diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13488 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2243,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882356e-07,
+      "loss": 0.8753,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.882352941176471e-07,
+      "loss": 0.8944,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.823529411764707e-07,
+      "loss": 0.9009,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1764705882352942e-06,
+      "loss": 0.9096,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4705882352941177e-06,
+      "loss": 0.8881,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7647058823529414e-06,
+      "loss": 0.8936,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.058823529411765e-06,
+      "loss": 0.8414,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.8584,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.647058823529412e-06,
+      "loss": 0.7688,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9411764705882355e-06,
+      "loss": 0.7206,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2352941176470594e-06,
+      "loss": 0.7156,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.618,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8235294117647055e-06,
+      "loss": 0.627,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 0.5852,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.411764705882353e-06,
+      "loss": 0.5701,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.5131,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.4972,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.4866,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.588235294117647e-06,
+      "loss": 0.4938,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 0.4772,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.176470588235295e-06,
+      "loss": 0.4596,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 0.4497,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.764705882352942e-06,
+      "loss": 0.4333,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.429,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.352941176470589e-06,
+      "loss": 0.4347,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 0.4308,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.941176470588236e-06,
+      "loss": 0.4212,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 0.4244,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.529411764705883e-06,
+      "loss": 0.4095,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 0.3913,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.11764705882353e-06,
+      "loss": 0.3941,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.3876,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.705882352941177e-06,
+      "loss": 0.3884,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.3781,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0294117647058823e-05,
+      "loss": 0.3765,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0588235294117648e-05,
+      "loss": 0.3748,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0882352941176471e-05,
+      "loss": 0.3692,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1176470588235295e-05,
+      "loss": 0.3659,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1470588235294118e-05,
+      "loss": 0.3517,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.3543,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2058823529411765e-05,
+      "loss": 0.353,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.235294117647059e-05,
+      "loss": 0.3466,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2647058823529412e-05,
+      "loss": 0.337,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2941176470588238e-05,
+      "loss": 0.3453,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.323529411764706e-05,
+      "loss": 0.3251,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3529411764705885e-05,
+      "loss": 0.333,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3823529411764706e-05,
+      "loss": 0.3245,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.323,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4411764705882353e-05,
+      "loss": 0.3351,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4705882352941179e-05,
+      "loss": 0.3175,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.3166,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5294117647058822e-05,
+      "loss": 0.3249,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.558823529411765e-05,
+      "loss": 0.3199,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5882352941176473e-05,
+      "loss": 0.3152,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6176470588235296e-05,
+      "loss": 0.3146,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.3167,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6764705882352943e-05,
+      "loss": 0.3142,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7058823529411767e-05,
+      "loss": 0.3031,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.735294117647059e-05,
+      "loss": 0.3002,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7647058823529414e-05,
+      "loss": 0.3126,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7941176470588237e-05,
+      "loss": 0.3085,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.823529411764706e-05,
+      "loss": 0.3163,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8529411764705884e-05,
+      "loss": 0.3152,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.3109,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.911764705882353e-05,
+      "loss": 0.3059,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9411764705882355e-05,
+      "loss": 0.3064,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9705882352941178e-05,
+      "loss": 0.313,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.3083,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999989568394573e-05,
+      "loss": 0.2984,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999958273600055e-05,
+      "loss": 0.3064,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999906115681737e-05,
+      "loss": 0.3065,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999833094748436e-05,
+      "loss": 0.2965,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999739210952493e-05,
+      "loss": 0.3056,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999962446448979e-05,
+      "loss": 0.2941,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999488855599718e-05,
+      "loss": 0.3083,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999332384565204e-05,
+      "loss": 0.3034,
+      "step": 76
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999915505171269e-05,
+      "loss": 0.2947,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999895685741216e-05,
+      "loss": 0.308,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99987378020771e-05,
+      "loss": 0.2962,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999849788616454e-05,
+      "loss": 0.3046,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998237110175018e-05,
+      "loss": 0.3027,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999795547465259e-05,
+      "loss": 0.304,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999765298018484e-05,
+      "loss": 0.3059,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999732962740288e-05,
+      "loss": 0.3035,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996985416981314e-05,
+      "loss": 0.2976,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996620349638286e-05,
+      "loss": 0.2906,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996234426135435e-05,
+      "loss": 0.2992,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995827647277926e-05,
+      "loss": 0.2896,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999540001391443e-05,
+      "loss": 0.2956,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994951526937126e-05,
+      "loss": 0.2887,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994482187281702e-05,
+      "loss": 0.2992,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999399199592735e-05,
+      "loss": 0.3025,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999348095389677e-05,
+      "loss": 0.3025,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999294906225616e-05,
+      "loss": 0.3033,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992396322115213e-05,
+      "loss": 0.2936,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991822734627126e-05,
+      "loss": 0.2968,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.3026,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999061302243977e-05,
+      "loss": 0.2997,
+      "step": 98
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989976900264348e-05,
+      "loss": 0.298,
+      "step": 99
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989319935789475e-05,
+      "loss": 0.2889,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998864213038579e-05,
+      "loss": 0.2897,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987943485467413e-05,
+      "loss": 0.2972,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998722400249194e-05,
+      "loss": 0.2836,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986483682960445e-05,
+      "loss": 0.286,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985722528417477e-05,
+      "loss": 0.2948,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984940540451037e-05,
+      "loss": 0.2914,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984137720692615e-05,
+      "loss": 0.2933,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983314070817142e-05,
+      "loss": 0.2932,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982469592543022e-05,
+      "loss": 0.2961,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981604287632104e-05,
+      "loss": 0.2974,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980718157889695e-05,
+      "loss": 0.2831,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979811205164544e-05,
+      "loss": 0.2814,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978883431348845e-05,
+      "loss": 0.2933,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977934838378236e-05,
+      "loss": 0.2912,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997696542823178e-05,
+      "loss": 0.2912,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975975202931982e-05,
+      "loss": 0.2985,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974964164544775e-05,
+      "loss": 0.2946,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973932315179502e-05,
+      "loss": 0.2848,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997287965698893e-05,
+      "loss": 0.2911,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971806192169252e-05,
+      "loss": 0.2895,
+      "step": 120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997071192296005e-05,
+      "loss": 0.2844,
+      "step": 121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996959685164433e-05,
+      "loss": 0.2904,
+      "step": 122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996846098054848e-05,
+      "loss": 0.2817,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967304312042298e-05,
+      "loss": 0.291,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996612684853896e-05,
+      "loss": 0.2843,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.2872,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996370954641049e-05,
+      "loss": 0.2988,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962469712828613e-05,
+      "loss": 0.2896,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961209094336115e-05,
+      "loss": 0.2774,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959927693563044e-05,
+      "loss": 0.2964,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958625513182816e-05,
+      "loss": 0.2863,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99573025559122e-05,
+      "loss": 0.2892,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955958824511302e-05,
+      "loss": 0.3015,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954594321783585e-05,
+      "loss": 0.3044,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953209050575838e-05,
+      "loss": 0.2828,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995180301377818e-05,
+      "loss": 0.2922,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995037621432405e-05,
+      "loss": 0.288,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994892865519022e-05,
+      "loss": 0.2919,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947460339396756e-05,
+      "loss": 0.2906,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994597127000704e-05,
+      "loss": 0.2996,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994446145012775e-05,
+      "loss": 0.2781,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942930882908852e-05,
+      "loss": 0.2838,
+      "step": 142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941379571543597e-05,
+      "loss": 0.2916,
+      "step": 143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939807519268525e-05,
+      "loss": 0.2838,
+      "step": 144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938214729363442e-05,
+      "loss": 0.2909,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936601205151415e-05,
+      "loss": 0.2823,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934966949998772e-05,
+      "loss": 0.272,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99333119673151e-05,
+      "loss": 0.2825,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931636260553226e-05,
+      "loss": 0.2828,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99299398332092e-05,
+      "loss": 0.287,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928222688822324e-05,
+      "loss": 0.2857,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926484830975116e-05,
+      "loss": 0.2928,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924726263293297e-05,
+      "loss": 0.2813,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922946989445804e-05,
+      "loss": 0.2844,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.2864,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991932633814555e-05,
+      "loss": 0.2851,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917484968246625e-05,
+      "loss": 0.2865,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915622907289695e-05,
+      "loss": 0.2851,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991374015915962e-05,
+      "loss": 0.2853,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911836727784413e-05,
+      "loss": 0.2855,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909912617135246e-05,
+      "loss": 0.2927,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907967831226428e-05,
+      "loss": 0.2831,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990600237411541e-05,
+      "loss": 0.2878,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904016249902765e-05,
+      "loss": 0.2807,
+      "step": 164
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990200946273219e-05,
+      "loss": 0.283,
+      "step": 165
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989998201679048e-05,
+      "loss": 0.2715,
+      "step": 166
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897933916307542e-05,
+      "loss": 0.2798,
+      "step": 167
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895865165556375e-05,
+      "loss": 0.287,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989377576885305e-05,
+      "loss": 0.2823,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891665730556727e-05,
+      "loss": 0.2744,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988953505506962e-05,
+      "loss": 0.2747,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887383746836995e-05,
+      "loss": 0.283,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885211810347185e-05,
+      "loss": 0.2826,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883019250131538e-05,
+      "loss": 0.2769,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988080607076444e-05,
+      "loss": 0.2817,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878572276863296e-05,
+      "loss": 0.2851,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876317873088512e-05,
+      "loss": 0.2796,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987404286414351e-05,
+      "loss": 0.2912,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871747254774673e-05,
+      "loss": 0.2785,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869431049771393e-05,
+      "loss": 0.2877,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986709425396601e-05,
+      "loss": 0.2815,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986473687223383e-05,
+      "loss": 0.2719,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862358909493113e-05,
+      "loss": 0.285,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.2833,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857541260873764e-05,
+      "loss": 0.2883,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855101585046292e-05,
+      "loss": 0.2783,
+      "step": 186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852641348312588e-05,
+      "loss": 0.2845,
+      "step": 187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850160555805485e-05,
+      "loss": 0.2806,
+      "step": 188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984765921270072e-05,
+      "loss": 0.2831,
+      "step": 189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98451373242169e-05,
+      "loss": 0.2952,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842594895615487e-05,
+      "loss": 0.2787,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984003193220081e-05,
+      "loss": 0.286,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837448439320027e-05,
+      "loss": 0.2847,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834844422363142e-05,
+      "loss": 0.2852,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832219886762965e-05,
+      "loss": 0.2798,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982957483799512e-05,
+      "loss": 0.2819,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826909281578027e-05,
+      "loss": 0.2818,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824223223072895e-05,
+      "loss": 0.2835,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821516668083705e-05,
+      "loss": 0.2834,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818789622257197e-05,
+      "loss": 0.2692,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816042091282867e-05,
+      "loss": 0.2832,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981327408089295e-05,
+      "loss": 0.2736,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981048559686239e-05,
+      "loss": 0.2738,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980767664500888e-05,
+      "loss": 0.2874,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980484723119278e-05,
+      "loss": 0.2689,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801997361317163e-05,
+      "loss": 0.283,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799127041327773e-05,
+      "loss": 0.2679,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796236277213013e-05,
+      "loss": 0.2718,
+      "step": 208
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793325075003952e-05,
+      "loss": 0.2718,
+      "step": 209
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979039344077429e-05,
+      "loss": 0.2861,
+      "step": 210
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787441380640355e-05,
+      "loss": 0.273,
+      "step": 211
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784468900761097e-05,
+      "loss": 0.2776,
+      "step": 212
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.282,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977846270661538e-05,
+      "loss": 0.2781,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977542900487977e-05,
+      "loss": 0.2867,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772374908460508e-05,
+      "loss": 0.2802,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769300423729422e-05,
+      "loss": 0.282,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 0.2821,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763090315031736e-05,
+      "loss": 0.284,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975995470402142e-05,
+      "loss": 0.282,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756798730611814e-05,
+      "loss": 0.2773,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975362240138728e-05,
+      "loss": 0.2795,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975042572297468e-05,
+      "loss": 0.2701,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747208702043297e-05,
+      "loss": 0.2801,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743971345304877e-05,
+      "loss": 0.2737,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974071365951358e-05,
+      "loss": 0.2803,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737435651465987e-05,
+      "loss": 0.2773,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734137328001077e-05,
+      "loss": 0.2671,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730818696000207e-05,
+      "loss": 0.2754,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727479762387115e-05,
+      "loss": 0.2762,
+      "step": 230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724120534127885e-05,
+      "loss": 0.2728,
+      "step": 231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972074101823095e-05,
+      "loss": 0.2929,
+      "step": 232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717341221747056e-05,
+      "loss": 0.27,
+      "step": 233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713921151769278e-05,
+      "loss": 0.2784,
+      "step": 234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710480815432983e-05,
+      "loss": 0.2728,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707020219915805e-05,
+      "loss": 0.2827,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970353937243767e-05,
+      "loss": 0.2803,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700038280260734e-05,
+      "loss": 0.2744,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696516950689404e-05,
+      "loss": 0.2714,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969297539107031e-05,
+      "loss": 0.2786,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968941360879227e-05,
+      "loss": 0.2737,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.2688,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682229406025635e-05,
+      "loss": 0.2753,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967860700052559e-05,
+      "loss": 0.2721,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674964402343684e-05,
+      "loss": 0.2731,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671301619079547e-05,
+      "loss": 0.2684,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667618658374912e-05,
+      "loss": 0.2772,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663915527913628e-05,
+      "loss": 0.2844,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660192235421607e-05,
+      "loss": 0.2821,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656448788666833e-05,
+      "loss": 0.2816,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652685195459346e-05,
+      "loss": 0.2673,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96489014636512e-05,
+      "loss": 0.279,
+      "step": 252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645097601136475e-05,
+      "loss": 0.2686,
+      "step": 253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641273615851256e-05,
+      "loss": 0.28,
+      "step": 254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637429515773603e-05,
+      "loss": 0.281,
+      "step": 255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963356530892354e-05,
+      "loss": 0.2843,
+      "step": 256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629681003363043e-05,
+      "loss": 0.2648,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9625776607196024e-05,
+      "loss": 0.2825,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621852128568303e-05,
+      "loss": 0.2783,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617907575667602e-05,
+      "loss": 0.2744,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961394295672353e-05,
+      "loss": 0.2752,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609958280007556e-05,
+      "loss": 0.2766,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605953553832987e-05,
+      "loss": 0.2614,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601928786554977e-05,
+      "loss": 0.2829,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597883986570473e-05,
+      "loss": 0.2777,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593819162318232e-05,
+      "loss": 0.2688,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589734322278784e-05,
+      "loss": 0.2673,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585629474974413e-05,
+      "loss": 0.2781,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581504628969153e-05,
+      "loss": 0.2814,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577359792868752e-05,
+      "loss": 0.2778,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.2782,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569010185014062e-05,
+      "loss": 0.2686,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564805430679733e-05,
+      "loss": 0.2716,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560580721090155e-05,
+      "loss": 0.2775,
+      "step": 274
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955633606505943e-05,
+      "loss": 0.2809,
+      "step": 275
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955207147144327e-05,
+      "loss": 0.2689,
+      "step": 276
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547786949138992e-05,
+      "loss": 0.2724,
+      "step": 277
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543482507085484e-05,
+      "loss": 0.2678,
+      "step": 278
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539158154263188e-05,
+      "loss": 0.2658,
+      "step": 279
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534813899694097e-05,
+      "loss": 0.2667,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953044975244172e-05,
+      "loss": 0.2666,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526065721611075e-05,
+      "loss": 0.2716,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521661816348647e-05,
+      "loss": 0.2714,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517238045842403e-05,
+      "loss": 0.2649,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951279441932175e-05,
+      "loss": 0.2719,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950833094605752e-05,
+      "loss": 0.2675,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503847635361944e-05,
+      "loss": 0.2756,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499344496588656e-05,
+      "loss": 0.2738,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949482153913265e-05,
+      "loss": 0.2762,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490278772430255e-05,
+      "loss": 0.2844,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485716205959154e-05,
+      "loss": 0.2764,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481133849238322e-05,
+      "loss": 0.2647,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476531711828027e-05,
+      "loss": 0.275,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94719098033298e-05,
+      "loss": 0.2703,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467268133386434e-05,
+      "loss": 0.2695,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462606711681936e-05,
+      "loss": 0.2766,
