860465116279e-08,
+      "loss": 0.2457,
+      "step": 6637
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.437707641196013e-08,
+      "loss": 0.2643,
+      "step": 6638
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4335548172757473e-08,
+      "loss": 0.3219,
+      "step": 6639
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4294019933554817e-08,
+      "loss": 0.2519,
+      "step": 6640
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.425249169435216e-08,
+      "loss": 0.2543,
+      "step": 6641
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.42109634551495e-08,
+      "loss": 0.3035,
+      "step": 6642
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4169435215946844e-08,
+      "loss": 0.277,
+      "step": 6643
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4127906976744187e-08,
+      "loss": 0.2954,
+      "step": 6644
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4086378737541524e-08,
+      "loss": 0.2425,
+      "step": 6645
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4044850498338867e-08,
+      "loss": 0.3153,
+      "step": 6646
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.400332225913621e-08,
+      "loss": 0.5174,
+      "step": 6647
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3961794019933554e-08,
+      "loss": 0.2705,
+      "step": 6648
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3920265780730894e-08,
+      "loss": 0.3048,
+      "step": 6649
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3878737541528238e-08,
+      "loss": 0.3011,
+      "step": 6650
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.383720930232558e-08,
+      "loss": 0.2764,
+      "step": 6651
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3795681063122924e-08,
+      "loss": 0.2424,
+      "step": 6652
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3754152823920265e-08,
+      "loss": 0.2638,
+      "step": 6653
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3712624584717608e-08,
+      "loss": 0.3108,
+      "step": 6654
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.367109634551495e-08,
+      "loss": 0.3665,
+      "step": 6655
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3629568106312288e-08,
+      "loss": 0.2717,
+      "step": 6656
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.358803986710963e-08,
+      "loss": 0.3049,
+      "step": 6657
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3546511627906975e-08,
+      "loss": 0.2965,
+      "step": 6658
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.350498338870432e-08,
+      "loss": 0.2441,
+      "step": 6659
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.346345514950166e-08,
+      "loss": 0.2554,
+      "step": 6660
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3421926910299002e-08,
+      "loss": 0.3164,
+      "step": 6661
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3380398671096345e-08,
+      "loss": 0.3121,
+      "step": 6662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.333887043189369e-08,
+      "loss": 0.2449,
+      "step": 6663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.329734219269103e-08,
+      "loss": 0.2538,
+      "step": 6664
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3255813953488372e-08,
+      "loss": 0.3099,
+      "step": 6665
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3214285714285716e-08,
+      "loss": 0.2985,
+      "step": 6666
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3172757475083056e-08,
+      "loss": 0.2999,
+      "step": 6667
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3131229235880396e-08,
+      "loss": 0.2728,
+      "step": 6668
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.308970099667774e-08,
+      "loss": 0.2391,
+      "step": 6669
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3048172757475083e-08,
+      "loss": 0.2781,
+      "step": 6670
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3006644518272423e-08,
+      "loss": 0.244,
+      "step": 6671
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2965116279069766e-08,
+      "loss": 0.3088,
+      "step": 6672
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.292358803986711e-08,
+      "loss": 0.3313,
+      "step": 6673
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2882059800664453e-08,
+      "loss": 0.3121,
+      "step": 6674
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.284053156146179e-08,
+      "loss": 0.2801,
+      "step": 6675
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2799003322259133e-08,
+      "loss": 0.2762,
+      "step": 6676
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2757475083056476e-08,
+      "loss": 0.3065,
+      "step": 6677
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.271594684385382e-08,
+      "loss": 0.3239,
+      "step": 6678
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.267441860465116e-08,
+      "loss": 0.3516,
+      "step": 6679
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2632890365448503e-08,
+      "loss": 0.2678,
+      "step": 6680
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2591362126245847e-08,
+      "loss": 0.3063,
+      "step": 6681
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.254983388704319e-08,
+      "loss": 0.2964,
+      "step": 6682
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.250830564784053e-08,
+      "loss": 0.3065,
+      "step": 6683
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2466777408637874e-08,
+      "loss": 0.2774,
+      "step": 6684
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2425249169435217e-08,
+      "loss": 0.3074,
+      "step": 6685
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2383720930232554e-08,
+      "loss": 0.296,
+      "step": 6686
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2342192691029897e-08,
+      "loss": 0.3366,
+      "step": 6687
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.230066445182724e-08,
+      "loss": 0.3309,
+      "step": 6688
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2259136212624584e-08,
+      "loss": 0.2304,
+      "step": 6689
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2217607973421924e-08,
+      "loss": 0.338,
+      "step": 6690
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2176079734219268e-08,
+      "loss": 0.2734,
+      "step": 6691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.213455149501661e-08,
+      "loss": 0.291,
+      "step": 6692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2093023255813954e-08,
+      "loss": 0.3044,
+      "step": 6693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2051495016611295e-08,
+      "loss": 0.3455,
+      "step": 6694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.2009966777408638e-08,
+      "loss": 0.268,
+      "step": 6695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.196843853820598e-08,
+      "loss": 0.2738,
+      "step": 6696
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1926910299003325e-08,
+      "loss": 0.2845,
+      "step": 6697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.188538205980066e-08,
+      "loss": 0.2305,
+      "step": 6698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1843853820598005e-08,
+      "loss": 0.2759,
+      "step": 6699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.1802325581395348e-08,
+      "loss": 0.2915,
+      "step": 6700
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.176079734219269e-08,
+      "loss": 0.3062,
+      "step": 6701
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1719269102990032e-08,
+      "loss": 0.2563,
+      "step": 6702
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1677740863787375e-08,
+      "loss": 0.296,
+      "step": 6703
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.163621262458472e-08,
+      "loss": 0.3056,
+      "step": 6704
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.159468438538206e-08,
+      "loss": 0.2558,
+      "step": 6705
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1553156146179402e-08,
+      "loss": 0.2743,
+      "step": 6706
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1511627906976746e-08,
+      "loss": 0.3047,
+      "step": 6707
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1470099667774086e-08,
+      "loss": 0.3318,
+      "step": 6708
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1428571428571426e-08,
+      "loss": 0.2779,
+      "step": 6709
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.138704318936877e-08,
+      "loss": 0.3134,
+      "step": 6710
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1345514950166113e-08,