+      "step": 296
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457925547941535e-05,
+      "loss": 0.2726,
+      "step": 297
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453224651931638e-05,
+      "loss": 0.267,
+      "step": 298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448504033459817e-05,
+      "loss": 0.2754,
+      "step": 299
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944376370237481e-05,
+      "loss": 0.2733,
+      "step": 300
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439003668566466e-05,
+      "loss": 0.2712,
+      "step": 301
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434223941965738e-05,
+      "loss": 0.2778,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429424532544677e-05,
+      "loss": 0.266,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424605450316387e-05,
+      "loss": 0.2866,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419766705335025e-05,
+      "loss": 0.2691,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414908307695765e-05,
+      "loss": 0.2842,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410030267534785e-05,
+      "loss": 0.2693,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940513259502924e-05,
+      "loss": 0.2681,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400215300397255e-05,
+      "loss": 0.2745,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395278393897876e-05,
+      "loss": 0.2702,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390321885831082e-05,
+      "loss": 0.2684,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385345786537734e-05,
+      "loss": 0.2612,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380350106399578e-05,
+      "loss": 0.2751,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937533485583921e-05,
+      "loss": 0.2763,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370300045320044e-05,
+      "loss": 0.2697,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365245685346316e-05,
+      "loss": 0.268,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360171786463043e-05,
+      "loss": 0.2672,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935507835925601e-05,
+      "loss": 0.2719,
+      "step": 318
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934996541435174e-05,
+      "loss": 0.2798,
+      "step": 319
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344832962417475e-05,
+      "loss": 0.2817,
+      "step": 320
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339681014161163e-05,
+      "loss": 0.2755,
+      "step": 321
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334509580331417e-05,
+      "loss": 0.2717,
+      "step": 322
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932931867171751e-05,
+      "loss": 0.2721,
+      "step": 323
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324108299149345e-05,
+      "loss": 0.2717,
+      "step": 324
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318878473497437e-05,
+      "loss": 0.2731,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931362920567287e-05,
+      "loss": 0.2796,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930836050662731e-05,
+      "loss": 0.2638,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303072387352952e-05,
+      "loss": 0.2821,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.2786,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292437932289205e-05,
+      "loss": 0.2709,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92870916186867e-05,
+      "loss": 0.2727,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281725929229127e-05,
+      "loss": 0.2735,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927634087511104e-05,
+      "loss": 0.2728,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270936467567388e-05,
+      "loss": 0.2737,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92655127178735e-05,
+      "loss": 0.2711,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926006963734506e-05,
+      "loss": 0.2624,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254607237338082e-05,
+      "loss": 0.2708,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924912552924889e-05,
+      "loss": 0.2689,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243624524514082e-05,
+      "loss": 0.266,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923810423461052e-05,
+      "loss": 0.2655,
+      "step": 340
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232564671055308e-05,
+      "loss": 0.2696,
+      "step": 341
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227005845405747e-05,
+      "loss": 0.2694,
+      "step": 342
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221427769259333e-05,
+      "loss": 0.2643,
+      "step": 343
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9215830454253726e-05,
+      "loss": 0.2739,
+      "step": 344
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921021391206672e-05,
+      "loss": 0.2666,
+      "step": 345
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204578154416233e-05,
+      "loss": 0.2827,
+      "step": 346
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198923193060255e-05,
+      "loss": 0.2792,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9193249039796856e-05,
+      "loss": 0.2685,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918755570646414e-05,
+      "loss": 0.275,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181843204940232e-05,
+      "loss": 0.2652,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176111547143237e-05,
+      "loss": 0.2626,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170360745031243e-05,
+      "loss": 0.2737,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164590810602262e-05,
+      "loss": 0.2678,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915880175589423e-05,
+      "loss": 0.2689,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152993592984983e-05,
+      "loss": 0.2632,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147166333992205e-05,
+      "loss": 0.2717,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141319991073432e-05,
+      "loss": 0.2689,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.269,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129570102287084e-05,
+      "loss": 0.2722,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123666580933547e-05,
+      "loss": 0.2614,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117744024682044e-05,
+      "loss": 0.2681,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9111802445888936e-05,
+      "loss": 0.2696,
+      "step": 362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105841856950254e-05,
+      "loss": 0.2585,
+      "step": 363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099862270301704e-05,
+      "loss": 0.2721,
+      "step": 364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093863698418627e-05,
+      "loss": 0.266,
+      "step": 365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087846153815965e-05,
+      "loss": 0.2693,
+      "step": 366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081809649048254e-05,
+      "loss": 0.2684,
+      "step": 367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 0.2617,
+      "step": 368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069679809433545e-05,
+      "loss": 0.2661,
+      "step": 369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063586499893292e-05,
+      "loss": 0.2765,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057474280801414e-05,
+      "loss": 0.2661,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051343164909965e-05,
+      "loss": 0.2672,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045193165010417e-05,
+      "loss": 0.266,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039024293933647e-05,
+      "loss": 0.2693,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90328365645499e-05,
+      "loss": 0.2684,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026629989768764e-05,
+      "loss": 0.2705,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020404582539153e-05,
+      "loss": 0.275,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901416035584926e-05,
+      "loss": 0.264,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007897322726546e-05,
+      "loss": 0.272,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001615496237714e-05,
+      "loss": 0.2726,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899531488948867e-05,
+      "loss": 0.2623,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988995515624497e-05,
+      "loss": 0.2597,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8982657387829447e-05,
+      "loss": 0.2725,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897630051932688e-05,
+      "loss": 0.2693,
+      "step": 384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896992492337927e-05,
+      "loss": 0.2695,
+      "step": 385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896353061328816e-05,
+      "loss": 0.2779,
+      "step": 386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.2672,
+      "step": 387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895068590407678e-05,
+      "loss": 0.2753,
+      "step": 388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944235531754697e-05,
+      "loss": 0.2727,
+      "step": 389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937766498885435e-05,
+      "loss": 0.2617,
+      "step": 390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931278818965463e-05,
+      "loss": 0.2616,
+      "step": 391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8924772505530177e-05,
+      "loss": 0.2647,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918247572153822e-05,
+      "loss": 0.2759,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911704032449515e-05,
+      "loss": 0.2733,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890514190006918e-05,
+      "loss": 0.2679,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898561188703527e-05,
+      "loss": 0.2667,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891961912082033e-05,
+      "loss": 0.2677,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885344083972912e-05,
+      "loss": 0.2662,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887870771818308e-05,
+      "loss": 0.265,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887205282855812e-05,
+      "loss": 0.2703,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8865379428982272e-05,
+      "loss": 0.2699,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885868753337839e-05,
+      "loss": 0.2761,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851977155707917e-05,
+      "loss": 0.2678,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845248309970855e-05,
+      "loss": 0.271,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883850101020574e-05,
+      "loss": 0.2671,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88317352704896e-05,
+      "loss": 0.2687,
+      "step": 406
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882495110493795e-05,
+      "loss": 0.2718,
+      "step": 407
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881814852770472e-05,
+      "loss": 0.2665,
+      "step": 408
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881132755298229e-05,
+      "loss": 0.2672,
+      "step": 409
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8804488195001394e-05,
+      "loss": 0.2708,
+      "step": 410
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797630468031128e-05,
+      "loss": 0.2675,
+      "step": 411
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8790754386378914e-05,
+      "loss": 0.2725,
+      "step": 412
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783859964390466e-05,
+      "loss": 0.2596,
+      "step": 413
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877694721644976e-05,
+      "loss": 0.2679,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8770016156979016e-05,
+      "loss": 0.271,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.2571,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756099161327222e-05,
+      "loss": 0.2707,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749113254181498e-05,
+      "loss": 0.2726,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742109093576313e-05,
+      "loss": 0.2706,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735086694124596e-05,
+      "loss": 0.2699,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728046070477322e-05,
+      "loss": 0.2706,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720987237323497e-05,
+      "loss": 0.2732,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713910209390113e-05,
+      "loss": 0.2696,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706815001442124e-05,
+      "loss": 0.2693,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869970162828241e-05,
+      "loss": 0.2616,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692570104751744e-05,
+      "loss": 0.267,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685420445728785e-05,
+      "loss": 0.2709,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678252666130016e-05,
+      "loss": 0.2699,
+      "step": 428
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671066780909722e-05,
+      "loss": 0.2711,
+      "step": 429
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866386280505997e-05,
+      "loss": 0.2666,
+      "step": 430
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656640753610563e-05,
+      "loss": 0.2637,
+      "step": 431
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8649400641629028e-05,
+      "loss": 0.2605,
+      "step": 432
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864214248422055e-05,
+      "loss": 0.2599,
+      "step": 433
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863486629652799e-05,
+      "loss": 0.2685,
+      "step": 434
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86275720937318e-05,
+      "loss": 0.267,
+      "step": 435
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8620259891050037e-05,
+      "loss": 0.2694,
+      "step": 436
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612929703738302e-05,
+      "loss": 0.2487,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605581547089715e-05,
+      "loss": 0.2608,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598215436434894e-05,
+      "loss": 0.2634,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859083138714191e-05,
+      "loss": 0.2736,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858342941461626e-05,
+      "loss": 0.2639,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857600953430084e-05,
+      "loss": 0.2704,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568571761675893e-05,
+      "loss": 0.2753,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856111611225901e-05,
+      "loss": 0.2544,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855364260160507e-05,
+      "loss": 0.2663,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8546151245306205e-05,
+      "loss": 0.2633,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538642058991803e-05,
+      "loss": 0.2624,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853111505832843e-05,
+      "loss": 0.2615,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523570259019827e-05,
+      "loss": 0.2668,
+      "step": 449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851600767680687e-05,
+      "loss": 0.2531,
+      "step": 450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850842732746753e-05,
+      "loss": 0.2552,
+      "step": 451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500829226816853e-05,
+      "loss": 0.2618,
+      "step": 452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849321339070692e-05,
+      "loss": 0.2638,
+      "step": 453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84855798350268e-05,
+      "loss": 0.2575,
+      "step": 454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847792857570255e-05,
+      "loss": 0.2624,
+      "step": 455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847025962869715e-05,
+      "loss": 0.2656,
+      "step": 456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846257301001049e-05,
+      "loss": 0.2672,
+      "step": 457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845486873567932e-05,
+      "loss": 0.2592,
+      "step": 458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844714682177723e-05,
+      "loss": 0.2628,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8439407284414613e-05,
+      "loss": 0.2557,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431650139738632e-05,
+      "loss": 0.2627,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8423875403933177e-05,
+      "loss": 0.2722,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416083093218847e-05,
+      "loss": 0.2657,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84082732238529e-05,
+      "loss": 0.2623,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400445812129234e-05,
+      "loss": 0.259,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392600874378343e-05,
+      "loss": 0.2535,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384738426967283e-05,
+      "loss": 0.2627,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837685848629965e-05,
+      "loss": 0.2577,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368961068815525e-05,
+      "loss": 0.2567,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361046190991457e-05,
+      "loss": 0.2657,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353113869340425e-05,
+      "loss": 0.2693,
+      "step": 471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345164120411798e-05,
+      "loss": 0.2597,
+      "step": 472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337196960791305e-05,
+      "loss": 0.255,
+      "step": 473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329212407100996e-05,
+      "loss": 0.255,
+      "step": 474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832121047599922e-05,
+      "loss": 0.265,
+      "step": 475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313191184180567e-05,
+      "loss": 0.262,
+      "step": 476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8305154548375862e-05,
+      "loss": 0.2675,
+      "step": 477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82971005853521e-05,
+      "loss": 0.2681,
+      "step": 478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289029311912442e-05,
+      "loss": 0.2594,
+      "step": 479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280940744896152e-05,
+      "loss": 0.2641,
+      "step": 480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827283490117858e-05,
+      "loss": 0.2629,
+      "step": 481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826471179767111e-05,
+      "loss": 0.2623,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256571451321155e-05,
+      "loss": 0.2718,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824841387911209e-05,
+      "loss": 0.2592,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824023909806322e-05,
+      "loss": 0.2692,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232047125229775e-05,
+      "loss": 0.2584,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223837977702834e-05,
+      "loss": 0.2662,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215611672609316e-05,
+      "loss": 0.2673,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207368227111936e-05,
+      "loss": 0.253,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8199107658409168e-05,
+      "loss": 0.2671,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190829983735206e-05,
+      "loss": 0.2759,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182535220359944e-05,
+      "loss": 0.2468,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817422338558892e-05,
+      "loss": 0.274,
+      "step": 493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165894496763287e-05,
+      "loss": 0.2607,
+      "step": 494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157548571259784e-05,
+      "loss": 0.2634,
+      "step": 495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149185626490692e-05,
+      "loss": 0.2712,
+      "step": 496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140805679903793e-05,
+      "loss": 0.2618,
+      "step": 497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132408748982356e-05,
+      "loss": 0.2672,
+      "step": 498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123994851245067e-05,
+      "loss": 0.2595,
+      "step": 499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115564004246025e-05,
+      "loss": 0.2713,
+      "step": 500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107116225574674e-05,
+      "loss": 0.2567,
+      "step": 501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098651532855806e-05,
+      "loss": 0.2686,
+      "step": 502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.2721,
+      "step": 503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808167147595101e-05,
+      "loss": 0.2557,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8073156147190933e-05,
+      "loss": 0.2561,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064623975234964e-05,
+      "loss": 0.2616,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056074977883946e-05,
+      "loss": 0.2566,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047509172973832e-05,
+      "loss": 0.2605,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038926578375654e-05,
+      "loss": 0.2602,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030327211995443e-05,
+      "loss": 0.2651,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802171109177425e-05,
+      "loss": 0.2625,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801307823568806e-05,
+      "loss": 0.2623,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800442866174779e-05,
+      "loss": 0.2586,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7995762387999228e-05,
+      "loss": 0.2617,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987079432522997e-05,
+      "loss": 0.2551,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978379813434535e-05,
+      "loss": 0.2719,
+      "step": 516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796966354888404e-05,
+      "loss": 0.2658,
+      "step": 517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 0.2522,
+      "step": 518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795218115617135e-05,
+      "loss": 0.2544,
+      "step": 519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943415064483035e-05,
+      "loss": 0.2539,
+      "step": 520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793463240028038e-05,
+      "loss": 0.2654,
+      "step": 521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925833181886845e-05,
+      "loss": 0.2676,
+      "step": 522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917017427660424e-05,
+      "loss": 0.2534,
+      "step": 523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908185155993604e-05,
+      "loss": 0.2632,
+      "step": 524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789933638531335e-05,
+      "loss": 0.2578,
+      "step": 525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789047113408103e-05,
+      "loss": 0.2633,
+      "step": 526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788158942079241e-05,
+      "loss": 0.2482,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78726912639776e-05,
+      "loss": 0.257,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863776682200997e-05,
+      "loss": 0.2585,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854845694061294e-05,
+      "loss": 0.254,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845898318191395e-05,
+      "loss": 0.2649,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78369345732584e-05,
+      "loss": 0.2601,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782795447796356e-05,
+      "loss": 0.2707,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7818958051042234e-05,
+      "loss": 0.2529,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809945311263863e-05,
+      "loss": 0.2527,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780091627743191e-05,
+      "loss": 0.2614,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779187096838384e-05,
+      "loss": 0.2556,
+      "step": 537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782809402991076e-05,
+      "loss": 0.2597,
+      "step": 538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773731600158947e-05,
+      "loss": 0.2556,
+      "step": 539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776463757882667e-05,
+      "loss": 0.2636,
+      "step": 540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755527357967288e-05,
+      "loss": 0.2485,
+      "step": 541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746400956587653e-05,
+      "loss": 0.2687,
+      "step": 542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737258393728363e-05,
+      "loss": 0.2533,
+      "step": 543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728099688463746e-05,
+      "loss": 0.2488,
+      "step": 544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718924859901793e-05,
+      "loss": 0.261,
+      "step": 545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770973392718415e-05,
+      "loss": 0.2596,
+      "step": 546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770052690948605e-05,
+      "loss": 0.2588,
+      "step": 547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769130382601629e-05,
+      "loss": 0.2562,
+      "step": 548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7682064696017184e-05,
+      "loss": 0.2627,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767280953876452e-05,
+      "loss": 0.26,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766353837356753e-05,
+      "loss": 0.2632,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765425121976884e-05,
+      "loss": 0.2546,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764494809674444e-05,
+      "loss": 0.2649,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763562902390363e-05,
+      "loss": 0.2626,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626294020688984e-05,
+      "loss": 0.2652,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616943106576316e-05,
+      "loss": 0.2628,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760757630107464e-05,
+      "loss": 0.2618,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598193623726115e-05,
+      "loss": 0.2617,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588795094106024e-05,
+      "loss": 0.2608,
+      "step": 559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579380731822712e-05,
+      "loss": 0.2756,
+      "step": 560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7569950556517566e-05,
+      "loss": 0.2537,
+      "step": 561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756050458786496e-05,
+      "loss": 0.2656,
+      "step": 562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755104284557221e-05,
+      "loss": 0.2586,
+      "step": 563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541565349379555e-05,
+      "loss": 0.2621,
+      "step": 564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532072119060094e-05,
+      "loss": 0.2575,
+      "step": 565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522563174419753e-05,
+      "loss": 0.2545,
+      "step": 566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513038535297247e-05,
+      "loss": 0.2593,
+      "step": 567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503498221564026e-05,
+      "loss": 0.2587,
+      "step": 568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749394225312425e-05,
+      "loss": 0.2705,
+      "step": 569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748437064991474e-05,
+      "loss": 0.2557,
+      "step": 570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747478343190493e-05,
+      "loss": 0.2557,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7465180619096834e-05,
+      "loss": 0.2563,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455562231525006e-05,
+      "loss": 0.2642,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744592828925649e-05,
+      "loss": 0.2592,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436278812390788e-05,
+      "loss": 0.2567,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426613821059796e-05,
+      "loss": 0.253,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416933335427798e-05,
+      "loss": 0.2568,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407237375691394e-05,
+      "loss": 0.2568,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397525962079465e-05,
+      "loss": 0.2703,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738779911485314e-05,
+      "loss": 0.2579,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737805685430575e-05,
+      "loss": 0.2601,
+      "step": 581
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368299200762768e-05,
+      "loss": 0.2612,
+      "step": 582
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358526174581806e-05,
+      "loss": 0.2608,
+      "step": 583
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7348737796152525e-05,
+      "loss": 0.2542,
+      "step": 584
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338934085896626e-05,
+      "loss": 0.262,
+      "step": 585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7329115064267802e-05,
+      "loss": 0.2646,
+      "step": 586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731928075175168e-05,
+      "loss": 0.2592,
+      "step": 587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730943116886579e-05,
+      "loss": 0.2562,
+      "step": 588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7299566336159536e-05,
+      "loss": 0.2596,
+      "step": 589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7289686274214116e-05,
+      "loss": 0.2594,
+      "step": 590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279791003642515e-05,
+      "loss": 0.2576,
+      "step": 591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726988054508945e-05,
+      "loss": 0.2545,
+      "step": 592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725995491923131e-05,
+      "loss": 0.2424,
+      "step": 593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250014146776143e-05,
+      "loss": 0.2694,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240058248463593e-05,
+      "loss": 0.2631,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230087245064857e-05,
+      "loss": 0.2631,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7220101157382653e-05,
+      "loss": 0.2613,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210100006251164e-05,
+      "loss": 0.2549,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200083812536e-05,
+      "loss": 0.2653,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190052597134164e-05,
+      "loss": 0.2598,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718000638097399e-05,
+      "loss": 0.2537,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7169945185015106e-05,
+      "loss": 0.2656,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7159869030248404e-05,
+      "loss": 0.2603,
+      "step": 603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714977793769597e-05,
+      "loss": 0.2574,
+      "step": 604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139671928411074e-05,
+      "loss": 0.2501,
+      "step": 605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129551023478085e-05,
+      "loss": 0.2565,
+      "step": 606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7119415244012464e-05,
+      "loss": 0.248,
+      "step": 607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710926461116071e-05,
+      "loss": 0.2616,
+      "step": 608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709909914610029e-05,
+      "loss": 0.2576,
+      "step": 609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088918870039634e-05,
+      "loss": 0.2565,
+      "step": 610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078723804218067e-05,
+      "loss": 0.2553,
+      "step": 611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706851396990577e-05,
+      "loss": 0.2576,
+      "step": 612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705828938840373e-05,
+      "loss": 0.2681,
+      "step": 613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048050081043713e-05,
+      "loss": 0.2578,
+      "step": 614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037796069188203e-05,
+      "loss": 0.2605,
+      "step": 615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7027527374230357e-05,
+      "loss": 0.2607,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701724401759397e-05,
+      "loss": 0.2616,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006946020733426e-05,
+      "loss": 0.2601,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996633405133656e-05,
+      "loss": 0.2598,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986306192310086e-05,
+      "loss": 0.2575,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975964403808595e-05,
+      "loss": 0.2613,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965608061205477e-05,
+      "loss": 0.2628,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955237186107387e-05,
+      "loss": 0.2662,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69448518001513e-05,
+      "loss": 0.2594,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693445192500447e-05,
+      "loss": 0.264,
+      "step": 625
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692403758236437e-05,
+      "loss": 0.262,
+      "step": 626
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913608793958664e-05,
+      "loss": 0.2495,
+      "step": 627
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690316558154516e-05,
+      "loss": 0.2496,
+      "step": 628
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689270796691174e-05,
+      "loss": 0.2568,
+      "step": 629
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882235971876352e-05,
+      "loss": 0.2609,
+      "step": 630
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6871749618286943e-05,
+      "loss": 0.2661,
+      "step": 631
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686124892802141e-05,
+      "loss": 0.2523,
+      "step": 632
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685073392298757e-05,
+      "loss": 0.2533,
+      "step": 633
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840204625123088e-05,
+      "loss": 0.2535,
+      "step": 634
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829661056395473e-05,
+      "loss": 0.2543,
+      "step": 635
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6819103238801987e-05,
+      "loss": 0.2535,
+      "step": 636
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808531194369628e-05,
+      "loss": 0.2568,
+      "step": 637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679794494515508e-05,
+      "loss": 0.2507,
+      "step": 638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6787344513244654e-05,
+      "loss": 0.2494,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776729920754254e-05,
+      "loss": 0.2596,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766101189829328e-05,
+      "loss": 0.2768,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755458342644825e-05,
+      "loss": 0.2528,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744801401405138e-05,
+      "loss": 0.2529,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734130388344073e-05,
+      "loss": 0.2562,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723445325724785e-05,
+      "loss": 0.2471,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671274623583975e-05,
+      "loss": 0.2574,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6702033141010695e-05,
+      "loss": 0.2537,
+      "step": 647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.2545,
+      "step": 648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680565025953542e-05,
+      "loss": 0.2546,
+      "step": 649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669810050514827e-05,
+      "loss": 0.2556,
+      "step": 650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659041159710762e-05,
+      "loss": 0.2575,
+      "step": 651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648258376008723e-05,
+      "loss": 0.2468,
+      "step": 652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637461721905047e-05,
+      "loss": 0.2586,
+      "step": 653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626651219925027e-05,
+      "loss": 0.2531,
+      "step": 654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661582689262284e-05,
+      "loss": 0.2442,
+      "step": 655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660498876258151e-05,
+      "loss": 0.2537,
+      "step": 656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6594136852412853e-05,
+      "loss": 0.2571,
+      "step": 657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583271184757444e-05,
+      "loss": 0.2516,
+      "step": 658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657239178228455e-05,
+      "loss": 0.2572,
+      "step": 659
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561498667692096e-05,
+      "loss": 0.2506,
+      "step": 660
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655059186370662e-05,
+      "loss": 0.2654,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6539671393083218e-05,
+      "loss": 0.2608,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528737278605492e-05,
+      "loss": 0.2524,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651778954308552e-05,
+      "loss": 0.2539,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506828209363796e-05,
+      "loss": 0.2596,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649585330030918e-05,
+      "loss": 0.2542,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484864838818856e-05,
+      "loss": 0.256,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 0.2504,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6462847350261137e-05,
+      "loss": 0.2593,
+      "step": 669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6451818369129298e-05,
+      "loss": 0.2555,
+      "step": 670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440775927432754e-05,
+      "loss": 0.2424,
+      "step": 671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6429720048209586e-05,
+      "loss": 0.2457,
+      "step": 672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641865075452591e-05,
+      "loss": 0.2629,
+      "step": 673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407568069475822e-05,
+      "loss": 0.2479,
+      "step": 674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396472016181363e-05,
+      "loss": 0.246,
+      "step": 675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385362617792465e-05,
+      "loss": 0.2501,
+      "step": 676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.2503,
+      "step": 677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636310387847023e-05,
+      "loss": 0.248,
+      "step": 678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635195458397577e-05,
+      "loss": 0.2506,
+      "step": 679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634079203726453e-05,
+      "loss": 0.2578,
+      "step": 680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329616261625158e-05,
+      "loss": 0.2559,
+      "step": 681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631842728037392e-05,
+      "loss": 0.259,
+      "step": 682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307225116854623e-05,
+      "loss": 0.2532,
+      "step": 683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6296009794438567e-05,
+      "loss": 0.25,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284781336524528e-05,
+      "loss": 0.2478,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627353976653866e-05,
+      "loss": 0.2519,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6262285107934506e-05,
+      "loss": 0.2559,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251017384192885e-05,
+      "loss": 0.2504,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239736618821886e-05,
+      "loss": 0.2527,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228442835356815e-05,
+      "loss": 0.2571,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217136057360122e-05,
+      "loss": 0.2529,
+      "step": 691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205816308421386e-05,
+      "loss": 0.2661,
+      "step": 692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194483612157232e-05,
+      "loss": 0.2519,
+      "step": 693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6183137992211303e-05,
+      "loss": 0.254,
+      "step": 694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171779472254206e-05,
+      "loss": 0.2562,
+      "step": 695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616040807598346e-05,
+      "loss": 0.2489,
+      "step": 696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614902382712345e-05,
+      "loss": 0.2648,
+      "step": 697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137626749425377e-05,
+      "loss": 0.2663,
+      "step": 698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61262168666672e-05,
+      "loss": 0.2574,
+      "step": 699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6114794202653602e-05,
+      "loss": 0.2504,
+      "step": 700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610335878121592e-05,
+      "loss": 0.2498,
+      "step": 701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6091910626212124e-05,
+      "loss": 0.2468,
+      "step": 702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080449761526737e-05,
+      "loss": 0.2582,
+      "step": 703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.60689762110708e-05,
+      "loss": 0.2527,
+      "step": 704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605748999878183e-05,
+      "loss": 0.2567,
+      "step": 705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045991148623752e-05,
+      "loss": 0.2543,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6034479684586854e-05,
+      "loss": 0.2472,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022955630687752e-05,
+      "loss": 0.2508,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011419010969316e-05,
+      "loss": 0.2472,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599986984950065e-05,
+      "loss": 0.2565,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5988308170377004e-05,
+      "loss": 0.2552,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5976733997719755e-05,
+      "loss": 0.2464,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965147355676344e-05,
+      "loss": 0.245,
+      "step": 713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953548268420228e-05,
+      "loss": 0.2455,
+      "step": 714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5941936760150825e-05,
+      "loss": 0.2468,
+      "step": 715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593031285509347e-05,
+      "loss": 0.2557,
+      "step": 716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591867657749936e-05,
+      "loss": 0.2572,
+      "step": 717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590702795164551e-05,
+      "loss": 0.253,
+      "step": 718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895367001834694e-05,
+      "loss": 0.2496,
+      "step": 719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5883693752395392e-05,
+      "loss": 0.2529,
+      "step": 720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5872008227681757e-05,
+      "loss": 0.2452,
+      "step": 721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586031045207354e-05,
+      "loss": 0.2504,
+      "step": 722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584860044997606e-05,
+      "loss": 0.2587,
+      "step": 723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583687824582014e-05,
+      "loss": 0.2521,
+      "step": 724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5825143864062063e-05,
+      "loss": 0.2413,
+      "step": 725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5813397329183514e-05,
+      "loss": 0.2534,
+      "step": 726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580163866569154e-05,
+      "loss": 0.2488,
+      "step": 727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578986789811849e-05,
+      "loss": 0.2621,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5778085051021957e-05,
+      "loss": 0.2547,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576629014898475e-05,
+      "loss": 0.2531,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5754483216614822e-05,
+      "loss": 0.2457,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574266427854522e-05,
+      "loss": 0.2542,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573083335943405e-05,
+      "loss": 0.2449,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57189904839644e-05,
+      "loss": 0.246,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 0.2536,
+      "step": 735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695268962806736e-05,
+      "loss": 0.2425,
+      "step": 736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683390366609428e-05,
+      "loss": 0.256,
+      "step": 737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671499913034962e-05,
+      "loss": 0.248,
+      "step": 738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565959762689064e-05,
+      "loss": 0.2563,
+      "step": 739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647683533008455e-05,
+      "loss": 0.2549,
+      "step": 740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5635757656245026e-05,
+      "loss": 0.2474,
+      "step": 741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562382002148157e-05,
+      "loss": 0.2553,
+      "step": 742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611870653623826e-05,
+      "loss": 0.2559,
+      "step": 743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5599909577602003e-05,
+      "loss": 0.2488,
+      "step": 744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587936818370754e-05,
+      "loss": 0.25,
+      "step": 745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575952400909093e-05,
+      "loss": 0.2486,
+      "step": 746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563956350220366e-05,
+      "loss": 0.2501,
+      "step": 747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551948691332192e-05,
+      "loss": 0.2524,
+      "step": 748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5539929449296392e-05,
+      "loss": 0.2511,
+      "step": 749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527898649188974e-05,
+      "loss": 0.2511,
+      "step": 750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5515856316110044e-05,
+      "loss": 0.2502,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503802475183773e-05,
+      "loss": 0.2515,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491737151558352e-05,
+      "loss": 0.256,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5479660370405913e-05,
+      "loss": 0.2465,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5467572156922504e-05,
+      "loss": 0.2477,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5455472536328016e-05,
+      "loss": 0.242,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443361533866146e-05,
+      "loss": 0.2467,
+      "step": 757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543123917480433e-05,
+      "loss": 0.256,
+      "step": 758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419105484433704e-05,
+      "loss": 0.2548,
+      "step": 759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540696048806904e-05,
+      "loss": 0.2535,
+      "step": 760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5394804211048702e-05,
+      "loss": 0.2429,
+      "step": 761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382636678734584e-05,
+      "loss": 0.2582,
+      "step": 762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370457916512072e-05,
+      "loss": 0.254,
+      "step": 763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.2517,
+      "step": 764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534606680400046e-05,
+      "loss": 0.2545,
+      "step": 765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533385450459905e-05,
+      "loss": 0.249,
+      "step": 766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321631077064522e-05,
+      "loss": 0.2415,
+      "step": 767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530939654689887e-05,
+      "loss": 0.248,
+      "step": 768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529715093962725e-05,
+      "loss": 0.2521,
+      "step": 769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528489428079793e-05,
+      "loss": 0.2514,
+      "step": 770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527262659598224e-05,
+      "loss": 0.2566,
+      "step": 771
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5260347910774503e-05,
+      "loss": 0.2413,
+      "step": 772
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248058250792008e-05,
+      "loss": 0.2486,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235757641674922e-05,
+      "loss": 0.2519,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223446109086273e-05,
+      "loss": 0.255,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211123678711865e-05,
+      "loss": 0.2499,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198790376260247e-05,
+      "loss": 0.249,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186446227462646e-05,
+      "loss": 0.2475,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174091258072926e-05,
+      "loss": 0.2368,
+      "step": 779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161725493867514e-05,
+      "loss": 0.2425,
+      "step": 780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149348960645367e-05,
+      "loss": 0.2528,
+      "step": 781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136961684227905e-05,
+      "loss": 0.2505,
+      "step": 782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124563690458966e-05,
+      "loss": 0.2512,
+      "step": 783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112155005204749e-05,
+      "loss": 0.2545,
+      "step": 784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509973565435375e-05,
+      "loss": 0.2453,
+      "step": 785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087305663816723e-05,
+      "loss": 0.2506,
+      "step": 786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074865059526622e-05,
+      "loss": 0.2458,
+      "step": 787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506241386743854e-05,
+      "loss": 0.2473,
+      "step": 788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049952113529666e-05,
+      "loss": 0.2498,
+      "step": 789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5037479823799217e-05,
+      "loss": 0.2515,
+      "step": 790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024997024268388e-05,
+      "loss": 0.247,
+      "step": 791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.501250374098032e-05,
+      "loss": 0.247,
+      "step": 792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.252,
+      "step": 793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987485827414257e-05,
+      "loss": 0.2492,
+      "step": 794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974961249331666e-05,
+      "loss": 0.2434,
+      "step": 795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962426291882524e-05,
+      "loss": 0.249,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949880981218768e-05,
+      "loss": 0.2498,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937325343513953e-05,
+      "loss": 0.2503,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492475940496317e-05,
+      "loss": 0.2544,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912183191782995e-05,
+      "loss": 0.254,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899596730211454e-05,
+      "loss": 0.2413,
+      "step": 801
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887000046507945e-05,
+      "loss": 0.238,
+      "step": 802
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874393166953194e-05,
+      "loss": 0.2543,
+      "step": 803
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861776117849198e-05,
+      "loss": 0.2405,
+      "step": 804
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849148925519174e-05,
+      "loss": 0.2509,
+      "step": 805
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836511616307501e-05,
+      "loss": 0.2388,
+      "step": 806
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823864216579664e-05,
+      "loss": 0.2579,
+      "step": 807
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811206752722196e-05,
+      "loss": 0.2466,
+      "step": 808
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798539251142634e-05,
+      "loss": 0.2522,
+      "step": 809
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4785861738269451e-05,
+      "loss": 0.2561,
+      "step": 810
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773174240552016e-05,
+      "loss": 0.2517,
+      "step": 811
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760476784460514e-05,
+      "loss": 0.2477,
+      "step": 812
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4747769396485917e-05,
+      "loss": 0.2426,
+      "step": 813
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4735052103139927e-05,
+      "loss": 0.2438,
+      "step": 814
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722324930954885e-05,
+      "loss": 0.2373,
+      "step": 815
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470958790648377e-05,
+      "loss": 0.2387,
+      "step": 816
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696841056300102e-05,
+      "loss": 0.2443,
+      "step": 817
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 0.252,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.467131798519164e-05,
+      "loss": 0.2401,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4658541817516166e-05,
+      "loss": 0.2453,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645755930626671e-05,
+      "loss": 0.2426,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463296035119862e-05,
+      "loss": 0.2508,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620155105927699e-05,
+      "loss": 0.2482,
+      "step": 823
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4607340221529762e-05,
+      "loss": 0.2366,
+      "step": 824
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594515724740775e-05,
+      "loss": 0.2409,
+      "step": 825
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581681642316752e-05,
+      "loss": 0.2462,
+      "step": 826
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568838001033711e-05,
+      "loss": 0.247,
+      "step": 827
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555984827687613e-05,
+      "loss": 0.2436,
+      "step": 828
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543122149094304e-05,
+      "loss": 0.2434,
+      "step": 829
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453024999208946e-05,
+      "loss": 0.2507,
+      "step": 830
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517368383528534e-05,
+      "loss": 0.2473,
+      "step": 831