+      "loss": 0.2148,
+      "step": 6711
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1303986710963453e-08,
+      "loss": 0.2527,
+      "step": 6712
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1262458471760796e-08,
+      "loss": 0.2907,
+      "step": 6713
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.122093023255814e-08,
+      "loss": 0.3321,
+      "step": 6714
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1179401993355483e-08,
+      "loss": 0.2955,
+      "step": 6715
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.113787375415282e-08,
+      "loss": 0.2542,
+      "step": 6716
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1096345514950163e-08,
+      "loss": 0.3189,
+      "step": 6717
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.1054817275747506e-08,
+      "loss": 0.2628,
+      "step": 6718
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.101328903654485e-08,
+      "loss": 0.3292,
+      "step": 6719
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.097176079734219e-08,
+      "loss": 0.2643,
+      "step": 6720
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0930232558139533e-08,
+      "loss": 0.2928,
+      "step": 6721
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0888704318936877e-08,
+      "loss": 0.3212,
+      "step": 6722
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.084717607973422e-08,
+      "loss": 0.2948,
+      "step": 6723
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.080564784053156e-08,
+      "loss": 0.3232,
+      "step": 6724
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0764119601328904e-08,
+      "loss": 0.2618,
+      "step": 6725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0722591362126247e-08,
+      "loss": 0.254,
+      "step": 6726
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0681063122923584e-08,
+      "loss": 0.2608,
+      "step": 6727
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0639534883720927e-08,
+      "loss": 0.282,
+      "step": 6728
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.059800664451827e-08,
+      "loss": 0.2491,
+      "step": 6729
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0556478405315614e-08,
+      "loss": 0.3017,
+      "step": 6730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0514950166112954e-08,
+      "loss": 0.2448,
+      "step": 6731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0473421926910298e-08,
+      "loss": 0.2391,
+      "step": 6732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.043189368770764e-08,
+      "loss": 0.2575,
+      "step": 6733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0390365448504984e-08,
+      "loss": 0.3222,
+      "step": 6734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0348837209302324e-08,
+      "loss": 0.3009,
+      "step": 6735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.0307308970099668e-08,
+      "loss": 0.2749,
+      "step": 6736
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.026578073089701e-08,
+      "loss": 0.2762,
+      "step": 6737
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0224252491694355e-08,
+      "loss": 0.3134,
+      "step": 6738
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.018272425249169e-08,
+      "loss": 0.2191,
+      "step": 6739
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0141196013289035e-08,
+      "loss": 0.3165,
+      "step": 6740
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0099667774086378e-08,
+      "loss": 0.2987,
+      "step": 6741
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0058139534883718e-08,
+      "loss": 0.3076,
+      "step": 6742
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.0016611295681062e-08,
+      "loss": 0.3398,
+      "step": 6743
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9975083056478405e-08,
+      "loss": 0.3593,
+      "step": 6744
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.993355481727575e-08,
+      "loss": 0.297,
+      "step": 6745
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.989202657807309e-08,
+      "loss": 0.2865,
+      "step": 6746
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9850498338870432e-08,
+      "loss": 0.3097,
+      "step": 6747
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9808970099667775e-08,
+      "loss": 0.2453,
+      "step": 6748
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9767441860465116e-08,
+      "loss": 0.3041,
+      "step": 6749
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9725913621262456e-08,
+      "loss": 0.2851,
+      "step": 6750
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.96843853820598e-08,
+      "loss": 0.3085,
+      "step": 6751
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9642857142857142e-08,
+      "loss": 0.2716,
+      "step": 6752
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9601328903654483e-08,
+      "loss": 0.313,
+      "step": 6753
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9559800664451826e-08,
+      "loss": 0.2469,
+      "step": 6754
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.951827242524917e-08,
+      "loss": 0.3566,
+      "step": 6755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9476744186046513e-08,
+      "loss": 0.3586,
+      "step": 6756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.943521594684385e-08,
+      "loss": 0.2694,
+      "step": 6757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9393687707641193e-08,
+      "loss": 0.3421,
+      "step": 6758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9352159468438536e-08,
+      "loss": 0.2355,
+      "step": 6759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.931063122923588e-08,
+      "loss": 0.2845,
+      "step": 6760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.926910299003322e-08,
+      "loss": 0.3149,
+      "step": 6761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9227574750830563e-08,
+      "loss": 0.2419,
+      "step": 6762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9186046511627907e-08,
+      "loss": 0.311,
+      "step": 6763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.914451827242525e-08,
+      "loss": 0.3559,
+      "step": 6764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.914451827242525e-08,
+      "loss": 0.2483,
+      "step": 6765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.910299003322259e-08,
+      "loss": 0.3263,
+      "step": 6766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9061461794019934e-08,
+      "loss": 0.2734,
+      "step": 6767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.9019933554817277e-08,
+      "loss": 0.222,
+      "step": 6768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8978405315614614e-08,
+      "loss": 0.3164,
+      "step": 6769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8936877076411957e-08,
+      "loss": 0.2254,
+      "step": 6770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.88953488372093e-08,
+      "loss": 0.3025,
+      "step": 6771
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8853820598006644e-08,
+      "loss": 0.27,
+      "step": 6772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8812292358803984e-08,
+      "loss": 0.2643,
+      "step": 6773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8770764119601327e-08,
+      "loss": 0.303,
+      "step": 6774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.872923588039867e-08,
+      "loss": 0.2858,
+      "step": 6775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.868770764119601e-08,
+      "loss": 0.2087,
+      "step": 6776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8646179401993354e-08,
+      "loss": 0.2496,
+      "step": 6777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8604651162790698e-08,
+      "loss": 0.2034,
+      "step": 6778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.856312292358804e-08,
+      "loss": 0.3893,
+      "step": 6779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.852159468438538e-08,
+      "loss": 0.2582,
+      "step": 6780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8480066445182725e-08,
+      "loss": 0.2797,
+      "step": 6781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8438538205980065e-08,
+      "loss": 0.3073,
+      "step": 6782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8397009966777408e-08,
+      "loss": 0.2752,
+      "step": 6783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.835548172757475e-08,
+      "loss": 0.2871,
+      "step": 6784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8313953488372092e-08,
+      "loss": 0.261,
+      "step": 6785
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8272425249169435e-08,
+      "loss": 0.2426,
+      "step": 6786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8230897009966775e-08,
+      "loss": 0.2786,
+      "step": 6787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.818936877076412e-08,
+      "loss": 0.2795,
+      "step": 6788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8147840531561462e-08,
+      "loss": 0.4143,
+      "step": 6789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8106312292358805e-08,
+      "loss": 0.2605,
+      "step": 6790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8064784053156145e-08,
+      "loss": 0.3016,
+      "step": 6791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.802325581395349e-08,
+      "loss": 0.2854,
+      "step": 6792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.798172757475083e-08,
+      "loss": 0.2326,
+      "step": 6793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7940199335548172e-08,
+      "loss": 0.2792,
+      "step": 6794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7898671096345516e-08,
+      "loss": 0.3023,
+      "step": 6795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7857142857142856e-08,
+      "loss": 0.2643,
+      "step": 6796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.78156146179402e-08,
+      "loss": 0.278,
+      "step": 6797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.777408637873754e-08,
+      "loss": 0.2302,
+      "step": 6798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7732558139534883e-08,
+      "loss": 0.3271,
+      "step": 6799
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7691029900332223e-08,
+      "loss": 0.2968,
+      "step": 6800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7649501661129566e-08,
+      "loss": 0.2812,
+      "step": 6801
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.760797342192691e-08,
+      "loss": 0.3195,
+      "step": 6802
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7566445182724253e-08,
+      "loss": 0.2539,
+      "step": 6803
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7524916943521593e-08,
+      "loss": 0.3111,
+      "step": 6804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7483388704318937e-08,
+      "loss": 0.3192,
+      "step": 6805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7441860465116277e-08,
+      "loss": 0.2938,
+      "step": 6806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.740033222591362e-08,
+      "loss": 0.2301,
+      "step": 6807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.7358803986710963e-08,
+      "loss": 0.2661,
+      "step": 6808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7317275747508307e-08,
+      "loss": 0.2482,
+      "step": 6809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7275747508305647e-08,
+      "loss": 0.3163,
+      "step": 6810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7234219269102987e-08,
+      "loss": 0.293,
+      "step": 6811
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.719269102990033e-08,
+      "loss": 0.3217,
+      "step": 6812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7151162790697674e-08,
+      "loss": 0.3226,
+      "step": 6813
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7109634551495017e-08,
+      "loss": 0.3402,
+      "step": 6814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.7068106312292357e-08,
+      "loss": 0.2689,
+      "step": 6815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.70265780730897e-08,
+      "loss": 0.2979,
+      "step": 6816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.698504983388704e-08,
+      "loss": 0.3236,
+      "step": 6817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6943521594684384e-08,
+      "loss": 0.2782,
+      "step": 6818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6901993355481728e-08,
+      "loss": 0.2788,
+      "step": 6819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.686046511627907e-08,
+      "loss": 0.2932,
+      "step": 6820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.681893687707641e-08,
+      "loss": 0.3009,
+      "step": 6821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6777408637873755e-08,
+      "loss": 0.316,
+      "step": 6822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6735880398671095e-08,
+      "loss": 0.311,
+      "step": 6823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6694352159468438e-08,
+      "loss": 0.2976,
+      "step": 6824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.665282392026578e-08,
+      "loss": 0.2811,
+      "step": 6825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.661129568106312e-08,
+      "loss": 0.2967,
+      "step": 6826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6569767441860465e-08,
+      "loss": 0.2994,
+      "step": 6827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6528239202657805e-08,
+      "loss": 0.2811,
+      "step": 6828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.648671096345515e-08,
+      "loss": 0.2581,
+      "step": 6829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6445182724252492e-08,
+      "loss": 0.2319,
+      "step": 6830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6403654485049835e-08,
+      "loss": 0.2392,
+      "step": 6831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6362126245847175e-08,
+      "loss": 0.199,
+      "step": 6832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.632059800664452e-08,
+      "loss": 0.3642,
+      "step": 6833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.627906976744186e-08,
+      "loss": 0.3396,
+      "step": 6834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6237541528239202e-08,
+      "loss": 0.3157,
+      "step": 6835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6196013289036546e-08,
+      "loss": 0.2757,
+      "step": 6836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6154485049833886e-08,
+      "loss": 0.2809,
+      "step": 6837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.611295681063123e-08,
+      "loss": 0.3373,
+      "step": 6838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.607142857142857e-08,
+      "loss": 0.2944,
+      "step": 6839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.6029900332225913e-08,
+      "loss": 0.285,
+      "step": 6840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5988372093023253e-08,
+      "loss": 0.2992,
+      "step": 6841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5946843853820596e-08,
+      "loss": 0.3042,
+      "step": 6842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.590531561461794e-08,
+      "loss": 0.2753,
+      "step": 6843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.5863787375415283e-08,
+      "loss": 0.2695,
+      "step": 6844
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5822259136212623e-08,
+      "loss": 0.2955,
+      "step": 6845
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5780730897009966e-08,
+      "loss": 0.277,
+      "step": 6846
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5739202657807307e-08,
+      "loss": 0.308,
+      "step": 6847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.569767441860465e-08,
+      "loss": 0.2741,
+      "step": 6848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5656146179401993e-08,
+      "loss": 0.2767,
+      "step": 6849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5614617940199337e-08,
+      "loss": 0.2784,
+      "step": 6850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5573089700996677e-08,
+      "loss": 0.2359,
+      "step": 6851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5531561461794017e-08,
+      "loss": 0.2858,
+      "step": 6852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.549003322259136e-08,
+      "loss": 0.2772,
+      "step": 6853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5448504983388704e-08,
+      "loss": 0.3577,
+      "step": 6854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5406976744186047e-08,
+      "loss": 0.2784,
+      "step": 6855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5365448504983387e-08,
+      "loss": 0.2342,
+      "step": 6856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.532392026578073e-08,
+      "loss": 0.3062,
+      "step": 6857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.528239202657807e-08,
+      "loss": 0.3088,
+      "step": 6858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5240863787375414e-08,