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504477350286701e-05,
+      "loss": 0.25,
+      "step": 832
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491576919258794e-05,
+      "loss": 0.2504,
+      "step": 833
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478667117359251e-05,
+      "loss": 0.245,
+      "step": 834
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465747971522067e-05,
+      "loss": 0.252,
+      "step": 835
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452819508700724e-05,
+      "loss": 0.2497,
+      "step": 836
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4439881755868153e-05,
+      "loss": 0.2525,
+      "step": 837
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426934740016655e-05,
+      "loss": 0.2367,
+      "step": 838
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4413978488157867e-05,
+      "loss": 0.2393,
+      "step": 839
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440101302732269e-05,
+      "loss": 0.2458,
+      "step": 840
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4388038384561234e-05,
+      "loss": 0.2492,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375054586942771e-05,
+      "loss": 0.2492,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362061661555675e-05,
+      "loss": 0.2482,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4349059635507359e-05,
+      "loss": 0.2402,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336048535924223e-05,
+      "loss": 0.2524,
+      "step": 845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4323028389951597e-05,
+      "loss": 0.2488,
+      "step": 846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4309999224753688e-05,
+      "loss": 0.2477,
+      "step": 847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296961067513519e-05,
+      "loss": 0.2479,
+      "step": 848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283913945432871e-05,
+      "loss": 0.2451,
+      "step": 849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427085788573223e-05,
+      "loss": 0.2465,
+      "step": 850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257792915650728e-05,
+      "loss": 0.2481,
+      "step": 851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424471906244609e-05,
+      "loss": 0.2502,
+      "step": 852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423163635339457e-05,
+      "loss": 0.2426,
+      "step": 853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4218544815790897e-05,
+      "loss": 0.2403,
+      "step": 854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205444476948225e-05,
+      "loss": 0.2492,
+      "step": 855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419233536419807e-05,
+      "loss": 0.2459,
+      "step": 856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417921750489024e-05,
+      "loss": 0.2409,
+      "step": 857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166090926392815e-05,
+      "loss": 0.2475,
+      "step": 858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4152955656092046e-05,
+      "loss": 0.2438,
+      "step": 859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139811721392325e-05,
+      "loss": 0.2495,
+      "step": 860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4126659149716115e-05,
+      "loss": 0.2414,
+      "step": 861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411349796850391e-05,
+      "loss": 0.2393,
+      "step": 862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100328205214161e-05,
+      "loss": 0.2411,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408714988732322e-05,
+      "loss": 0.2551,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073963042325284e-05,
+      "loss": 0.2456,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060767697732356e-05,
+      "loss": 0.2445,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047563881074157e-05,
+      "loss": 0.2574,
+      "step": 867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034351619898088e-05,
+      "loss": 0.241,
+      "step": 868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021130941769165e-05,
+      "loss": 0.2467,
+      "step": 869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400790187426997e-05,
+      "loss": 0.2522,
+      "step": 870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399466444500059e-05,
+      "loss": 0.2533,
+      "step": 871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981418681578546e-05,
+      "loss": 0.2357,
+      "step": 872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3968164611638759e-05,
+      "loss": 0.251,
+      "step": 873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395490226283347e-05,
+      "loss": 0.2381,
+      "step": 874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941631662832201e-05,
+      "loss": 0.2434,
+      "step": 875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928352839321679e-05,
+      "loss": 0.244,
+      "step": 876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915065820005795e-05,
+      "loss": 0.2375,
+      "step": 877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901770632605546e-05,
+      "loss": 0.2401,
+      "step": 878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888467304858952e-05,
+      "loss": 0.2452,
+      "step": 879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875155864521031e-05,
+      "loss": 0.2532,
+      "step": 880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861836339363718e-05,
+      "loss": 0.2378,
+      "step": 881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848508757175825e-05,
+      "loss": 0.2471,
+      "step": 882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3835173145762961e-05,
+      "loss": 0.2491,
+      "step": 883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821829532947497e-05,
+      "loss": 0.2544,
+      "step": 884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808477946568494e-05,
+      "loss": 0.2453,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3795118414481648e-05,
+      "loss": 0.2436,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781750964559228e-05,
+      "loss": 0.246,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3768375624690033e-05,
+      "loss": 0.2491,
+      "step": 888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754992422779317e-05,
+      "loss": 0.2427,
+      "step": 889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741601386748728e-05,
+      "loss": 0.2475,
+      "step": 890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728202544536276e-05,
+      "loss": 0.2426,
+      "step": 891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714795924096243e-05,
+      "loss": 0.2442,
+      "step": 892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701381553399147e-05,
+      "loss": 0.2561,
+      "step": 893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687959460431668e-05,
+      "loss": 0.2447,
+      "step": 894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674529673196606e-05,
+      "loss": 0.2428,
+      "step": 895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661092219712803e-05,
+      "loss": 0.2503,
+      "step": 896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647647128015111e-05,
+      "loss": 0.2544,
+      "step": 897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634194426154301e-05,
+      "loss": 0.2514,
+      "step": 898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620734142197031e-05,
+      "loss": 0.2523,
+      "step": 899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607266304225779e-05,
+      "loss": 0.2426,
+      "step": 900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359379094033877e-05,
+      "loss": 0.2462,
+      "step": 901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580308078649948e-05,
+      "loss": 0.2436,
+      "step": 902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566817747288887e-05,
+      "loss": 0.245,
+      "step": 903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3553319974400756e-05,
+      "loss": 0.2486,
+      "step": 904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539814788146235e-05,
+      "loss": 0.2461,
+      "step": 905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.352630221670148e-05,
+      "loss": 0.2411,
+      "step": 906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512782288258061e-05,
+      "loss": 0.2402,
+      "step": 907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499255031022887e-05,
+      "loss": 0.2434,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485720473218153e-05,
+      "loss": 0.2484,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472178643081303e-05,
+      "loss": 0.2457,
+      "step": 910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458629568864931e-05,
+      "loss": 0.2354,
+      "step": 911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445073278836767e-05,
+      "loss": 0.2386,
+      "step": 912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3431509801279578e-05,
+      "loss": 0.2389,
+      "step": 913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417939164491136e-05,
+      "loss": 0.2419,
+      "step": 914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404361396784148e-05,
+      "loss": 0.2466,
+      "step": 915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390776526486193e-05,
+      "loss": 0.2414,
+      "step": 916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377184581939672e-05,
+      "loss": 0.2511,
+      "step": 917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3363585591501751e-05,
+      "loss": 0.2396,
+      "step": 918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349979583544291e-05,
+      "loss": 0.2456,
+      "step": 919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336366586453783e-05,
+      "loss": 0.2382,
+      "step": 920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332274662863132e-05,
+      "loss": 0.2506,
+      "step": 921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309119738492503e-05,
+      "loss": 0.2409,
+      "step": 922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295485944467405e-05,
+      "loss": 0.2393,
+      "step": 923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281845275000492e-05,
+      "loss": 0.2439,
+      "step": 924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326819775855058e-05,
+      "loss": 0.2386,
+      "step": 925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254543423590773e-05,
+      "loss": 0.2443,
+      "step": 926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240882298608402e-05,
+      "loss": 0.2426,
+      "step": 927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227214412104952e-05,
+      "loss": 0.2406,
+      "step": 928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213539792596027e-05,
+      "loss": 0.2465,
+      "step": 929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.319985846861127e-05,
+      "loss": 0.2383,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3186170468694327e-05,
+      "loss": 0.2409,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172475821402748e-05,
+      "loss": 0.2496,
+      "step": 932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158774555307972e-05,
+      "loss": 0.2482,
+      "step": 933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3145066698995233e-05,
+      "loss": 0.2395,
+      "step": 934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313135228106353e-05,
+      "loss": 0.2436,
+      "step": 935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311763133012553e-05,
+      "loss": 0.2483,
+      "step": 936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103903874807554e-05,
+      "loss": 0.2459,
+      "step": 937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.2426,
+      "step": 938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3076429565604685e-05,
+      "loss": 0.241,
+      "step": 939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062682769040026e-05,
+      "loss": 0.2405,
+      "step": 940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048929582735724e-05,
+      "loss": 0.2454,
+      "step": 941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3035170035385347e-05,
+      "loss": 0.2377,
+      "step": 942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021404155695728e-05,
+      "loss": 0.2436,
+      "step": 943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007631972386912e-05,
+      "loss": 0.2451,
+      "step": 944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299385351419209e-05,
+      "loss": 0.2419,
+      "step": 945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298006880985756e-05,
+      "loss": 0.2465,
+      "step": 946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2966277888142633e-05,
+      "loss": 0.2447,
+      "step": 947
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29524807778196e-05,
+      "loss": 0.245,
+      "step": 948
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293867750767367e-05,
+      "loss": 0.2339,
+      "step": 949
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292486810650289e-05,
+      "loss": 0.2313,
+      "step": 950
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911052603118109e-05,
+      "loss": 0.2421,
+      "step": 951
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2897231026342898e-05,
+      "loss": 0.2424,
+      "step": 952
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288340340501351e-05,
+      "loss": 0.2388,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28695697679788e-05,
+      "loss": 0.2467,
+      "step": 954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2855730144100174e-05,
+      "loss": 0.2387,
+      "step": 955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841884562251532e-05,
+      "loss": 0.2387,
+      "step": 956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282803305131921e-05,
+      "loss": 0.2409,
+      "step": 957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28141756402019e-05,
+      "loss": 0.2451,
+      "step": 958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800312357810616e-05,
+      "loss": 0.2419,
+      "step": 959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786443233068613e-05,
+      "loss": 0.2358,
+      "step": 960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772568294911339e-05,
+      "loss": 0.2407,
+      "step": 961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758687572286367e-05,
+      "loss": 0.2383,
+      "step": 962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274480109415335e-05,
+      "loss": 0.2358,
+      "step": 963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730908889483934e-05,
+      "loss": 0.2448,
+      "step": 964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717010987261716e-05,
+      "loss": 0.2443,
+      "step": 965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2703107416482188e-05,
+      "loss": 0.2401,
+      "step": 966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689198206152657e-05,
+      "loss": 0.2286,
+      "step": 967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 0.2347,
+      "step": 968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661362982931625e-05,
+      "loss": 0.2403,
+      "step": 969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647437028113332e-05,
+      "loss": 0.2421,
+      "step": 970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2633505549891345e-05,
+      "loss": 0.2406,
+      "step": 971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26195685773312e-05,
+      "loss": 0.2417,
+      "step": 972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605626139509899e-05,
+      "loss": 0.2401,
+      "step": 973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591678265515838e-05,
+      "loss": 0.2437,
+      "step": 974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257772498444877e-05,
+      "loss": 0.2429,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2563766325419708e-05,
+      "loss": 0.2461,
+      "step": 976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549802317550907e-05,
+      "loss": 0.2468,
+      "step": 977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535832989975766e-05,
+      "loss": 0.2424,
+      "step": 978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2521858371838792e-05,
+      "loss": 0.2486,
+      "step": 979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250787849229552e-05,
+      "loss": 0.2353,
+      "step": 980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2493893380512467e-05,
+      "loss": 0.2353,
+      "step": 981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479903065667072e-05,
+      "loss": 0.2374,
+      "step": 982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465907576947623e-05,
+      "loss": 0.2317,
+      "step": 983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451906943553195e-05,
+      "loss": 0.2406,
+      "step": 984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437901194693616e-05,
+      "loss": 0.2501,
+      "step": 985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423890359589368e-05,
+      "loss": 0.2433,
+      "step": 986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2409874467471556e-05,
+      "loss": 0.2441,
+      "step": 987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395853547581829e-05,
+      "loss": 0.2393,
+      "step": 988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2381827629172325e-05,
+      "loss": 0.2423,
+      "step": 989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367796741505618e-05,
+      "loss": 0.2462,
+      "step": 990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2353760913854644e-05,
+      "loss": 0.2387,
+      "step": 991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339720175502643e-05,
+      "loss": 0.2407,
+      "step": 992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325674555743106e-05,
+      "loss": 0.2447,
+      "step": 993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311624083879706e-05,
+      "loss": 0.2469,
+      "step": 994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229756878922624e-05,
+      "loss": 0.2409,
+      "step": 995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283508701106559e-05,
+      "loss": 0.2368,
+      "step": 996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269443848854528e-05,
+      "loss": 0.2368,
+      "step": 997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2255374261813944e-05,
+      "loss": 0.2375,
+      "step": 998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.224129996933848e-05,
+      "loss": 0.2463,
+      "step": 999
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227221000791631e-05,
+      "loss": 0.2363,
+      "step": 1000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2213137385546648e-05,
+      "loss": 0.2342,
+      "step": 1001
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199049152986473e-05,
+      "loss": 0.2405,
+      "step": 1002
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184956332503679e-05,
+      "loss": 0.2496,
+      "step": 1003
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170858953500417e-05,
+      "loss": 0.236,
+      "step": 1004
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215675704538835e-05,
+      "loss": 0.2409,
+      "step": 1005
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142650637588582e-05,
+      "loss": 0.2343,
+      "step": 1006
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212853975953161e-05,
+      "loss": 0.2406,
+      "step": 1007
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114424440657255e-05,
+      "loss": 0.2421,
+      "step": 1008
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100304710414606e-05,
+      "loss": 0.2602,
+      "step": 1009
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086180598261956e-05,
+      "loss": 0.2365,
+      "step": 1010
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072052133666733e-05,
+      "loss": 0.2395,
+      "step": 1011
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057919346105454e-05,
+      "loss": 0.2327,
+      "step": 1012
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.204378226506365e-05,
+      "loss": 0.2484,
+      "step": 1013
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2029640920035814e-05,
+      "loss": 0.2422,
+      "step": 1014
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015495340525329e-05,
+      "loss": 0.2505,
+      "step": 1015
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2001345556044417e-05,
+      "loss": 0.2362,
+      "step": 1016
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987191596114073e-05,
+      "loss": 0.2431,
+      "step": 1017
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973033490264e-05,
+      "loss": 0.2407,
+      "step": 1018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1958871268032555e-05,
+      "loss": 0.2503,
+      "step": 1019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194470495896668e-05,
+      "loss": 0.241,
+      "step": 1020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930534592621845e-05,
+      "loss": 0.2444,
+      "step": 1021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191636019856198e-05,
+      "loss": 0.2352,
+      "step": 1022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902181806359427e-05,
+      "loss": 0.234,
+      "step": 1023
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1887999445594866e-05,
+      "loss": 0.2395,
+      "step": 1024
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.2516,
+      "step": 1025
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1859622936743756e-05,
+      "loss": 0.2392,
+      "step": 1026
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845428847859718e-05,
+      "loss": 0.2383,
+      "step": 1027
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831230908818563e-05,
+      "loss": 0.2422,
+      "step": 1028
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817029149241752e-05,
+      "loss": 0.2415,
+      "step": 1029
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180282359875871e-05,
+      "loss": 0.2366,
+      "step": 1030
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788614287006785e-05,
+      "loss": 0.2411,
+      "step": 1031
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177440124363116e-05,
+      "loss": 0.2361,
+      "step": 1032
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1760184498284801e-05,
+      "loss": 0.236,
+      "step": 1033
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1745964080628411e-05,
+      "loss": 0.244,
+      "step": 1034
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1731740020330347e-05,
+      "loss": 0.2395,
+      "step": 1035
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717512347066562e-05,
+      "loss": 0.2353,
+      "step": 1036
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1703281090520553e-05,
+      "loss": 0.2348,
+      "step": 1037
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168904628038329e-05,
+      "loss": 0.2432,
+      "step": 1038
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674807946353159e-05,
+      "loss": 0.2422,
+      "step": 1039
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660566118135894e-05,
+      "loss": 0.2272,
+      "step": 1040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646320825444524e-05,
+      "loss": 0.238,
+      "step": 1041
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632072097999301e-05,
+      "loss": 0.2447,
+      "step": 1042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161781996552765e-05,
+      "loss": 0.2379,
+      "step": 1043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160356445776409e-05,
+      "loss": 0.24,
+      "step": 1044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158930560445019e-05,
+      "loss": 0.2434,
+      "step": 1045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575043435334494e-05,
+      "loss": 0.2378,
+      "step": 1046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1560777980172473e-05,
+      "loss": 0.2442,
+      "step": 1047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1546509268726437e-05,
+      "loss": 0.2373,
+      "step": 1048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532237330765508e-05,
+      "loss": 0.2487,
+      "step": 1049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1517962196065527e-05,
+      "loss": 0.2346,
+      "step": 1050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503683894409011e-05,
+      "loss": 0.2384,
+      "step": 1051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489402455585078e-05,
+      "loss": 0.244,
+      "step": 1052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475117909389396e-05,
+      "loss": 0.2387,
+      "step": 1053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460830285624119e-05,
+      "loss": 0.2408,
+      "step": 1054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1446539614097814e-05,
+      "loss": 0.2397,
+      "step": 1055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432245924625407e-05,
+      "loss": 0.225,
+      "step": 1056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.141794924702813e-05,
+      "loss": 0.2443,
+      "step": 1057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1403649611133444e-05,
+      "loss": 0.2358,
+      "step": 1058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389347046774977e-05,
+      "loss": 0.2418,
+      "step": 1059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375041583792467e-05,
+      "loss": 0.2451,
+      "step": 1060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136073325203171e-05,
+      "loss": 0.2431,
+      "step": 1061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134642208134448e-05,
+      "loss": 0.2347,
+      "step": 1062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332108101588468e-05,
+      "loss": 0.2269,
+      "step": 1063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317791342627239e-05,
+      "loss": 0.2442,
+      "step": 1064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303471834330142e-05,
+      "loss": 0.2362,
+      "step": 1065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289149606572273e-05,
+      "loss": 0.2402,
+      "step": 1066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12748246892344e-05,
+      "loss": 0.2366,
+      "step": 1067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260497112202895e-05,
+      "loss": 0.2397,
+      "step": 1068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1246166905369686e-05,
+      "loss": 0.2356,
+      "step": 1069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123183409863219e-05,
+      "loss": 0.2357,
+      "step": 1070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217498721893238e-05,
+      "loss": 0.2362,
+      "step": 1071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120316080506103e-05,
+      "loss": 0.2415,
+      "step": 1072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1188820378049066e-05,
+      "loss": 0.235,
+      "step": 1073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1174477470776082e-05,
+      "loss": 0.2418,
+      "step": 1074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160132113165985e-05,
+      "loss": 0.2407,
+      "step": 1075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145784335147793e-05,
+      "loss": 0.238,
+      "step": 1076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113143416665559e-05,
+      "loss": 0.2379,
+      "step": 1077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117081637628426e-05,
+      "loss": 0.2387,
+      "step": 1078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102726778010288e-05,
+      "loss": 0.2336,