+      "loss": 0.2709,
+      "step": 6859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5199335548172758e-08,
+      "loss": 0.3108,
+      "step": 6860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.51578073089701e-08,
+      "loss": 0.279,
+      "step": 6861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.511627906976744e-08,
+      "loss": 0.2657,
+      "step": 6862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5074750830564785e-08,
+      "loss": 0.2629,
+      "step": 6863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.5033222591362125e-08,
+      "loss": 0.2707,
+      "step": 6864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4991694352159468e-08,
+      "loss": 0.333,
+      "step": 6865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.495016611295681e-08,
+      "loss": 0.312,
+      "step": 6866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.490863787375415e-08,
+      "loss": 0.2946,
+      "step": 6867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4867109634551495e-08,
+      "loss": 0.2562,
+      "step": 6868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4825581395348835e-08,
+      "loss": 0.2971,
+      "step": 6869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4784053156146178e-08,
+      "loss": 0.3242,
+      "step": 6870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.474252491694352e-08,
+      "loss": 0.2715,
+      "step": 6871
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4700996677740864e-08,
+      "loss": 0.2935,
+      "step": 6872
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4659468438538205e-08,
+      "loss": 0.3,
+      "step": 6873
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4617940199335549e-08,
+      "loss": 0.2534,
+      "step": 6874
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4576411960132889e-08,
+      "loss": 0.3049,
+      "step": 6875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4534883720930232e-08,
+      "loss": 0.2739,
+      "step": 6876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4493355481727574e-08,
+      "loss": 0.2788,
+      "step": 6877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4451827242524917e-08,
+      "loss": 0.3166,
+      "step": 6878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4410299003322259e-08,
+      "loss": 0.2456,
+      "step": 6879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.43687707641196e-08,
+      "loss": 0.2784,
+      "step": 6880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4327242524916943e-08,
+      "loss": 0.2759,
+      "step": 6881
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4285714285714284e-08,
+      "loss": 0.2447,
+      "step": 6882
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4244186046511628e-08,
+      "loss": 0.2571,
+      "step": 6883
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.420265780730897e-08,
+      "loss": 0.2822,
+      "step": 6884
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4161129568106313e-08,
+      "loss": 0.2656,
+      "step": 6885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4119601328903653e-08,
+      "loss": 0.2681,
+      "step": 6886
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4078073089700996e-08,
+      "loss": 0.365,
+      "step": 6887
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4036544850498338e-08,
+      "loss": 0.2881,
+      "step": 6888
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3995016611295682e-08,
+      "loss": 0.2948,
+      "step": 6889
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3953488372093022e-08,
+      "loss": 0.2628,
+      "step": 6890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3911960132890365e-08,
+      "loss": 0.2674,
+      "step": 6891
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3870431893687707e-08,
+      "loss": 0.2954,
+      "step": 6892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3828903654485049e-08,
+      "loss": 0.2993,
+      "step": 6893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3787375415282392e-08,
+      "loss": 0.243,
+      "step": 6894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3745847176079732e-08,
+      "loss": 0.2835,
+      "step": 6895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3704318936877075e-08,
+      "loss": 0.2597,
+      "step": 6896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3662790697674417e-08,
+      "loss": 0.3068,
+      "step": 6897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.362126245847176e-08,
+      "loss": 0.2995,
+      "step": 6898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3579734219269102e-08,
+      "loss": 0.3335,
+      "step": 6899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3538205980066446e-08,
+      "loss": 0.364,
+      "step": 6900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3496677740863786e-08,
+      "loss": 0.2796,
+      "step": 6901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.345514950166113e-08,
+      "loss": 0.348,
+      "step": 6902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3413621262458471e-08,
+      "loss": 0.2952,
+      "step": 6903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3372093023255814e-08,
+      "loss": 0.302,
+      "step": 6904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3330564784053156e-08,
+      "loss": 0.2915,
+      "step": 6905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3289036544850498e-08,
+      "loss": 0.2811,
+      "step": 6906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.324750830564784e-08,
+      "loss": 0.2554,
+      "step": 6907
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3205980066445181e-08,
+      "loss": 0.2916,
+      "step": 6908
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3164451827242525e-08,
+      "loss": 0.2966,
+      "step": 6909
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3122923588039865e-08,
+      "loss": 0.3015,
+      "step": 6910
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.3081395348837208e-08,
+      "loss": 0.3844,
+      "step": 6911
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.303986710963455e-08,
+      "loss": 0.3153,
+      "step": 6912
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2998338870431893e-08,
+      "loss": 0.3263,
+      "step": 6913
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2956810631229235e-08,
+      "loss": 0.2544,
+      "step": 6914
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2915282392026579e-08,
+      "loss": 0.2982,
+      "step": 6915
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2873754152823919e-08,
+      "loss": 0.3034,
+      "step": 6916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2832225913621262e-08,
+      "loss": 0.2587,
+      "step": 6917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2790697674418604e-08,
+      "loss": 0.302,
+      "step": 6918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2749169435215947e-08,
+      "loss": 0.229,
+      "step": 6919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2707641196013289e-08,
+      "loss": 0.2665,
+      "step": 6920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2666112956810629e-08,
+      "loss": 0.3124,
+      "step": 6921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2624584717607973e-08,
+      "loss": 0.2733,
+      "step": 6922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2583056478405314e-08,
+      "loss": 0.3222,
+      "step": 6923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2541528239202658e-08,
+      "loss": 0.3029,
+      "step": 6924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.25e-08,
+      "loss": 0.2125,
+      "step": 6925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2458471760797343e-08,
+      "loss": 0.3011,
+      "step": 6926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2416943521594683e-08,
+      "loss": 0.2936,
+      "step": 6927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2375415282392026e-08,
+      "loss": 0.3159,
+      "step": 6928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2333887043189368e-08,
+      "loss": 0.2872,
+      "step": 6929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2292358803986711e-08,
+      "loss": 0.2851,
+      "step": 6930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2250830564784052e-08,
+      "loss": 0.2421,
+      "step": 6931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2209302325581395e-08,
+      "loss": 0.2597,
+      "step": 6932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2167774086378737e-08,
+      "loss": 0.2922,
+      "step": 6933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.212624584717608e-08,
+      "loss": 0.3291,
+      "step": 6934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2084717607973422e-08,
+      "loss": 0.3298,
+      "step": 6935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2043189368770762e-08,
+      "loss": 0.3197,
+      "step": 6936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2001661129568105e-08,
+      "loss": 0.283,
+      "step": 6937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1960132890365447e-08,
+      "loss": 0.3495,
+      "step": 6938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.191860465116279e-08,
+      "loss": 0.2489,
+      "step": 6939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1877076411960132e-08,
+      "loss": 0.3009,
+      "step": 6940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1835548172757476e-08,
+      "loss": 0.2853,
+      "step": 6941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1794019933554816e-08,
+      "loss": 0.239,
+      "step": 6942
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.175249169435216e-08,
+      "loss": 0.3254,
+      "step": 6943
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1710963455149501e-08,
+      "loss": 0.2397,
+      "step": 6944
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1669435215946844e-08,
+      "loss": 0.2701,
+      "step": 6945
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1627906976744186e-08,
+      "loss": 0.2628,
+      "step": 6946
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1586378737541528e-08,
+      "loss": 0.2908,
+      "step": 6947
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.154485049833887e-08,
+      "loss": 0.2803,
+      "step": 6948
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1503322259136211e-08,
+      "loss": 0.2584,
+      "step": 6949
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1461794019933555e-08,
+      "loss": 0.3003,
+      "step": 6950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1420265780730895e-08,
+      "loss": 0.2907,
+      "step": 6951
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1378737541528238e-08,
+      "loss": 0.3309,
+      "step": 6952
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.133720930232558e-08,
+      "loss": 0.2622,
+      "step": 6953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1295681063122923e-08,
+      "loss": 0.2978,
+      "step": 6954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1254152823920265e-08,
+      "loss": 0.2684,
+      "step": 6955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1212624584717609e-08,
+      "loss": 0.3082,
+      "step": 6956
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1171096345514949e-08,
+      "loss": 0.3557,
+      "step": 6957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1129568106312292e-08,
+      "loss": 0.2771,
+      "step": 6958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1088039867109634e-08,
+      "loss": 0.2701,
+      "step": 6959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1046511627906977e-08,
+      "loss": 0.3275,
+      "step": 6960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1004983388704319e-08,
+      "loss": 0.3443,
+      "step": 6961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0963455149501662e-08,
+      "loss": 0.3199,
+      "step": 6962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0921926910299002e-08,
+      "loss": 0.2748,
+      "step": 6963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0880398671096344e-08,
+      "loss": 0.312,
+      "step": 6964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0838870431893688e-08,
+      "loss": 0.3364,
+      "step": 6965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.079734219269103e-08,
+      "loss": 0.3147,
+      "step": 6966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0755813953488373e-08,
+      "loss": 0.3226,
+      "step": 6967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0714285714285713e-08,
+      "loss": 0.2762,
+      "step": 6968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0672757475083056e-08,
+      "loss": 0.2843,
+      "step": 6969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0631229235880398e-08,
+      "loss": 0.295,
+      "step": 6970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0589700996677741e-08,
+      "loss": 0.2415,
+      "step": 6971
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0548172757475082e-08,
+      "loss": 0.3093,
+      "step": 6972
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0506644518272425e-08,
+      "loss": 0.3038,
+      "step": 6973
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0465116279069767e-08,
+      "loss": 0.2561,
+      "step": 6974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.042358803986711e-08,
+      "loss": 0.243,
+      "step": 6975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0382059800664452e-08,
+      "loss": 0.2218,
+      "step": 6976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0340531561461792e-08,
+      "loss": 0.2739,
+      "step": 6977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0299003322259135e-08,
+      "loss": 0.3127,
+      "step": 6978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0257475083056477e-08,
+      "loss": 0.2852,
+      "step": 6979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.021594684385382e-08,
+      "loss": 0.2941,
+      "step": 6980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0174418604651162e-08,
+      "loss": 0.3539,
+      "step": 6981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0132890365448506e-08,
+      "loss": 0.2979,
+      "step": 6982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0091362126245846e-08,
+      "loss": 0.3188,
+      "step": 6983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0049833887043189e-08,
+      "loss": 0.2777,
+      "step": 6984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.0008305647840531e-08,
+      "loss": 0.2476,
+      "step": 6985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.966777408637874e-09,
+      "loss": 0.3037,
+      "step": 6986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.925249169435216e-09,
+      "loss": 0.2887,
+      "step": 6987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.883720930232558e-09,
+      "loss": 0.3146,
+      "step": 6988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.8421926910299e-09,
+      "loss": 0.2767,
+      "step": 6989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.800664451827241e-09,
+      "loss": 0.2913,
+      "step": 6990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.759136212624585e-09,
+      "loss": 0.2877,
+      "step": 6991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.717607973421925e-09,
+      "loss": 0.2395,
+      "step": 6992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.676079734219268e-09,
+      "loss": 0.3518,
+      "step": 6993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.63455149501661e-09,
+      "loss": 0.2724,
+      "step": 6994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.593023255813953e-09,
+      "loss": 0.3566,
+      "step": 6995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.551495016611295e-09,
+      "loss": 0.2885,
+      "step": 6996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.509966777408638e-09,
+      "loss": 0.3381,
+      "step": 6997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.468438538205979e-09,
+      "loss": 0.3347,
+      "step": 6998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.426910299003322e-09,
+      "loss": 0.3128,
+      "step": 6999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.385382059800664e-09,
+      "loss": 0.2859,
+      "step": 7000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.343853820598005e-09,
+      "loss": 0.3052,
+      "step": 7001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.302325581395349e-09,
+      "loss": 0.2858,
+      "step": 7002
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.26079734219269e-09,
+      "loss": 0.2811,
+      "step": 7003
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.219269102990032e-09,
+      "loss": 0.306,
+      "step": 7004
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.177740863787376e-09,
+      "loss": 0.2445,
+      "step": 7005
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.136212624584718e-09,
+      "loss": 0.2268,
+      "step": 7006
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.09468438538206e-09,
+      "loss": 0.2698,
+      "step": 7007
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.053156146179403e-09,