+      "step": 1079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088369617750021e-05,
+      "loss": 0.2498,
+      "step": 1080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074010186801273e-05,
+      "loss": 0.232,
+      "step": 1081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1059648515122426e-05,
+      "loss": 0.2409,
+      "step": 1082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.2362,
+      "step": 1083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030918569431277e-05,
+      "loss": 0.233,
+      "step": 1084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1016550355358872e-05,
+      "loss": 0.2384,
+      "step": 1085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1002180020436025e-05,
+      "loss": 0.2375,
+      "step": 1086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987807594643867e-05,
+      "loss": 0.2413,
+      "step": 1087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0973433107967901e-05,
+      "loss": 0.2364,
+      "step": 1088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959056590397912e-05,
+      "loss": 0.2339,
+      "step": 1089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944678071927936e-05,
+      "loss": 0.234,
+      "step": 1090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930297582556182e-05,
+      "loss": 0.2324,
+      "step": 1091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915915152284966e-05,
+      "loss": 0.235,
+      "step": 1092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901530811120655e-05,
+      "loss": 0.2264,
+      "step": 1093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0887144589073601e-05,
+      "loss": 0.2435,
+      "step": 1094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872756516158087e-05,
+      "loss": 0.2267,
+      "step": 1095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0858366622392252e-05,
+      "loss": 0.2373,
+      "step": 1096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084397493779803e-05,
+      "loss": 0.2263,
+      "step": 1097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0829581492401103e-05,
+      "loss": 0.2372,
+      "step": 1098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0815186316230814e-05,
+      "loss": 0.2243,
+      "step": 1099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800789439320128e-05,
+      "loss": 0.2326,
+      "step": 1100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786390891705544e-05,
+      "loss": 0.2356,
+      "step": 1101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771990703427067e-05,
+      "loss": 0.2342,
+      "step": 1102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757588904528106e-05,
+      "loss": 0.233,
+      "step": 1103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0743185525055437e-05,
+      "loss": 0.2411,
+      "step": 1104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728780595059138e-05,
+      "loss": 0.2257,
+      "step": 1105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0714374144592516e-05,
+      "loss": 0.2414,
+      "step": 1106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699966203712057e-05,
+      "loss": 0.2327,
+      "step": 1107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685556802477347e-05,
+      "loss": 0.2393,
+      "step": 1108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.067114597095102e-05,
+      "loss": 0.2384,
+      "step": 1109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656733739198705e-05,
+      "loss": 0.2276,
+      "step": 1110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642320137288944e-05,
+      "loss": 0.2374,
+      "step": 1111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627905195293135e-05,
+      "loss": 0.2393,
+      "step": 1112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061348894328548e-05,
+      "loss": 0.2361,
+      "step": 1113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0599071411342903e-05,
+      "loss": 0.2391,
+      "step": 1114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584652629545011e-05,
+      "loss": 0.2341,
+      "step": 1115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570232627974008e-05,
+      "loss": 0.235,
+      "step": 1116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0555811436714651e-05,
+      "loss": 0.2307,
+      "step": 1117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 0.2389,
+      "step": 1118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0526965605482236e-05,
+      "loss": 0.2366,
+      "step": 1119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512541025690838e-05,
+      "loss": 0.2319,
+      "step": 1120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498115376574294e-05,
+      "loss": 0.2352,
+      "step": 1121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483688688229137e-05,
+      "loss": 0.2264,
+      "step": 1122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.046926099075407e-05,
+      "loss": 0.2266,
+      "step": 1123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0454832314249904e-05,
+      "loss": 0.2348,
+      "step": 1124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440402688819486e-05,
+      "loss": 0.2275,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042597214456766e-05,
+      "loss": 0.2307,
+      "step": 1126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041154071160116e-05,
+      "loss": 0.2385,
+      "step": 1127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0397108420028598e-05,
+      "loss": 0.2317,
+      "step": 1128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0382675299960367e-05,
+      "loss": 0.2366,
+      "step": 1129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036824138150859e-05,
+      "loss": 0.2334,
+      "step": 1130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353806694787053e-05,
+      "loss": 0.2401,
+      "step": 1131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339371269911147e-05,
+      "loss": 0.2368,
+      "step": 1132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324935136997807e-05,
+      "loss": 0.2385,
+      "step": 1133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310498326165439e-05,
+      "loss": 0.2378,
+      "step": 1134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296060867533865e-05,
+      "loss": 0.2416,
+      "step": 1135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281622791224257e-05,
+      "loss": 0.244,
+      "step": 1136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267184127359088e-05,
+      "loss": 0.2388,
+      "step": 1137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252744906062038e-05,
+      "loss": 0.226,
+      "step": 1138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.023830515745796e-05,
+      "loss": 0.2383,
+      "step": 1139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223864911672807e-05,
+      "loss": 0.2323,
+      "step": 1140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.2405,
+      "step": 1141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194983049068212e-05,
+      "loss": 0.2309,
+      "step": 1142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180541492505605e-05,
+      "loss": 0.2479,
+      "step": 1143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166099559275475e-05,
+      "loss": 0.2381,
+      "step": 1144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151657279508335e-05,
+      "loss": 0.2281,
+      "step": 1145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0137214683335414e-05,
+      "loss": 0.2287,
+      "step": 1146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122771800888606e-05,
+      "loss": 0.2302,
+      "step": 1147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0108328662300399e-05,
+      "loss": 0.2446,
+      "step": 1148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093885297703825e-05,
+      "loss": 0.2314,
+      "step": 1149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079441737232369e-05,
+      "loss": 0.2374,
+      "step": 1150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0064998011019945e-05,
+      "loss": 0.2313,
+      "step": 1151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050554149200797e-05,
+      "loss": 0.2273,
+      "step": 1152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036110181909466e-05,
+      "loss": 0.2429,
+      "step": 1153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021666139280697e-05,
+      "loss": 0.2335,
+      "step": 1154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0007222051449406e-05,
+      "loss": 0.2446,
+      "step": 1155
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.992777948550599e-06,
+      "loss": 0.2298,
+      "step": 1156
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978333860719307e-06,
+      "loss": 0.2329,
+      "step": 1157
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963889818090539e-06,
+      "loss": 0.225,
+      "step": 1158
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949445850799206e-06,
+      "loss": 0.236,
+      "step": 1159
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93500198898006e-06,
+      "loss": 0.2301,
+      "step": 1160
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920558262767633e-06,
+      "loss": 0.2317,
+      "step": 1161
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90611470229618e-06,
+      "loss": 0.2342,
+      "step": 1162
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891671337699603e-06,
+      "loss": 0.2261,
+      "step": 1163
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877228199111399e-06,
+      "loss": 0.2244,
+      "step": 1164
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.862785316664587e-06,
+      "loss": 0.2318,
+      "step": 1165
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848342720491666e-06,
+      "loss": 0.2311,
+      "step": 1166
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833900440724526e-06,
+      "loss": 0.2314,
+      "step": 1167
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819458507494395e-06,
+      "loss": 0.2289,
+      "step": 1168
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805016950931791e-06,
+      "loss": 0.2266,
+      "step": 1169
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790575801166432e-06,
+      "loss": 0.2326,
+      "step": 1170
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.776135088327195e-06,
+      "loss": 0.2392,
+      "step": 1171
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761694842542042e-06,
+      "loss": 0.2289,
+      "step": 1172
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747255093937966e-06,
+      "loss": 0.2364,
+      "step": 1173
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732815872640915e-06,
+      "loss": 0.232,
+      "step": 1174
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718377208775744e-06,
+      "loss": 0.2328,
+      "step": 1175
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703939132466138e-06,
+      "loss": 0.2291,
+      "step": 1176
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689501673834566e-06,
+      "loss": 0.2322,
+      "step": 1177
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675064863002196e-06,
+      "loss": 0.2422,
+      "step": 1178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660628730088858e-06,
+      "loss": 0.2253,
+      "step": 1179
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646193305212949e-06,
+      "loss": 0.2368,
+      "step": 1180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631758618491415e-06,
+      "loss": 0.2341,
+      "step": 1181
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617324700039635e-06,
+      "loss": 0.2297,
+      "step": 1182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.602891579971405e-06,
+      "loss": 0.2384,
+      "step": 1183
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588459288398841e-06,
+      "loss": 0.2433,
+      "step": 1184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574027855432347e-06,
+      "loss": 0.2262,
+      "step": 1185
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559597311180515e-06,
+      "loss": 0.2341,
+      "step": 1186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.545167685750098e-06,
+      "loss": 0.2246,
+      "step": 1187
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.530739009245934e-06,
+      "loss": 0.2175,
+      "step": 1188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516311311770867e-06,
+      "loss": 0.2318,
+      "step": 1189
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50188462342571e-06,
+      "loss": 0.2359,
+      "step": 1190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487458974309163e-06,
+      "loss": 0.2252,
+      "step": 1191
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.473034394517769e-06,
+      "loss": 0.221,
+      "step": 1192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458610914145826e-06,
+      "loss": 0.236,
+      "step": 1193
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.444188563285352e-06,
+      "loss": 0.2387,
+      "step": 1194
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429767372025992e-06,
+      "loss": 0.2343,
+      "step": 1195
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415347370454992e-06,
+      "loss": 0.2322,
+      "step": 1196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4009285886571e-06,
+      "loss": 0.2283,
+      "step": 1197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386511056714527e-06,
+      "loss": 0.2366,
+      "step": 1198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.2356,
+      "step": 1199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.35767986271106e-06,
+      "loss": 0.2316,
+      "step": 1200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343266260801297e-06,
+      "loss": 0.2278,
+      "step": 1201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.328854029048985e-06,
+      "loss": 0.23,
+      "step": 1202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314443197522658e-06,
+      "loss": 0.2321,
+      "step": 1203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300033796287948e-06,
+      "loss": 0.228,
+      "step": 1204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.285625855407485e-06,
+      "loss": 0.2217,
+      "step": 1205
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271219404940862e-06,
+      "loss": 0.2357,
+      "step": 1206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.256814474944566e-06,
+      "loss": 0.2338,
+      "step": 1207
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242411095471897e-06,
+      "loss": 0.2354,
+      "step": 1208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228009296572937e-06,
+      "loss": 0.2427,
+      "step": 1209
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213609108294456e-06,
+      "loss": 0.2281,
+      "step": 1210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199210560679877e-06,
+      "loss": 0.2274,
+      "step": 1211
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.184813683769188e-06,
+      "loss": 0.2327,
+      "step": 1212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170418507598902e-06,
+      "loss": 0.2227,
+      "step": 1213
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156025062201971e-06,
+      "loss": 0.2251,
+      "step": 1214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.141633377607753e-06,
+      "loss": 0.228,
+      "step": 1215
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127243483841915e-06,
+      "loss": 0.222,
+      "step": 1216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.112855410926404e-06,
+      "loss": 0.2229,
+      "step": 1217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098469188879348e-06,
+      "loss": 0.2426,
+      "step": 1218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.084084847715037e-06,
+      "loss": 0.2245,
+      "step": 1219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.069702417443821e-06,
+      "loss": 0.2278,
+      "step": 1220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055321928072068e-06,
+      "loss": 0.2266,
+      "step": 1221
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040943409602092e-06,
+      "loss": 0.2352,
+      "step": 1222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026566892032104e-06,
+      "loss": 0.2385,
+      "step": 1223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012192405356134e-06,
+      "loss": 0.2334,
+      "step": 1224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.997819979563982e-06,
+      "loss": 0.2204,
+      "step": 1225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.983449644641131e-06,
+      "loss": 0.2345,
+      "step": 1226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969081430568724e-06,
+      "loss": 0.2285,
+      "step": 1227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.2262,
+      "step": 1228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940351484877576e-06,
+      "loss": 0.228,
+      "step": 1229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92598981319873e-06,
+      "loss": 0.2369,
+      "step": 1230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91163038224998e-06,
+      "loss": 0.219,
+      "step": 1231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897273221989715e-06,
+      "loss": 0.2308,
+      "step": 1232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882918362371573e-06,
+      "loss": 0.2312,
+      "step": 1233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.868565833344411e-06,
+      "loss": 0.2293,
+      "step": 1234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854215664852207e-06,
+      "loss": 0.2284,
+      "step": 1235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.839867886834022e-06,
+      "loss": 0.2317,
+      "step": 1236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.825522529223921e-06,
+      "loss": 0.2278,
+      "step": 1237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811179621950937e-06,
+      "loss": 0.2243,
+      "step": 1238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796839194938973e-06,
+      "loss": 0.2312,
+      "step": 1239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782501278106767e-06,
+      "loss": 0.2296,
+      "step": 1240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768165901367811e-06,
+      "loss": 0.2319,
+      "step": 1241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.753833094630317e-06,
+      "loss": 0.2271,
+      "step": 1242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739502887797108e-06,
+      "loss": 0.2275,
+      "step": 1243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725175310765606e-06,
+      "loss": 0.2245,
+      "step": 1244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.710850393427729e-06,
+      "loss": 0.2273,
+      "step": 1245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69652816566986e-06,
+      "loss": 0.2308,
+      "step": 1246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682208657372767e-06,
+      "loss": 0.2179,
+      "step": 1247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66789189841153e-06,
+      "loss": 0.2265,
+      "step": 1248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.653577918655522e-06,
+      "loss": 0.2331,
+      "step": 1249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63926674796829e-06,
+      "loss": 0.2177,
+      "step": 1250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624958416207534e-06,
+      "loss": 0.2166,
+      "step": 1251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610652953225025e-06,
+      "loss": 0.231,
+      "step": 1252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.596350388866557e-06,
+      "loss": 0.2306,
+      "step": 1253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58205075297187e-06,
+      "loss": 0.2335,
+      "step": 1254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567754075374596e-06,
+      "loss": 0.2245,
+      "step": 1255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55346038590219e-06,
+      "loss": 0.2243,
+      "step": 1256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539169714375885e-06,
+      "loss": 0.2238,
+      "step": 1257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.524882090610607e-06,
+      "loss": 0.2263,
+      "step": 1258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510597544414927e-06,
+      "loss": 0.233,
+      "step": 1259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496316105590992e-06,
+      "loss": 0.2323,
+      "step": 1260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.482037803934477e-06,
+      "loss": 0.2292,
+      "step": 1261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467762669234496e-06,
+      "loss": 0.2281,
+      "step": 1262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.453490731273566e-06,
+      "loss": 0.2232,
+      "step": 1263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.43922201982753e-06,
+      "loss": 0.2243,
+      "step": 1264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424956564665508e-06,
+      "loss": 0.2299,
+      "step": 1265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.410694395549815e-06,
+      "loss": 0.2262,
+      "step": 1266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39643554223591e-06,
+      "loss": 0.2356,
+      "step": 1267
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 0.2278,
+      "step": 1268
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3679279020007e-06,
+      "loss": 0.2335,
+      "step": 1269
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35367917455548e-06,
+      "loss": 0.2335,
+      "step": 1270
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.339433881864106e-06,
+      "loss": 0.2318,
+      "step": 1271
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325192053646843e-06,
+      "loss": 0.2305,
+      "step": 1272
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.310953719616712e-06,
+      "loss": 0.238,
+      "step": 1273
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29671890947945e-06,
+      "loss": 0.221,
+      "step": 1274
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28248765293344e-06,
+      "loss": 0.2264,
+      "step": 1275
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.268259979669656e-06,
+      "loss": 0.2266,
+      "step": 1276
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25403591937159e-06,
+      "loss": 0.2245,
+      "step": 1277
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.239815501715202e-06,
+      "loss": 0.2259,
+      "step": 1278
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225598756368844e-06,
+      "loss": 0.2329,
+      "step": 1279
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211385712993219e-06,
+      "loss": 0.2341,
+      "step": 1280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197176401241292e-06,
+      "loss": 0.2267,
+      "step": 1281
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.182970850758255e-06,
+      "loss": 0.2333,
+      "step": 1282
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168769091181439e-06,
+      "loss": 0.2227,
+      "step": 1283
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154571152140287e-06,
+      "loss": 0.2196,
+      "step": 1284
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.140377063256249e-06,
+      "loss": 0.2194,
+      "step": 1285
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126186854142752e-06,
+      "loss": 0.2255,
+      "step": 1286
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.112000554405137e-06,
+      "loss": 0.2207,
+      "step": 1287
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097818193640571e-06,
+      "loss": 0.2343,
+      "step": 1288
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083639801438021e-06,
+      "loss": 0.233,
+      "step": 1289
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069465407378155e-06,
+      "loss": 0.2202,
+      "step": 1290
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055295041033324e-06,
+      "loss": 0.2279,
+      "step": 1291
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.041128731967445e-06,
+      "loss": 0.2351,
+      "step": 1292
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026966509736001e-06,
+      "loss": 0.2305,
+      "step": 1293
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.012808403885927e-06,
+      "loss": 0.2267,
+      "step": 1294
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.998654443955586e-06,
+      "loss": 0.2242,
+      "step": 1295
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984504659474673e-06,
+      "loss": 0.2264,
+      "step": 1296
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97035907996419e-06,
+      "loss": 0.2245,
+      "step": 1297
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.956217734936353e-06,
+      "loss": 0.2261,
+      "step": 1298
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942080653894551e-06,
+      "loss": 0.2207,
+      "step": 1299
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92794786633327e-06,
+      "loss": 0.2257,
+      "step": 1300
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.91381940173805e-06,
+      "loss": 0.2291,
+      "step": 1301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899695289585397e-06,
+      "loss": 0.229,
+      "step": 1302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885575559342751e-06,
+      "loss": 0.2238,
+      "step": 1303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871460240468395e-06,
+      "loss": 0.2215,
+      "step": 1304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857349362411421e-06,
+      "loss": 0.2255,
+      "step": 1305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843242954611652e-06,
+      "loss": 0.2254,
+      "step": 1306
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829141046499583e-06,
+      "loss": 0.2286,
+      "step": 1307
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815043667496323e-06,
+      "loss": 0.2299,
+      "step": 1308
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80095084701353e-06,
+      "loss": 0.2257,
+      "step": 1309
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.786862614453356e-06,
+      "loss": 0.2215,
+      "step": 1310
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772778999208369e-06,
+      "loss": 0.2285,
+      "step": 1311
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758700030661523e-06,
+      "loss": 0.2198,
+      "step": 1312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74462573818606e-06,
+      "loss": 0.2342,
+      "step": 1313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730556151145477e-06,
+      "loss": 0.2234,
+      "step": 1314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716491298893443e-06,
+      "loss": 0.2295,
+      "step": 1315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702431210773767e-06,
+      "loss": 0.235,
+      "step": 1316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688375916120298e-06,
+      "loss": 0.2184,
+      "step": 1317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674325444256899e-06,
+      "loss": 0.2297,
+      "step": 1318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660279824497359e-06,
+      "loss": 0.2238,
+      "step": 1319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.646239086145361e-06,
+      "loss": 0.2284,
+      "step": 1320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632203258494385e-06,
+      "loss": 0.2276,
+      "step": 1321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61817237082768e-06,
+      "loss": 0.2349,
+      "step": 1322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604146452418175e-06,
+      "loss": 0.2245,
+      "step": 1323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.590125532528448e-06,
+      "loss": 0.2224,