+      "loss": 0.2504,
+      "step": 7008
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.011627906976744e-09,
+      "loss": 0.2686,
+      "step": 7009
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.970099667774086e-09,
+      "loss": 0.2905,
+      "step": 7010
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.928571428571428e-09,
+      "loss": 0.2722,
+      "step": 7011
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.88704318936877e-09,
+      "loss": 0.2724,
+      "step": 7012
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.845514950166111e-09,
+      "loss": 0.2594,
+      "step": 7013
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.803986710963455e-09,
+      "loss": 0.2828,
+      "step": 7014
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.762458471760797e-09,
+      "loss": 0.3228,
+      "step": 7015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.720930232558138e-09,
+      "loss": 0.2949,
+      "step": 7016
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.679401993355482e-09,
+      "loss": 0.2006,
+      "step": 7017
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.637873754152823e-09,
+      "loss": 0.3096,
+      "step": 7018
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.596345514950165e-09,
+      "loss": 0.303,
+      "step": 7019
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.554817275747509e-09,
+      "loss": 0.2592,
+      "step": 7020
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.51328903654485e-09,
+      "loss": 0.2557,
+      "step": 7021
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.471760797342192e-09,
+      "loss": 0.3173,
+      "step": 7022
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.430232558139536e-09,
+      "loss": 0.2948,
+      "step": 7023
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.388704318936877e-09,
+      "loss": 0.2324,
+      "step": 7024
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 8.347176079734219e-09,
+      "loss": 0.2432,
+      "step": 7025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.30564784053156e-09,
+      "loss": 0.2647,
+      "step": 7026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.264119601328903e-09,
+      "loss": 0.2884,
+      "step": 7027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.222591362126246e-09,
+      "loss": 0.2519,
+      "step": 7028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.181063122923588e-09,
+      "loss": 0.2829,
+      "step": 7029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.13953488372093e-09,
+      "loss": 0.3263,
+      "step": 7030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.098006644518273e-09,
+      "loss": 0.3182,
+      "step": 7031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.056478405315615e-09,
+      "loss": 0.3113,
+      "step": 7032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 8.014950166112956e-09,
+      "loss": 0.3209,
+      "step": 7033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.973421926910298e-09,
+      "loss": 0.2452,
+      "step": 7034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.931893687707641e-09,
+      "loss": 0.2899,
+      "step": 7035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.890365448504983e-09,
+      "loss": 0.2647,
+      "step": 7036
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.848837209302325e-09,
+      "loss": 0.2431,
+      "step": 7037
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.807308970099668e-09,
+      "loss": 0.2949,
+      "step": 7038
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.765780730897008e-09,
+      "loss": 0.2833,
+      "step": 7039
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.724252491694352e-09,
+      "loss": 0.2271,
+      "step": 7040
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.682724252491694e-09,
+      "loss": 0.2976,
+      "step": 7041
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.641196013289035e-09,
+      "loss": 0.3682,
+      "step": 7042
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.599667774086379e-09,
+      "loss": 0.3348,
+      "step": 7043
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.55813953488372e-09,
+      "loss": 0.314,
+      "step": 7044
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.516611295681062e-09,
+      "loss": 0.3127,
+      "step": 7045
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.475083056478406e-09,
+      "loss": 0.2884,
+      "step": 7046
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.4335548172757475e-09,
+      "loss": 0.2974,
+      "step": 7047
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.392026578073089e-09,
+      "loss": 0.287,
+      "step": 7048
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.350498338870432e-09,
+      "loss": 0.2759,
+      "step": 7049
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.308970099667774e-09,
+      "loss": 0.2279,
+      "step": 7050
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.267441860465116e-09,
+      "loss": 0.2785,
+      "step": 7051
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.225913621262459e-09,
+      "loss": 0.3633,
+      "step": 7052
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.1843853820598e-09,
+      "loss": 0.3599,
+      "step": 7053
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.142857142857142e-09,
+      "loss": 0.2432,
+      "step": 7054
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.101328903654485e-09,
+      "loss": 0.3037,
+      "step": 7055
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.0598006644518265e-09,
+      "loss": 0.2813,
+      "step": 7056
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 7.018272425249169e-09,
+      "loss": 0.2536,
+      "step": 7057
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.976744186046511e-09,
+      "loss": 0.3066,
+      "step": 7058
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.935215946843853e-09,
+      "loss": 0.3047,
+      "step": 7059
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.893687707641196e-09,
+      "loss": 0.3581,
+      "step": 7060
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.852159468438538e-09,
+      "loss": 0.3454,
+      "step": 7061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.81063122923588e-09,
+      "loss": 0.2875,
+      "step": 7062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.769102990033223e-09,
+      "loss": 0.2775,
+      "step": 7063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.727574750830565e-09,
+      "loss": 0.2686,
+      "step": 7064
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.686046511627907e-09,
+      "loss": 0.2943,
+      "step": 7065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.644518272425249e-09,
+      "loss": 0.3273,
+      "step": 7066
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.602990033222591e-09,
+      "loss": 0.273,
+      "step": 7067
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.5614617940199325e-09,
+      "loss": 0.2311,
+      "step": 7068
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.519933554817275e-09,
+      "loss": 0.2812,
+      "step": 7069
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.478405315614618e-09,
+      "loss": 0.2442,
+      "step": 7070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.436877076411959e-09,
+      "loss": 0.3146,
+      "step": 7071
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.395348837209302e-09,
+      "loss": 0.3051,
+      "step": 7072
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.3538205980066445e-09,
+      "loss": 0.2917,
+      "step": 7073
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.312292358803986e-09,
+      "loss": 0.2366,
+      "step": 7074
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.270764119601329e-09,
+      "loss": 0.2794,
+      "step": 7075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.2292358803986714e-09,
+      "loss": 0.2336,
+      "step": 7076
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.187707641196013e-09,
+      "loss": 0.3064,
+      "step": 7077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.146179401993356e-09,
+      "loss": 0.3139,
+      "step": 7078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.1046511627906975e-09,
+      "loss": 0.2886,
+      "step": 7079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.06312292358804e-09,
+      "loss": 0.3443,
+      "step": 7080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.021594684385381e-09,
+      "loss": 0.2824,
+      "step": 7081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.9800664451827236e-09,