+      "step": 1324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5761096404106335e-06,
+      "loss": 0.2225,
+      "step": 1325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562098805306385e-06,
+      "loss": 0.2157,
+      "step": 1326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.548093056446806e-06,
+      "loss": 0.2274,
+      "step": 1327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534092423052382e-06,
+      "loss": 0.2158,
+      "step": 1328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.520096934332931e-06,
+      "loss": 0.2274,
+      "step": 1329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506106619487533e-06,
+      "loss": 0.2179,
+      "step": 1330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492121507704484e-06,
+      "loss": 0.2195,
+      "step": 1331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478141628161211e-06,
+      "loss": 0.2223,
+      "step": 1332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464167010024236e-06,
+      "loss": 0.2267,
+      "step": 1333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450197682449093e-06,
+      "loss": 0.2252,
+      "step": 1334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436233674580293e-06,
+      "loss": 0.221,
+      "step": 1335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422275015551234e-06,
+      "loss": 0.2219,
+      "step": 1336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408321734484165e-06,
+      "loss": 0.2315,
+      "step": 1337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394373860490105e-06,
+      "loss": 0.227,
+      "step": 1338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.380431422668803e-06,
+      "loss": 0.2351,
+      "step": 1339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.366494450108659e-06,
+      "loss": 0.2264,
+      "step": 1340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352562971886673e-06,
+      "loss": 0.2289,
+      "step": 1341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338637017068377e-06,
+      "loss": 0.222,
+      "step": 1342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.324716614707794e-06,
+      "loss": 0.2235,
+      "step": 1343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310801793847344e-06,
+      "loss": 0.2278,
+      "step": 1344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.296892583517819e-06,
+      "loss": 0.2331,
+      "step": 1345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2829890127382866e-06,
+      "loss": 0.2281,
+      "step": 1346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26909111051607e-06,
+      "loss": 0.2234,
+      "step": 1347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2551989058466545e-06,
+      "loss": 0.2297,
+      "step": 1348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241312427713632e-06,
+      "loss": 0.2205,
+      "step": 1349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227431705088665e-06,
+      "loss": 0.2272,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213556766931389e-06,
+      "loss": 0.2258,
+      "step": 1351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199687642189388e-06,
+      "loss": 0.2253,
+      "step": 1352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185824359798099e-06,
+      "loss": 0.2227,
+      "step": 1353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171966948680792e-06,
+      "loss": 0.226,
+      "step": 1354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158115437748467e-06,
+      "loss": 0.2235,
+      "step": 1355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.14426985589983e-06,
+      "loss": 0.2264,
+      "step": 1356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.130430232021204e-06,
+      "loss": 0.2124,
+      "step": 1357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1165965949864934e-06,
+      "loss": 0.2217,
+      "step": 1358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102768973657104e-06,
+      "loss": 0.227,
+      "step": 1359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088947396881896e-06,
+      "loss": 0.2244,
+      "step": 1360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0751318934971116e-06,
+      "loss": 0.2234,
+      "step": 1361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.061322492326333e-06,
+      "loss": 0.2283,
+      "step": 1362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0475192221804014e-06,
+      "loss": 0.226,
+      "step": 1363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.033722111857373e-06,
+      "loss": 0.218,
+      "step": 1364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019931190142444e-06,
+      "loss": 0.2242,
+      "step": 1365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00614648580791e-06,
+      "loss": 0.223,
+      "step": 1366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992368027613092e-06,
+      "loss": 0.2248,
+      "step": 1367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978595844304272e-06,
+      "loss": 0.2179,
+      "step": 1368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964829964614654e-06,
+      "loss": 0.2176,
+      "step": 1369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951070417264278e-06,
+      "loss": 0.2313,
+      "step": 1370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937317230959979e-06,
+      "loss": 0.2224,
+      "step": 1371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.923570434395315e-06,
+      "loss": 0.2323,
+      "step": 1372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.2169,
+      "step": 1373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896096125192449e-06,
+      "loss": 0.2268,
+      "step": 1374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882368669874473e-06,
+      "loss": 0.2134,
+      "step": 1375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868647718936474e-06,
+      "loss": 0.2299,
+      "step": 1376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8549333010047694e-06,
+      "loss": 0.2252,
+      "step": 1377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.841225444692032e-06,
+      "loss": 0.2172,
+      "step": 1378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827524178597257e-06,
+      "loss": 0.2202,
+      "step": 1379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.813829531305676e-06,
+      "loss": 0.2197,
+      "step": 1380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800141531388731e-06,
+      "loss": 0.2161,
+      "step": 1381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7864602074039775e-06,
+      "loss": 0.2232,
+      "step": 1382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772785587895055e-06,
+      "loss": 0.2156,
+      "step": 1383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.759117701391602e-06,
+      "loss": 0.2234,
+      "step": 1384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745456576409227e-06,
+      "loss": 0.2194,
+      "step": 1385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7318022414494235e-06,
+      "loss": 0.2243,
+      "step": 1386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718154724999511e-06,
+      "loss": 0.2185,
+      "step": 1387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704514055532597e-06,
+      "loss": 0.2301,
+      "step": 1388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690880261507496e-06,
+      "loss": 0.226,
+      "step": 1389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.677253371368681e-06,
+      "loss": 0.2194,
+      "step": 1390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663633413546215e-06,
+      "loss": 0.2253,
+      "step": 1391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650020416455713e-06,
+      "loss": 0.222,
+      "step": 1392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.636414408498249e-06,
+      "loss": 0.2261,
+      "step": 1393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62281541806033e-06,
+      "loss": 0.2229,
+      "step": 1394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6092234735138094e-06,
+      "loss": 0.219,
+      "step": 1395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595638603215857e-06,
+      "loss": 0.2307,
+      "step": 1396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5820608355088664e-06,
+      "loss": 0.2179,
+      "step": 1397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.568490198720426e-06,
+      "loss": 0.2302,
+      "step": 1398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554926721163235e-06,
+      "loss": 0.2182,
+      "step": 1399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541370431135073e-06,
+      "loss": 0.2279,
+      "step": 1400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5278213569187025e-06,
+      "loss": 0.2292,
+      "step": 1401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5142795267818505e-06,
+      "loss": 0.2259,
+      "step": 1402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500744968977116e-06,
+      "loss": 0.2216,
+      "step": 1403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4872177117419415e-06,
+      "loss": 0.2188,
+      "step": 1404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47369778329852e-06,
+      "loss": 0.2245,
+      "step": 1405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460185211853767e-06,
+      "loss": 0.2228,
+      "step": 1406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446680025599248e-06,
+      "loss": 0.2306,
+      "step": 1407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433182252711113e-06,
+      "loss": 0.2153,
+      "step": 1408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419691921350055e-06,
+      "loss": 0.2234,
+      "step": 1409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40620905966123e-06,
+      "loss": 0.2267,
+      "step": 1410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392733695774225e-06,
+      "loss": 0.2275,
+      "step": 1411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37926585780297e-06,
+      "loss": 0.2305,
+      "step": 1412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365805573845701e-06,
+      "loss": 0.2252,
+      "step": 1413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35235287198489e-06,
+      "loss": 0.2252,
+      "step": 1414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338907780287198e-06,
+      "loss": 0.2245,
+      "step": 1415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325470326803398e-06,
+      "loss": 0.2214,
+      "step": 1416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312040539568337e-06,
+      "loss": 0.2299,
+      "step": 1417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 0.2216,
+      "step": 1418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2852040759037605e-06,
+      "loss": 0.2234,
+      "step": 1419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271797455463726e-06,
+      "loss": 0.2172,
+      "step": 1420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258398613251275e-06,
+      "loss": 0.2174,
+      "step": 1421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245007577220686e-06,
+      "loss": 0.2229,
+      "step": 1422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.231624375309971e-06,
+      "loss": 0.2187,
+      "step": 1423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218249035440773e-06,
+      "loss": 0.2166,
+      "step": 1424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204881585518354e-06,
+      "loss": 0.2316,
+      "step": 1425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191522053431508e-06,
+      "loss": 0.2149,
+      "step": 1426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178170467052504e-06,
+      "loss": 0.2183,
+      "step": 1427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.164826854237042e-06,
+      "loss": 0.2233,
+      "step": 1428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151491242824177e-06,
+      "loss": 0.2241,
+      "step": 1429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138163660636285e-06,
+      "loss": 0.2224,
+      "step": 1430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124844135478971e-06,
+      "loss": 0.2177,
+      "step": 1431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11153269514105e-06,
+      "loss": 0.2137,
+      "step": 1432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.098229367394455e-06,
+      "loss": 0.2264,
+      "step": 1433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084934179994206e-06,
+      "loss": 0.2247,
+      "step": 1434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071647160678323e-06,
+      "loss": 0.2169,
+      "step": 1435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058368337167805e-06,
+      "loss": 0.2182,
+      "step": 1436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045097737166533e-06,
+      "loss": 0.219,
+      "step": 1437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0318353883612465e-06,
+      "loss": 0.2246,
+      "step": 1438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0185813184214544e-06,
+      "loss": 0.2184,
+      "step": 1439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005335554999414e-06,
+      "loss": 0.2164,
+      "step": 1440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992098125730031e-06,
+      "loss": 0.2222,
+      "step": 1441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978869058230841e-06,
+      "loss": 0.2115,
+      "step": 1442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965648380101916e-06,
+      "loss": 0.2187,
+      "step": 1443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9524361189258476e-06,
+      "loss": 0.2201,
+      "step": 1444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939232302267646e-06,
+      "loss": 0.2216,
+      "step": 1445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926036957674715e-06,
+      "loss": 0.2206,
+      "step": 1446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912850112676784e-06,
+      "loss": 0.2173,
+      "step": 1447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8996717947858395e-06,
+      "loss": 0.2222,
+      "step": 1448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886502031496091e-06,
+      "loss": 0.2227,
+      "step": 1449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.873340850283884e-06,
+      "loss": 0.2247,
+      "step": 1450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8601882786076795e-06,
+      "loss": 0.2185,
+      "step": 1451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.847044343907956e-06,
+      "loss": 0.2235,
+      "step": 1452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833909073607187e-06,
+      "loss": 0.2252,
+      "step": 1453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82078249510976e-06,
+      "loss": 0.223,
+      "step": 1454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.807664635801935e-06,
+      "loss": 0.2164,
+      "step": 1455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794555523051777e-06,
+      "loss": 0.2179,
+      "step": 1456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781455184209106e-06,
+      "loss": 0.2187,
+      "step": 1457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768363646605435e-06,
+      "loss": 0.2097,
+      "step": 1458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755280937553917e-06,
+      "loss": 0.225,
+      "step": 1459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742207084349274e-06,
+      "loss": 0.2294,
+      "step": 1460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729142114267775e-06,
+      "loss": 0.2209,
+      "step": 1461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716086054567134e-06,
+      "loss": 0.2152,
+      "step": 1462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7030389324864845e-06,
+      "loss": 0.2157,
+      "step": 1463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.690000775246315e-06,
+      "loss": 0.2176,
+      "step": 1464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.676971610048406e-06,
+      "loss": 0.2205,
+      "step": 1465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663951464075781e-06,
+      "loss": 0.2155,
+      "step": 1466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.650940364492643e-06,
+      "loss": 0.225,
+      "step": 1467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637938338444325e-06,
+      "loss": 0.2187,
+      "step": 1468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624945413057229e-06,
+      "loss": 0.2209,
+      "step": 1469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6119616154387715e-06,
+      "loss": 0.2177,
+      "step": 1470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598986972677312e-06,
+      "loss": 0.2179,
+      "step": 1471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.586021511842136e-06,
+      "loss": 0.2228,
+      "step": 1472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573065259983347e-06,
+      "loss": 0.2201,
+      "step": 1473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.560118244131852e-06,
+      "loss": 0.2203,
+      "step": 1474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547180491299278e-06,
+      "loss": 0.2173,
+      "step": 1475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534252028477938e-06,
+      "loss": 0.2251,
+      "step": 1476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521332882640753e-06,
+      "loss": 0.2119,
+      "step": 1477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5084230807412135e-06,
+      "loss": 0.2206,
+      "step": 1478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4955226497133e-06,
+      "loss": 0.2198,
+      "step": 1479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48263161647147e-06,
+      "loss": 0.2185,
+      "step": 1480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4697500079105445e-06,
+      "loss": 0.211,
+      "step": 1481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456877850905701e-06,
+      "loss": 0.2285,
+      "step": 1482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444015172312391e-06,
+      "loss": 0.2213,
+      "step": 1483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431161998966292e-06,
+      "loss": 0.2159,
+      "step": 1484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418318357683252e-06,
+      "loss": 0.2264,
+      "step": 1485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4054842752592285e-06,
+      "loss": 0.2308,
+      "step": 1486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.392659778470239e-06,
+      "loss": 0.2181,
+      "step": 1487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379844894072303e-06,
+      "loss": 0.2197,
+      "step": 1488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367039648801386e-06,
+      "loss": 0.2223,
+      "step": 1489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.35424406937333e-06,
+      "loss": 0.2206,
+      "step": 1490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3414581824838385e-06,
+      "loss": 0.2179,
+      "step": 1491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.328682014808364e-06,
+      "loss": 0.2164,
+      "step": 1492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3159155930021e-06,
+      "loss": 0.2147,
+      "step": 1493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3031589436999e-06,
+      "loss": 0.2159,
+      "step": 1494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290412093516232e-06,
+      "loss": 0.2145,
+      "step": 1495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.277675069045116e-06,
+      "loss": 0.219,
+      "step": 1496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.264947896860081e-06,
+      "loss": 0.2217,
+      "step": 1497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252230603514082e-06,
+      "loss": 0.2128,
+      "step": 1498
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239523215539492e-06,
+      "loss": 0.2183,
+      "step": 1499
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22682575944799e-06,
+      "loss": 0.2245,
+      "step": 1500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21413826173055e-06,
+      "loss": 0.2191,
+      "step": 1501
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201460748857369e-06,
+      "loss": 0.2248,
+      "step": 1502
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1887932472778056e-06,
+      "loss": 0.2227,
+      "step": 1503
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176135783420339e-06,
+      "loss": 0.2096,
+      "step": 1504
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163488383692499e-06,
+      "loss": 0.2176,
+      "step": 1505
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150851074480826e-06,
+      "loss": 0.2134,
+      "step": 1506
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138223882150803e-06,
+      "loss": 0.2191,
+      "step": 1507
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12560683304681e-06,
+      "loss": 0.2183,
+      "step": 1508
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112999953492055e-06,
+      "loss": 0.2234,
+      "step": 1509
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.100403269788547e-06,
+      "loss": 0.2233,
+      "step": 1510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087816808217005e-06,
+      "loss": 0.2173,
+      "step": 1511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0752405950368325e-06,
+      "loss": 0.22,
+      "step": 1512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062674656486049e-06,
+      "loss": 0.2142,
+      "step": 1513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050119018781232e-06,
+      "loss": 0.2149,
+      "step": 1514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.03757370811748e-06,
+      "loss": 0.2211,
+      "step": 1515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025038750668338e-06,
+      "loss": 0.2144,
+      "step": 1516
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012514172585744e-06,
+      "loss": 0.2125,
+      "step": 1517
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.2146,
+      "step": 1518
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987496259019684e-06,
+      "loss": 0.2207,
+      "step": 1519
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975002975731613e-06,
+      "loss": 0.2205,
+      "step": 1520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9625201762007865e-06,
+      "loss": 0.2126,
+      "step": 1521
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.950047886470336e-06,
+      "loss": 0.221,
+      "step": 1522
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93758613256146e-06,
+      "loss": 0.2143,
+      "step": 1523
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.925134940473383e-06,
+      "loss": 0.2196,
+      "step": 1524
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912694336183278e-06,
+      "loss": 0.222,
+      "step": 1525
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.900264345646253e-06,
+      "loss": 0.2119,
+      "step": 1526
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887844994795257e-06,
+      "loss": 0.227,
+      "step": 1527
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875436309541033e-06,
+      "loss": 0.2173,
+      "step": 1528
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8630383157720985e-06,
+      "loss": 0.2228,
+      "step": 1529
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8506510393546365e-06,
+      "loss": 0.2132,
+      "step": 1530
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838274506132489e-06,
+      "loss": 0.2205,
+      "step": 1531
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825908741927076e-06,
+      "loss": 0.217,
+      "step": 1532
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.813553772537354e-06,
+      "loss": 0.219,
+      "step": 1533
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801209623739754e-06,
+      "loss": 0.2167,
+      "step": 1534
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78887632128814e-06,
+      "loss": 0.2191,
+      "step": 1535
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776553890913729e-06,
+      "loss": 0.213,
+      "step": 1536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764242358325081e-06,
+      "loss": 0.2115,
+      "step": 1537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751941749207996e-06,
+      "loss": 0.2183,
+      "step": 1538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739652089225497e-06,
+      "loss": 0.2123,
+      "step": 1539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.727373404017762e-06,
+      "loss": 0.2051,
+      "step": 1540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715105719202071e-06,
+      "loss": 0.2207,
+      "step": 1541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.702849060372752e-06,
+      "loss": 0.2104,
+      "step": 1542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690603453101134e-06,
+      "loss": 0.2177,
+      "step": 1543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678368922935478e-06,
+      "loss": 0.2096,
+      "step": 1544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.66614549540095e-06,
+      "loss": 0.2125,
+      "step": 1545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653933195999546e-06,
+      "loss": 0.2205,
+      "step": 1546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.641732050210032e-06,
+      "loss": 0.2103,
+      "step": 1547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629542083487931e-06,
+      "loss": 0.2128,
+      "step": 1548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617363321265416e-06,
+      "loss": 0.2124,
+      "step": 1549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6051957889513e-06,
+      "loss": 0.2237,
+      "step": 1550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593039511930963e-06,
+      "loss": 0.2219,
+      "step": 1551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5808945155662984e-06,
+      "loss": 0.2096,
+      "step": 1552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568760825195671e-06,
+      "loss": 0.2107,
+      "step": 1553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556638466133859e-06,
+      "loss": 0.2173,
+      "step": 1554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.544527463671984e-06,
+      "loss": 0.2167,
+      "step": 1555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5324278430775e-06,
+      "loss": 0.214,
+      "step": 1556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52033962959409e-06,
+      "loss": 0.2165,
+      "step": 1557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.508262848441651e-06,
+      "loss": 0.2213,
+      "step": 1558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.496197524816229e-06,
+      "loss": 0.216,
+      "step": 1559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484143683889964e-06,
+      "loss": 0.217,
+      "step": 1560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47210135081103e-06,
+      "loss": 0.2161,
+      "step": 1561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4600705507036125e-06,
+      "loss": 0.2177,
+      "step": 1562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.44805130866781e-06,
+      "loss": 0.2336,
+      "step": 1563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436043649779637e-06,
+      "loss": 0.2161,
+      "step": 1564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.42404759909091e-06,
+      "loss": 0.2222,
+      "step": 1565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412063181629247e-06,
+      "loss": 0.2154,
+      "step": 1566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400090422397999e-06,
+      "loss": 0.2225,
+      "step": 1567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 0.22,
+      "step": 1568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.376179978518429e-06,
+      "loss": 0.2125,
+      "step": 1569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.364242343754972e-06,
+      "loss": 0.2216,
+      "step": 1570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3523164669915495e-06,
+      "loss": 0.2206,
+      "step": 1571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3404023731093604e-06,
+      "loss": 0.2144,
+      "step": 1572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328500086965041e-06,
+      "loss": 0.2113,
+      "step": 1573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3166096333905745e-06,
+      "loss": 0.2204,
+      "step": 1574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304731037193267e-06,
+      "loss": 0.2168,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.2192,
+      "step": 1576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.281009516035602e-06,
+      "loss": 0.2135,
+      "step": 1577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269166640565953e-06,
+      "loss": 0.2139,
+      "step": 1578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257335721454785e-06,
+      "loss": 0.2259,
+      "step": 1579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2455167833851804e-06,
+      "loss": 0.2182,
+      "step": 1580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233709851015254e-06,
+      "loss": 0.21,
+      "step": 1581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221914948978047e-06,
+      "loss": 0.2139,
+      "step": 1582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210132101881516e-06,
+      "loss": 0.2238,
+      "step": 1583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198361334308463e-06,
+      "loss": 0.2161,
+      "step": 1584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.186602670816488e-06,
+      "loss": 0.2235,
+      "step": 1585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.17485613593794e-06,
+      "loss": 0.2154,
+      "step": 1586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1631217541798614e-06,
+      "loss": 0.221,
+      "step": 1587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151399550023941e-06,
+      "loss": 0.2031,
+      "step": 1588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139689547926461e-06,
+      "loss": 0.2118,
+      "step": 1589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1279917723182476e-06,
+      "loss": 0.2183,
+      "step": 1590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116306247604606e-06,
+      "loss": 0.2149,
+      "step": 1591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104632998165309e-06,
+      "loss": 0.2114,
+      "step": 1592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.092972048354491e-06,
+      "loss": 0.2151,
+      "step": 1593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08132342250064e-06,
+      "loss": 0.22,
+      "step": 1594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.069687144906532e-06,
+      "loss": 0.2135,
+      "step": 1595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058063239849178e-06,
+      "loss": 0.2145,
+      "step": 1596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046451731579775e-06,
+      "loss": 0.2147,
+      "step": 1597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034852644323661e-06,
+      "loss": 0.2232,
+      "step": 1598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.023266002280246e-06,
+      "loss": 0.2179,
+      "step": 1599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.011691829623001e-06,
+      "loss": 0.2107,
+      "step": 1600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.000130150499355e-06,
+      "loss": 0.207,
+      "step": 1601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988580989030686e-06,
+      "loss": 0.2123,
+      "step": 1602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977044369312253e-06,
+      "loss": 0.2252,
+      "step": 1603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.965520315413149e-06,
+      "loss": 0.2129,
+      "step": 1604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954008851376252e-06,
+      "loss": 0.2158,
+      "step": 1605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.942510001218171e-06,
+      "loss": 0.2217,
+      "step": 1606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9310237889292e-06,
+      "loss": 0.2155,
+      "step": 1607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919550238473265e-06,
+      "loss": 0.2246,
+      "step": 1608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9080893737878815e-06,
+      "loss": 0.2061,
+      "step": 1609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896641218784081e-06,
+      "loss": 0.2144,
+      "step": 1610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8852057973464045e-06,
+      "loss": 0.2115,
+      "step": 1611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873783133332803e-06,
+      "loss": 0.2119,
+      "step": 1612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862373250574626e-06,
+      "loss": 0.2236,
+      "step": 1613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850976172876551e-06,
+      "loss": 0.202,
+      "step": 1614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839591924016541e-06,
+      "loss": 0.2175,
+      "step": 1615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828220527745796e-06,
+      "loss": 0.2172,
+      "step": 1616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816862007788702e-06,
+      "loss": 0.2046,
+      "step": 1617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8055163878427703e-06,
+      "loss": 0.2182,
+      "step": 1618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.794183691578618e-06,
+      "loss": 0.2163,
+      "step": 1619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7828639426398793e-06,
+      "loss": 0.2197,
+      "step": 1620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771557164643189e-06,
+      "loss": 0.213,
+      "step": 1621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7602633811781165e-06,
+      "loss": 0.2118,
+      "step": 1622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7489826158071196e-06,
+      "loss": 0.2249,
+      "step": 1623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7377148920654973e-06,
+      "loss": 0.2142,
+      "step": 1624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7264602334613385e-06,
+      "loss": 0.2149,
+      "step": 1625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715218663475476e-06,
+      "loss": 0.2148,
+      "step": 1626
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.703990205561434e-06,
+      "loss": 0.2146,
+      "step": 1627
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6927748831453835e-06,
+      "loss": 0.2158,
+      "step": 1628
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6815727196260796e-06,
+      "loss": 0.2035,
+      "step": 1629
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.670383738374844e-06,
+      "loss": 0.2065,
+      "step": 1630
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.659207962735475e-06,
+      "loss": 0.2074,
+      "step": 1631
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.648045416024232e-06,
+      "loss": 0.2168,
+      "step": 1632
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6368961215297726e-06,
+      "loss": 0.2102,
+      "step": 1633
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.625760102513103e-06,
+      "loss": 0.2015,
+      "step": 1634
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6146373822075366e-06,
+      "loss": 0.2155,
+      "step": 1635
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603527983818641e-06,
+      "loss": 0.2153,
+      "step": 1636
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5924319305241797e-06,
+      "loss": 0.2031,
+      "step": 1637
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581349245474095e-06,
+      "loss": 0.2126,
+      "step": 1638
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.570279951790416e-06,
+      "loss": 0.213,
+      "step": 1639
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5592240725672476e-06,
+      "loss": 0.2193,
+      "step": 1640
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5481816308707052e-06,
+      "loss": 0.211,
+      "step": 1641
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.537152649738864e-06,
+      "loss": 0.2149,
+      "step": 1642
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5261371521817247e-06,
+      "loss": 0.2204,
+      "step": 1643
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515135161181151e-06,
+      "loss": 0.2097,
+      "step": 1644
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5041466996908203e-06,
+      "loss": 0.2123,
+      "step": 1645
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.493171790636203e-06,
+      "loss": 0.2149,
+      "step": 1646
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482210456914481e-06,
+      "loss": 0.2056,
+      "step": 1647
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4712627213945084e-06,
+      "loss": 0.2145,
+      "step": 1648
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4603286069167873e-06,
+      "loss": 0.2071,
+      "step": 1649
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449408136293383e-06,
+      "loss": 0.2059,
+      "step": 1650
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4385013323079066e-06,
+      "loss": 0.2089,
+      "step": 1651
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4276082177154536e-06,
+      "loss": 0.2023,
+      "step": 1652
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416728815242557e-06,
+      "loss": 0.2095,
+      "step": 1653
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.405863147587146e-06,
+      "loss": 0.2079,
+      "step": 1654
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.395011237418494e-06,
+      "loss": 0.21,
+      "step": 1655
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38417310737716e-06,
+      "loss": 0.202,
+      "step": 1656
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3733487800749775e-06,
+      "loss": 0.2042,
+      "step": 1657
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625382780949576e-06,
+      "loss": 0.212,
+      "step": 1658
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3517416239912826e-06,
+      "loss": 0.2162,
+      "step": 1659
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340958840289239e-06,
+      "loss": 0.215,
+      "step": 1660
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3301899494851764e-06,
+      "loss": 0.2099,
+      "step": 1661
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319434974046458e-06,
+      "loss": 0.2233,
+      "step": 1662
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.2042,
+      "step": 1663
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2979668589893064e-06,
+      "loss": 0.2086,
+      "step": 1664
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2872537641602542e-06,
+      "loss": 0.2116,
+      "step": 1665
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2765546742752173e-06,
+      "loss": 0.2153,
+      "step": 1666
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2658696116559253e-06,
+      "loss": 0.2173,
+      "step": 1667
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.255198598594862e-06,
+      "loss": 0.2108,
+      "step": 1668
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244541657355176e-06,
+      "loss": 0.2039,
+      "step": 1669
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2338988101706727e-06,
+      "loss": 0.2199,
+      "step": 1670
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2232700792457485e-06,
+      "loss": 0.2078,
+      "step": 1671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2126554867553483e-06,
+      "loss": 0.2132,
+      "step": 1672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2020550548449213e-06,
+      "loss": 0.2113,
+      "step": 1673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1914688056303743e-06,
+      "loss": 0.2193,
+      "step": 1674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1808967611980133e-06,
+      "loss": 0.2135,
+      "step": 1675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1703389436045307e-06,
+      "loss": 0.2166,
+      "step": 1676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1597953748769126e-06,
+      "loss": 0.2143,
+      "step": 1677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1492660770124337e-06,
+      "loss": 0.2116,
+      "step": 1678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1387510719785907e-06,
+      "loss": 0.2032,
+      "step": 1679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1282503817130583e-06,
+      "loss": 0.2125,
+      "step": 1680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1177640281236486e-06,
+      "loss": 0.2143,
+      "step": 1681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.107292033088265e-06,
+      "loss": 0.1964,
+      "step": 1682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.096834418454844e-06,
+      "loss": 0.2155,
+      "step": 1683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.086391206041338e-06,
+      "loss": 0.2104,
+      "step": 1684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075962417635634e-06,
+      "loss": 0.2042,
+      "step": 1685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0655480749955303e-06,
+      "loss": 0.2154,
+      "step": 1686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055148199848702e-06,
+      "loss": 0.2175,
+      "step": 1687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0447628138926153e-06,
+      "loss": 0.2142,
+      "step": 1688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0343919387945265e-06,
+      "loss": 0.205,
+      "step": 1689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024035596191408e-06,
+      "loss": 0.2102,
+      "step": 1690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.013693807689917e-06,
+      "loss": 0.2128,
+      "step": 1691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 0.2119,
+      "step": 1692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9930539792665767e-06,
+      "loss": 0.2154,
+      "step": 1693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9827559824060314e-06,
+      "loss": 0.2183,
+      "step": 1694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.972472625769648e-06,
+      "loss": 0.2127,
+      "step": 1695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9622039308118e-06,
+      "loss": 0.2141,
+      "step": 1696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9519499189562884e-06,
+      "loss": 0.2097,
+      "step": 1697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.941710611596271e-06,
+      "loss": 0.2223,
+      "step": 1698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9314860300942372e-06,
+      "loss": 0.2025,
+      "step": 1699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9212761957819347e-06,
+      "loss": 0.2127,
+      "step": 1700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91108112996037e-06,
+      "loss": 0.2066,
+      "step": 1701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9009008538997142e-06,
+      "loss": 0.21,
+      "step": 1702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8907353888392955e-06,
+      "loss": 0.2093,
+      "step": 1703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8805847559875365e-06,
+      "loss": 0.2198,
+      "step": 1704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8704489765219157e-06,
+      "loss": 0.2113,
+      "step": 1705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8603280715889304e-06,
+      "loss": 0.211,
+      "step": 1706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.850222062304031e-06,
+      "loss": 0.2136,
+      "step": 1707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8401309697515987e-06,
+      "loss": 0.2101,
+      "step": 1708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830054814984895e-06,
+      "loss": 0.2101,
+      "step": 1709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819993619026015e-06,
+      "loss": 0.2105,
+      "step": 1710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8099474028658356e-06,
+      "loss": 0.214,
+      "step": 1711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7999161874640026e-06,
+      "loss": 0.2009,
+      "step": 1712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7898999937488415e-06,
+      "loss": 0.2142,
+      "step": 1713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.779898842617351e-06,
+      "loss": 0.2064,
+      "step": 1714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.769912754935146e-06,
+      "loss": 0.2011,
+      "step": 1715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7599417515364103e-06,
+      "loss": 0.2215,
+      "step": 1716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7499858532238587e-06,
+      "loss": 0.2099,
+      "step": 1717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 0.2128,
+      "step": 1718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7301194549105523e-06,
+      "loss": 0.2122,
+      "step": 1719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7202089963574863e-06,
+      "loss": 0.2059,
+      "step": 1720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7103137257858867e-06,
+      "loss": 0.2119,
+      "step": 1721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7004336638404683e-06,
+      "loss": 0.2166,
+      "step": 1722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6905688311342105e-06,
+      "loss": 0.2201,
+      "step": 1723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6807192482483246e-06,
+      "loss": 0.2128,
+      "step": 1724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.670884935732202e-06,
+      "loss": 0.2075,
+      "step": 1725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6610659141033758e-06,
+      "loss": 0.2099,
+      "step": 1726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6512622038474778e-06,
+      "loss": 0.2123,
+      "step": 1727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641473825418197e-06,
+      "loss": 0.2009,
+      "step": 1728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6317007992372346e-06,
+      "loss": 0.2153,
+      "step": 1729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6219431456942536e-06,
+      "loss": 0.2071,
+      "step": 1730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6122008851468627e-06,
+      "loss": 0.2125,
+      "step": 1731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602474037920539e-06,
+      "loss": 0.2243,
+      "step": 1732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.59276262430861e-06,
+      "loss": 0.2111,
+      "step": 1733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5830666645722034e-06,
+      "loss": 0.2218,
+      "step": 1734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5733861789402047e-06,
+      "loss": 0.1986,
+      "step": 1735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.563721187609214e-06,
+      "loss": 0.209,
+      "step": 1736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.554071710743511e-06,
+      "loss": 0.2133,
+      "step": 1737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.544437768474993e-06,
+      "loss": 0.2099,
+      "step": 1738
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5348193809031695e-06,
+      "loss": 0.2093,
+      "step": 1739
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5252165680950747e-06,
+      "loss": 0.2128,
+      "step": 1740
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5156293500852645e-06,
+      "loss": 0.2124,
+      "step": 1741
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506057746875753e-06,
+      "loss": 0.2117,
+      "step": 1742
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.496501778435977e-06,
+      "loss": 0.2064,
+      "step": 1743
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.486961464702756e-06,
+      "loss": 0.2094,
+      "step": 1744
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4774368255802483e-06,
+      "loss": 0.2112,
+      "step": 1745
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4679278809399076e-06,
+      "loss": 0.2057,
+      "step": 1746
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.458434650620447e-06,
+      "loss": 0.2084,
+      "step": 1747
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4489571544277944e-06,
+      "loss": 0.2029,
+      "step": 1748
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4394954121350433e-06,
+      "loss": 0.2104,
+      "step": 1749
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4300494434824373e-06,
+      "loss": 0.2169,
+      "step": 1750
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.42061926817729e-06,
+      "loss": 0.2102,
+      "step": 1751
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411204905893979e-06,
+      "loss": 0.2121,
+      "step": 1752
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4018063762738863e-06,
+      "loss": 0.2151,
+      "step": 1753
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3924236989253623e-06,
+      "loss": 0.2014,
+      "step": 1754
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3830568934236854e-06,
+      "loss": 0.2108,
+      "step": 1755
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3737059793110216e-06,
+      "loss": 0.2045,
+      "step": 1756
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364370976096372e-06,
+      "loss": 0.2021,
+      "step": 1757
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.355051903255562e-06,
+      "loss": 0.2091,
+      "step": 1758
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3457487802311597e-06,
+      "loss": 0.2137,
+      "step": 1759
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3364616264324725e-06,
+      "loss": 0.2198,
+      "step": 1760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3271904612354823e-06,
+      "loss": 0.2121,
+      "step": 1761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.317935303982819e-06,
+      "loss": 0.203,
+      "step": 1762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.308696173983711e-06,
+      "loss": 0.2102,
+      "step": 1763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299473090513954e-06,
+      "loss": 0.2154,
+      "step": 1764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290266072815851e-06,
+      "loss": 0.2038,
+      "step": 1765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281075140098208e-06,
+      "loss": 0.2105,
+      "step": 1766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2719003115362593e-06,
+      "loss": 0.2013,
+      "step": 1767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2627416062716366e-06,
+      "loss": 0.2076,
+      "step": 1768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.25359904341235e-06,
+      "loss": 0.202,
+      "step": 1769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2444726420327136e-06,
+      "loss": 0.2203,
+      "step": 1770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2353624211733328e-06,
+      "loss": 0.2121,
+      "step": 1771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226268399841055e-06,
+      "loss": 0.207,
+      "step": 1772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217190597008927e-06,
+      "loss": 0.2206,
+      "step": 1773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2081290316161606e-06,
+      "loss": 0.2171,
+      "step": 1774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.199083722568095e-06,
+      "loss": 0.2146,
+      "step": 1775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1900546887361398e-06,
+      "loss": 0.2018,
+      "step": 1776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181041948957768e-06,
+      "loss": 0.2012,
+      "step": 1777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1720455220364443e-06,
+      "loss": 0.2045,
+      "step": 1778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163065426741603e-06,
+      "loss": 0.2077,
+      "step": 1779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154101681808608e-06,
+      "loss": 0.2038,
+      "step": 1780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451543059387094e-06,
+      "loss": 0.213,
+      "step": 1781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1362233177990055e-06,
+      "loss": 0.2112,
+      "step": 1782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1273087360224086e-06,
+      "loss": 0.2143,
+      "step": 1783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1184105792075894e-06,
+      "loss": 0.2046,
+      "step": 1784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1095288659189696e-06,
+      "loss": 0.2041,
+      "step": 1785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.100663614686653e-06,
+      "loss": 0.2125,
+      "step": 1786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0918148440063948e-06,
+      "loss": 0.2025,
+      "step": 1787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0829825723395803e-06,
+      "loss": 0.2095,
+      "step": 1788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0741668181131557e-06,
+      "loss": 0.2057,
+      "step": 1789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.065367599719621e-06,
+      "loss": 0.2097,
+      "step": 1790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0565849355169675e-06,
+      "loss": 0.2212,
+      "step": 1791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047818843828654e-06,
+      "loss": 0.2152,
+      "step": 1792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0390693429435626e-06,
+      "loss": 0.206,
+      "step": 1793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030336451115964e-06,
+      "loss": 0.2076,
+      "step": 1794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021620186565466e-06,
+      "loss": 0.2096,
+      "step": 1795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.012920567477007e-06,
+      "loss": 0.2033,
+      "step": 1796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0042376120007766e-06,
+      "loss": 0.2084,
+      "step": 1797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9955713382522113e-06,
+      "loss": 0.2059,
+      "step": 1798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.986921764311941e-06,
+      "loss": 0.2083,
+      "step": 1799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.978288908225754e-06,
+      "loss": 0.2069,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9696727880045587e-06,
+      "loss": 0.2044,
+      "step": 1801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961073421624352e-06,
+      "loss": 0.2103,
+      "step": 1802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9524908270261655e-06,
+      "loss": 0.2101,
+      "step": 1803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9439250221160576e-06,
+      "loss": 0.2102,
+      "step": 1804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.93537602476504e-06,
+      "loss": 0.2115,
+      "step": 1805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926843852809067e-06,
+      "loss": 0.2121,
+      "step": 1806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9183285240489945e-06,
+      "loss": 0.2077,
+      "step": 1807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.2082,
+      "step": 1808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9013484671441973e-06,
+      "loss": 0.2044,
+      "step": 1809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8928837744253259e-06,
+      "loss": 0.2021,
+      "step": 1810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8844359957539783e-06,
+      "loss": 0.2025,
+      "step": 1811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8760051487549325e-06,
+      "loss": 0.2061,
+      "step": 1812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8675912510176476e-06,