+      "loss": 0.2901,
+      "step": 7082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.938538205980066e-09,
+      "loss": 0.2266,
+      "step": 7083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.897009966777408e-09,
+      "loss": 0.2483,
+      "step": 7084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.8554817275747505e-09,
+      "loss": 0.2662,
+      "step": 7085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.813953488372093e-09,
+      "loss": 0.3099,
+      "step": 7086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.772425249169435e-09,
+      "loss": 0.2929,
+      "step": 7087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.730897009966777e-09,
+      "loss": 0.3106,
+      "step": 7088
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689368770764119e-09,
+      "loss": 0.2906,
+      "step": 7089
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647840531561462e-09,
+      "loss": 0.282,
+      "step": 7090
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.606312292358804e-09,
+      "loss": 0.3169,
+      "step": 7091
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.564784053156146e-09,
+      "loss": 0.3155,
+      "step": 7092
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.523255813953489e-09,
+      "loss": 0.2935,
+      "step": 7093
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.481727574750831e-09,
+      "loss": 0.2957,
+      "step": 7094
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.440199335548172e-09,
+      "loss": 0.3392,
+      "step": 7095
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.398671096345515e-09,
+      "loss": 0.2444,
+      "step": 7096
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.3571428571428564e-09,
+      "loss": 0.2574,
+      "step": 7097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.315614617940199e-09,
+      "loss": 0.3096,
+      "step": 7098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.274086378737541e-09,
+      "loss": 0.2208,
+      "step": 7099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.232558139534883e-09,
+      "loss": 0.2267,
+      "step": 7100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.191029900332226e-09,
+      "loss": 0.2671,
+      "step": 7101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.149501661129568e-09,
+      "loss": 0.2861,
+      "step": 7102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.10797342192691e-09,
+      "loss": 0.2415,
+      "step": 7103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.066445182724253e-09,
+      "loss": 0.285,
+      "step": 7104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.0249169435215946e-09,
+      "loss": 0.2848,
+      "step": 7105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.983388704318937e-09,
+      "loss": 0.3146,
+      "step": 7106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.941860465116279e-09,
+      "loss": 0.2477,
+      "step": 7107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.900332225913621e-09,
+      "loss": 0.2739,
+      "step": 7108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.858803986710962e-09,
+      "loss": 0.233,
+      "step": 7109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.817275747508305e-09,
+      "loss": 0.2731,
+      "step": 7110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.7757475083056475e-09,
+      "loss": 0.2911,
+      "step": 7111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.734219269102989e-09,
+      "loss": 0.2719,
+      "step": 7112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.692691029900332e-09,
+      "loss": 0.348,
+      "step": 7113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.6511627906976744e-09,
+      "loss": 0.3692,
+      "step": 7114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.609634551495016e-09,
+      "loss": 0.2587,
+      "step": 7115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.568106312292359e-09,
+      "loss": 0.2438,
+      "step": 7116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.526578073089701e-09,
+      "loss": 0.2231,
+      "step": 7117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.485049833887043e-09,
+      "loss": 0.2734,
+      "step": 7118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.443521594684385e-09,
+      "loss": 0.3445,
+      "step": 7119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.4019933554817274e-09,
+      "loss": 0.2421,
+      "step": 7120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.360465116279069e-09,
+      "loss": 0.2635,
+      "step": 7121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.318936877076412e-09,
+      "loss": 0.3151,
+      "step": 7122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.277408637873754e-09,
+      "loss": 0.3268,
+      "step": 7123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.235880398671096e-09,
+      "loss": 0.2863,
+      "step": 7124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.194352159468439e-09,
+      "loss": 0.3146,
+      "step": 7125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.15282392026578e-09,
+      "loss": 0.3335,
+      "step": 7126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.111295681063123e-09,
+      "loss": 0.3068,
+      "step": 7127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.069767441860465e-09,
+      "loss": 0.2975,
+      "step": 7128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.028239202657807e-09,
+      "loss": 0.3067,
+      "step": 7129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.986710963455149e-09,
+      "loss": 0.319,
+      "step": 7130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.945182724252492e-09,
+      "loss": 0.2875,
+      "step": 7131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.903654485049834e-09,
+      "loss": 0.304,
+      "step": 7132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.862126245847176e-09,
+      "loss": 0.2877,
+      "step": 7133
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.820598006644518e-09,
+      "loss": 0.3634,
+      "step": 7134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.77906976744186e-09,
+      "loss": 0.295,
+      "step": 7135
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.737541528239203e-09,
+      "loss": 0.3018,
+      "step": 7136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6960132890365446e-09,
+      "loss": 0.2413,
+      "step": 7137
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.654485049833887e-09,
+      "loss": 0.2778,
+      "step": 7138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.6129568106312293e-09,
+      "loss": 0.3352,
+      "step": 7139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.571428571428571e-09,
+      "loss": 0.2628,
+      "step": 7140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.5299003322259133e-09,
+      "loss": 0.2553,
+      "step": 7141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4883720930232554e-09,
+      "loss": 0.2231,
+      "step": 7142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.446843853820598e-09,
+      "loss": 0.2863,
+      "step": 7143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.40531561461794e-09,
+      "loss": 0.2521,
+      "step": 7144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3637873754152823e-09,
+      "loss": 0.2965,
+      "step": 7145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3222591362126245e-09,
+      "loss": 0.3123,
+      "step": 7146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.2807308970099662e-09,
+      "loss": 0.2701,
+      "step": 7147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.239202657807309e-09,
+      "loss": 0.2837,
+      "step": 7148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.197674418604651e-09,
+      "loss": 0.2787,
+      "step": 7149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.156146179401993e-09,
+      "loss": 0.2791,
+      "step": 7150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.1146179401993357e-09,
+      "loss": 0.2621,
+      "step": 7151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.073089700996678e-09,
+      "loss": 0.353,
+      "step": 7152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.03156146179402e-09,
+      "loss": 0.3349,
+      "step": 7153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9900332225913618e-09,
+      "loss": 0.2396,
+      "step": 7154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.948504983388704e-09,
+      "loss": 0.2895,
+      "step": 7155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.9069767441860465e-09,
+      "loss": 0.2284,
+      "step": 7156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.8654485049833887e-09,
+      "loss": 0.318,
+      "step": 7157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.823920265780731e-09,
+      "loss": 0.2727,
+      "step": 7158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.782392026578073e-09,
+      "loss": 0.2673,
+      "step": 7159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.7408637873754156e-09,
+      "loss": 0.2787,
+      "step": 7160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6993355481727573e-09,
+      "loss": 0.3058,
+      "step": 7161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6578073089700995e-09,
+      "loss": 0.3162,
+      "step": 7162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.6162790697674417e-09,
+      "loss": 0.2916,
+      "step": 7163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.574750830564784e-09,
+      "loss": 0.2941,
+      "step": 7164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.5332225913621264e-09,
+      "loss": 0.2855,
+      "step": 7165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4916943521594686e-09,
+      "loss": 0.2779,
+      "step": 7166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4501661129568103e-09,
+      "loss": 0.297,
+      "step": 7167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.4086378737541525e-09,
+      "loss": 0.2689,
+      "step": 7168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3671096345514946e-09,
+      "loss": 0.2785,
+      "step": 7169
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3255813953488372e-09,
+      "loss": 0.2776,
+      "step": 7170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2840531561461794e-09,
+      "loss": 0.275,
+      "step": 7171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2425249169435215e-09,
+      "loss": 0.3033,
+      "step": 7172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.2009966777408637e-09,
+      "loss": 0.247,
+      "step": 7173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.159468438538206e-09,
+      "loss": 0.2329,
+      "step": 7174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.117940199335548e-09,
+      "loss": 0.3541,
+      "step": 7175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.07641196013289e-09,
+      "loss": 0.2571,
+      "step": 7176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.0348837209302324e-09,
+      "loss": 0.2641,
+      "step": 7177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.9933554817275745e-09,
+      "loss": 0.313,
+      "step": 7178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.951827242524917e-09,
+      "loss": 0.3678,
+      "step": 7179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.910299003322259e-09,
+      "loss": 0.2965,
+      "step": 7180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8687707641196014e-09,
+      "loss": 0.2996,
+      "step": 7181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8272425249169436e-09,
+      "loss": 0.2695,
+      "step": 7182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7857142857142855e-09,
+      "loss": 0.4055,
+      "step": 7183
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.7441860465116277e-09,
+      "loss": 0.2818,
+      "step": 7184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.70265780730897e-09,
+      "loss": 0.3084,
+      "step": 7185
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6611295681063122e-09,
+      "loss": 0.3661,
+      "step": 7186
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6196013289036544e-09,
+      "loss": 0.3291,
+      "step": 7187
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.5780730897009966e-09,
+      "loss": 0.3386,
+      "step": 7188
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.536544850498339e-09,
+      "loss": 0.2521,
+      "step": 7189
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4950166112956809e-09,
+      "loss": 0.3109,
+      "step": 7190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4534883720930233e-09,
+      "loss": 0.2582,
+      "step": 7191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.4119601328903654e-09,
+      "loss": 0.3588,
+      "step": 7192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3704318936877078e-09,
+      "loss": 0.2454,
+      "step": 7193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3289036544850498e-09,
+      "loss": 0.2762,
+      "step": 7194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.287375415282392e-09,
+      "loss": 0.2693,
+      "step": 7195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2458471760797343e-09,
+      "loss": 0.3258,
+      "step": 7196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2043189368770762e-09,
+      "loss": 0.3613,
+      "step": 7197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1627906976744186e-09,
+      "loss": 0.2381,
+      "step": 7198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.1212624584717608e-09,
+      "loss": 0.2776,
+      "step": 7199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.079734219269103e-09,
+      "loss": 0.3166,
+      "step": 7200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.038205980066445e-09,
+      "loss": 0.2813,
+      "step": 7201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.966777408637873e-10,
+      "loss": 0.2582,
+      "step": 7202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.551495016611294e-10,
+      "loss": 0.2782,
+      "step": 7203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.136212624584718e-10,
+      "loss": 0.3252,
+      "step": 7204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 8.720930232558139e-10,
+      "loss": 0.3356,
+      "step": 7205
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.305647840531561e-10,
+      "loss": 0.3312,
+      "step": 7206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.890365448504983e-10,
+      "loss": 0.2446,
+      "step": 7207
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.475083056478404e-10,
+      "loss": 0.2146,
+      "step": 7208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.059800664451827e-10,
+      "loss": 0.317,
+      "step": 7209
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.644518272425249e-10,
+      "loss": 0.3405,
+      "step": 7210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.229235880398671e-10,
+      "loss": 0.2873,
+      "step": 7211
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.813953488372093e-10,
+      "loss": 0.2779,
+      "step": 7212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.398671096345515e-10,
+      "loss": 0.2883,
+      "step": 7213
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.983388704318936e-10,
+      "loss": 0.2662,
+      "step": 7214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.568106312292359e-10,
+      "loss": 0.2478,
+      "step": 7215
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.1528239202657806e-10,
+      "loss": 0.2971,
+      "step": 7216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.737541528239202e-10,
+      "loss": 0.3216,
+      "step": 7217
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3222591362126244e-10,
+      "loss": 0.3156,
+      "step": 7218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.9069767441860465e-10,
+      "loss": 0.3208,
+      "step": 7219
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.491694352159468e-10,
+      "loss": 0.2909,
+      "step": 7220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.0764119601328903e-10,
+      "loss": 0.2781,
+      "step": 7221
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6611295681063122e-10,
+      "loss": 0.2405,
+      "step": 7222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.245847176079734e-10,
+      "loss": 0.3091,
+      "step": 7223
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 8.305647840531561e-11,
+      "loss": 0.2672,
+      "step": 7224
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.47963303327560425,
+      "eval_runtime": 3146.7479,
+      "eval_samples_per_second": 5.066,
+      "eval_steps_per_second": 0.079,
+      "eval_wer": 0.392462442022833,
+      "step": 7224
+    },
+    {
+      "epoch": 2.0,
+      "step": 7224,
+      "total_flos": 1.3873486530426556e+20,
+      "train_loss": 0.2911061938534411,
+      "train_runtime": 204479.9012,
+      "train_samples_per_second": 4.522,
+      "train_steps_per_second": 0.035
+    }
+  ],
+  "max_steps": 7224,
+  "num_train_epochs": 2,
+  "total_flos": 1.3873486530426556e+20,
+  "trial_name": null,
+  "trial_params": null
+}