+      "loss": 0.205,
+      "step": 1813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8591943200962059e-06,
+      "loss": 0.2144,
+      "step": 1814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850814373509312e-06,
+      "loss": 0.2152,
+      "step": 1815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8424514287402184e-06,
+      "loss": 0.2061,
+      "step": 1816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8341055032367149e-06,
+      "loss": 0.2023,
+      "step": 1817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8257766144110823e-06,
+      "loss": 0.2005,
+      "step": 1818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8174647796400568e-06,
+      "loss": 0.2111,
+      "step": 1819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.809170016264794e-06,
+      "loss": 0.2097,
+      "step": 1820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8008923415908375e-06,
+      "loss": 0.2043,
+      "step": 1821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792631772888065e-06,
+      "loss": 0.2149,
+      "step": 1822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.784388327390687e-06,
+      "loss": 0.2046,
+      "step": 1823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7761620222971698e-06,
+      "loss": 0.2042,
+      "step": 1824
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7679528747702257e-06,
+      "loss": 0.2091,
+      "step": 1825
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597609019367822e-06,
+      "loss": 0.2034,
+      "step": 1826
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7515861208879147e-06,
+      "loss": 0.2133,
+      "step": 1827
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743428548678847e-06,
+      "loss": 0.2138,
+      "step": 1828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.735288202328891e-06,
+      "loss": 0.2134,
+      "step": 1829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.727165098821425e-06,
+      "loss": 0.2104,
+      "step": 1830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.71905925510385e-06,
+      "loss": 0.2083,
+      "step": 1831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.710970688087561e-06,
+      "loss": 0.2073,
+      "step": 1832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028994146478983e-06,
+      "loss": 0.1984,
+      "step": 1833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.694845451624142e-06,
+      "loss": 0.2043,
+      "step": 1834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6868088158194351e-06,
+      "loss": 0.2046,
+      "step": 1835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6787895240007833e-06,
+      "loss": 0.213,
+      "step": 1836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6707875928990059e-06,
+      "loss": 0.2047,
+      "step": 1837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6628030392087001e-06,
+      "loss": 0.2087,
+      "step": 1838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6548358795882037e-06,
+      "loss": 0.2113,
+      "step": 1839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6468861306595786e-06,
+      "loss": 0.2018,
+      "step": 1840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6389538090085467e-06,
+      "loss": 0.2047,
+      "step": 1841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6310389311844787e-06,
+      "loss": 0.2072,
+      "step": 1842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6231415137003536e-06,
+      "loss": 0.2003,
+      "step": 1843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6152615730327159e-06,
+      "loss": 0.2124,
+      "step": 1844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6073991256216603e-06,
+      "loss": 0.1999,
+      "step": 1845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5995541878707677e-06,
+      "loss": 0.2144,
+      "step": 1846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5917267761471022e-06,
+      "loss": 0.2118,
+      "step": 1847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5839169067811555e-06,
+      "loss": 0.2059,
+      "step": 1848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.576124596066826e-06,
+      "loss": 0.2134,
+      "step": 1849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5683498602613689e-06,
+      "loss": 0.1977,
+      "step": 1850
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5605927155853884e-06,
+      "loss": 0.2034,
+      "step": 1851
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5528531782227718e-06,
+      "loss": 0.2031,
+      "step": 1852
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5451312643206828e-06,
+      "loss": 0.2138,
+      "step": 1853
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5374269899895122e-06,
+      "loss": 0.2068,
+      "step": 1854
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5297403713028502e-06,
+      "loss": 0.201,
+      "step": 1855
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.522071424297451e-06,
+      "loss": 0.203,
+      "step": 1856
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5144201649732037e-06,
+      "loss": 0.2042,
+      "step": 1857
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.506786609293084e-06,
+      "loss": 0.2066,
+      "step": 1858
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4991707731831494e-06,
+      "loss": 0.2039,
+      "step": 1859
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.491572672532473e-06,
+      "loss": 0.208,
+      "step": 1860
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4839923231931342e-06,
+      "loss": 0.2039,
+      "step": 1861
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4764297409801764e-06,
+      "loss": 0.2024,
+      "step": 1862
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4688849416715734e-06,
+      "loss": 0.1975,
+      "step": 1863
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.461357941008199e-06,
+      "loss": 0.2056,
+      "step": 1864
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4538487546937951e-06,
+      "loss": 0.2005,
+      "step": 1865
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.446357398394934e-06,
+      "loss": 0.2053,
+      "step": 1866
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4388838877409906e-06,
+      "loss": 0.2027,
+      "step": 1867
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 0.2049,
+      "step": 1868
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4239904656991633e-06,
+      "loss": 0.2106,
+      "step": 1869
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416570585383742e-06,
+      "loss": 0.1982,
+      "step": 1870
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4091686128580927e-06,
+      "loss": 0.2089,
+      "step": 1871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4017845635651096e-06,
+      "loss": 0.2069,
+      "step": 1872
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3944184529102878e-06,
+      "loss": 0.2114,
+      "step": 1873
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.387070296261701e-06,
+      "loss": 0.1935,
+      "step": 1874
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.379740108949964e-06,
+      "loss": 0.2124,
+      "step": 1875
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3724279062682022e-06,
+      "loss": 0.2128,
+      "step": 1876
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3651337034720113e-06,
+      "loss": 0.2142,
+      "step": 1877
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3578575157794504e-06,
+      "loss": 0.2076,
+      "step": 1878
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3505993583709753e-06,
+      "loss": 0.1991,
+      "step": 1879
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3433592463894375e-06,
+      "loss": 0.2057,
+      "step": 1880
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3361371949400325e-06,
+      "loss": 0.2069,
+      "step": 1881
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3289332190902793e-06,
+      "loss": 0.1985,
+      "step": 1882
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.321747333869986e-06,
+      "loss": 0.2025,
+      "step": 1883
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3145795542712147e-06,
+      "loss": 0.2073,
+      "step": 1884
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074298952482568e-06,
+      "loss": 0.2063,
+      "step": 1885
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3002983717175955e-06,
+      "loss": 0.2023,
+      "step": 1886
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2931849985578804e-06,
+      "loss": 0.2001,
+      "step": 1887
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2860897906098869e-06,
+      "loss": 0.2097,
+      "step": 1888
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2790127626765048e-06,
+      "loss": 0.212,
+      "step": 1889
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2719539295226802e-06,
+      "loss": 0.211,
+      "step": 1890
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2649133058754071e-06,
+      "loss": 0.207,
+      "step": 1891
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2578909064236887e-06,
+      "loss": 0.2047,
+      "step": 1892
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2508867458185037e-06,
+      "loss": 0.2042,
+      "step": 1893
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2439008386727802e-06,
+      "loss": 0.2161,
+      "step": 1894
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2369331995613664e-06,
+      "loss": 0.2025,
+      "step": 1895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2299838430209876e-06,
+      "loss": 0.2002,
+      "step": 1896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2230527835502414e-06,
+      "loss": 0.2181,
+      "step": 1897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2161400356095376e-06,
+      "loss": 0.2063,
+      "step": 1898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2092456136210894e-06,
+      "loss": 0.2081,
+      "step": 1899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2023695319688755e-06,
+      "loss": 0.2006,
+      "step": 1900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1955118049986091e-06,
+      "loss": 0.203,
+      "step": 1901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1886724470177113e-06,
+      "loss": 0.1963,
+      "step": 1902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1818514722952812e-06,
+      "loss": 0.204,
+      "step": 1903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1750488950620543e-06,
+      "loss": 0.2079,
+      "step": 1904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1682647295104e-06,
+      "loss": 0.2036,
+      "step": 1905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161498989794263e-06,
+      "loss": 0.1988,
+      "step": 1906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1547516900291445e-06,
+      "loss": 0.205,
+      "step": 1907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1480228442920848e-06,
+      "loss": 0.2012,
+      "step": 1908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1413124666216124e-06,
+      "loss": 0.2008,
+      "step": 1909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1346205710177304e-06,
+      "loss": 0.2053,
+      "step": 1910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1279471714418833e-06,
+      "loss": 0.2033,
+      "step": 1911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212922818169224e-06,
+      "loss": 0.2059,
+      "step": 1912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1146559160270875e-06,
+      "loss": 0.2124,
+      "step": 1913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1080380879179686e-06,
+      "loss": 0.2055,
+      "step": 1914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1014388112964747e-06,
+      "loss": 0.2083,
+      "step": 1915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0948580999308234e-06,
+      "loss": 0.209,
+      "step": 1916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0882959675504845e-06,
+      "loss": 0.2107,
+      "step": 1917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0817524278461777e-06,
+      "loss": 0.1943,
+      "step": 1918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.075227494469826e-06,
+      "loss": 0.2064,
+      "step": 1919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0687211810345366e-06,
+      "loss": 0.1984,
+      "step": 1920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0622335011145678e-06,
+      "loss": 0.2035,
+      "step": 1921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055764468245304e-06,
+      "loss": 0.2078,
+      "step": 1922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0493140959232206e-06,
+      "loss": 0.2029,
+      "step": 1923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042882397605871e-06,
+      "loss": 0.2035,
+      "step": 1924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0364693867118425e-06,
+      "loss": 0.2024,
+      "step": 1925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.030075076620729e-06,
+      "loss": 0.2067,
+      "step": 1926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0236994806731225e-06,
+      "loss": 0.2031,
+      "step": 1927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0173426121705577e-06,
+      "loss": 0.197,
+      "step": 1928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011004484375504e-06,
+      "loss": 0.2145,
+      "step": 1929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0046851105113331e-06,
+      "loss": 0.208,
+      "step": 1930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.983845037622876e-07,
+      "loss": 0.2116,
+      "step": 1931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921026772734543e-07,
+      "loss": 0.2067,
+      "step": 1932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.858396441507445e-07,
+      "loss": 0.2159,
+      "step": 1933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.795954174608491e-07,
+      "loss": 0.2035,
+      "step": 1934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.733700102312372e-07,
+      "loss": 0.2057,
+      "step": 1935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.671634354501036e-07,
+      "loss": 0.2019,
+      "step": 1936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60975706066356e-07,
+      "loss": 0.2048,
+      "step": 1937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.548068349895845e-07,
+      "loss": 0.2042,
+      "step": 1938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.48656835090036e-07,
+      "loss": 0.2043,
+      "step": 1939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425257191985859e-07,
+      "loss": 0.2026,
+      "step": 1940
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.364135001067098e-07,
+      "loss": 0.209,
+      "step": 1941
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.30320190566456e-07,
+      "loss": 0.2057,
+      "step": 1942
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.242458032904311e-07,
+      "loss": 0.2162,
+      "step": 1943
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.181903509517509e-07,
+      "loss": 0.2056,
+      "step": 1944
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121538461840352e-07,
+      "loss": 0.2058,
+      "step": 1945
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.061363015813763e-07,
+      "loss": 0.1997,
+      "step": 1946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.001377296982972e-07,
+      "loss": 0.2054,
+      "step": 1947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.941581430497481e-07,
+      "loss": 0.2142,
+      "step": 1948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.881975541110666e-07,
+      "loss": 0.1956,
+      "step": 1949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.82255975317956e-07,
+      "loss": 0.2148,
+      "step": 1950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.763334190664552e-07,
+      "loss": 0.2031,
+      "step": 1951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.70429897712921e-07,
+      "loss": 0.2076,
+      "step": 1952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.2016,
+      "step": 1953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.586800089265712e-07,
+      "loss": 0.195,
+      "step": 1954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.528336660077974e-07,
+      "loss": 0.2077,
+      "step": 1955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470064070150186e-07,
+      "loss": 0.2021,
+      "step": 1956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.411982441057687e-07,
+      "loss": 0.2093,
+      "step": 1957
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.354091893977401e-07,
+      "loss": 0.2111,
+      "step": 1958
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.296392549687593e-07,
+      "loss": 0.2056,
+      "step": 1959
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.238884528567648e-07,
+      "loss": 0.2113,
+      "step": 1960
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181567950597713e-07,
+      "loss": 0.2074,
+      "step": 1961
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.124442935358623e-07,
+      "loss": 0.197,
+      "step": 1962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.067509602031465e-07,
+      "loss": 0.2109,
+      "step": 1963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.010768069397457e-07,
+      "loss": 0.2027,
+      "step": 1964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.954218455837703e-07,
+      "loss": 0.199,
+      "step": 1965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.897860879332797e-07,
+      "loss": 0.2047,
+      "step": 1966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.84169545746275e-07,
+      "loss": 0.2084,
+      "step": 1967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.785722307406685e-07,
+      "loss": 0.2022,
+      "step": 1968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729941545942577e-07,
+      "loss": 0.2061,
+      "step": 1969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.674353289446946e-07,
+      "loss": 0.2018,
+      "step": 1970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618957653894809e-07,
+      "loss": 0.2025,
+      "step": 1971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.563754754859187e-07,
+      "loss": 0.208,
+      "step": 1972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.508744707511118e-07,
+      "loss": 0.2056,
+      "step": 1973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.453927626619184e-07,
+      "loss": 0.2043,
+      "step": 1974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.399303626549415e-07,
+      "loss": 0.1978,
+      "step": 1975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.344872821265025e-07,
+      "loss": 0.1989,
+      "step": 1976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.290635324326168e-07,
+      "loss": 0.202,
+      "step": 1977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.236591248889624e-07,
+      "loss": 0.2038,
+      "step": 1978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.182740707708757e-07,
+      "loss": 0.2117,
+      "step": 1979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129083813133031e-07,
+      "loss": 0.205,
+      "step": 1980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.075620677107975e-07,
+      "loss": 0.2063,
+      "step": 1981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.022351411174866e-07,
+      "loss": 0.2065,
+      "step": 1982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969276126470493e-07,
+      "loss": 0.2008,
+      "step": 1983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.916394933726923e-07,
+      "loss": 0.1997,
+      "step": 1984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.863707943271325e-07,
+      "loss": 0.2054,
+      "step": 1985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.811215265025673e-07,
+      "loss": 0.2037,
+      "step": 1986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.758917008506549e-07,
+      "loss": 0.2064,
+      "step": 1987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70681328282492e-07,
+      "loss": 0.1979,
+      "step": 1988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.654904196685841e-07,
+      "loss": 0.2094,
+      "step": 1989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.603189858388393e-07,
+      "loss": 0.2081,
+      "step": 1990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.551670375825259e-07,
+      "loss": 0.204,
+      "step": 1991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.500345856482615e-07,
+      "loss": 0.2061,
+      "step": 1992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.449216407439906e-07,
+      "loss": 0.1999,
+      "step": 1993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.398282135369571e-07,
+      "loss": 0.1995,
+      "step": 1994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34754314653685e-07,
+      "loss": 0.2119,
+      "step": 1995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.296999546799599e-07,
+      "loss": 0.2061,
+      "step": 1996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.246651441607931e-07,
+      "loss": 0.2046,
+      "step": 1997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.196498936004236e-07,
+      "loss": 0.2063,
+      "step": 1998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.14654213462268e-07,
+      "loss": 0.2088,
+      "step": 1999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.096781141689223e-07,
+      "loss": 0.2042,
+      "step": 2000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.047216061021266e-07,
+      "loss": 0.2029,
+      "step": 2001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.99784699602749e-07,
+      "loss": 0.208,
+      "step": 2002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.948674049707604e-07,
+      "loss": 0.2041,
+      "step": 2003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.899697324652165e-07,
+      "loss": 0.2089,
+      "step": 2004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.85091692304236e-07,
+      "loss": 0.2063,
+      "step": 2005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.802332946649757e-07,
+      "loss": 0.2041,
+      "step": 2006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.753945496836144e-07,
+      "loss": 0.2,
+      "step": 2007
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.705754674553243e-07,
+      "loss": 0.2077,
+      "step": 2008
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.65776058034263e-07,
+      "loss": 0.2089,
+      "step": 2009
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.609963314335365e-07,
+      "loss": 0.2008,
+      "step": 2010
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.562362976251901e-07,
+      "loss": 0.2057,
+      "step": 2011
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.51495966540182e-07,
+      "loss": 0.2064,
+      "step": 2012
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.467753480683658e-07,
+      "loss": 0.1999,
+      "step": 2013
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.420744520584664e-07,
+      "loss": 0.2124,
+      "step": 2014
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373932883180655e-07,
+      "loss": 0.1984,
+      "step": 2015
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.327318666135672e-07,
+      "loss": 0.1916,
+      "step": 2016
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280901966702012e-07,
+      "loss": 0.2085,
+      "step": 2017
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 0.1954,
+      "step": 2018
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.188661507616799e-07,
+      "loss": 0.2024,
+      "step": 2019
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.142837940408474e-07,
+      "loss": 0.1978,
+      "step": 2020
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.097212275697461e-07,
+      "loss": 0.2054,
+      "step": 2021
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.051784608673549e-07,
+      "loss": 0.2074,
+      "step": 2022
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.006555034113459e-07,
+      "loss": 0.1985,
+      "step": 2023
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.961523646380562e-07,
+      "loss": 0.2052,
+      "step": 2024
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916690539424828e-07,
+      "loss": 0.1951,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.872055806782517e-07,
+      "loss": 0.1983,
+      "step": 2026
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.827619541575967e-07,
+      "loss": 0.1973,
+      "step": 2027
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.783381836513557e-07,
+      "loss": 0.2095,
+      "step": 2028
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7393427838893e-07,
+      "loss": 0.2029,
+      "step": 2029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.695502475582814e-07,
+      "loss": 0.2116,
+      "step": 2030
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.651861003059055e-07,
+      "loss": 0.1984,
+      "step": 2031
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6084184573681465e-07,
+      "loss": 0.2,
+      "step": 2032
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5651749291451884e-07,
+      "loss": 0.2099,
+      "step": 2033
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.52213050861009e-07,
+      "loss": 0.2024,
+      "step": 2034
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.479285285567281e-07,
+      "loss": 0.2019,
+      "step": 2035
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4366393494057135e-07,
+      "loss": 0.1958,
+      "step": 2036
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.39419278909845e-07,
+      "loss": 0.1994,
+      "step": 2037
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.351945693202686e-07,
+      "loss": 0.2056,
+      "step": 2038
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.309898149859393e-07,
+      "loss": 0.2007,
+      "step": 2039
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.268050246793276e-07,
+      "loss": 0.208,
+      "step": 2040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2264020713124854e-07,
+      "loss": 0.2071,
+      "step": 2041
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1849537103084924e-07,
+      "loss": 0.1974,
+      "step": 2042
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1437052502558693e-07,
+      "loss": 0.2067,
+      "step": 2043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.102656777212166e-07,
+      "loss": 0.2094,
+      "step": 2044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0618083768176996e-07,
+      "loss": 0.2096,
+      "step": 2045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0211601342952876e-07,
+      "loss": 0.2089,
+      "step": 2046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9807121344502706e-07,
+      "loss": 0.2004,
+      "step": 2047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.940464461670135e-07,
+      "loss": 0.211,
+      "step": 2048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9004171999244553e-07,
+      "loss": 0.2116,
+      "step": 2049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8605704327646877e-07,
+      "loss": 0.1926,
+      "step": 2050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8209242433239755e-07,
+      "loss": 0.2053,
+      "step": 2051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7814787143169885e-07,
+      "loss": 0.2055,
+      "step": 2052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.742233928039796e-07,
+      "loss": 0.2006,
+      "step": 2053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7031899663695804e-07,
+      "loss": 0.2009,
+      "step": 2054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.664346910764638e-07,
+      "loss": 0.2067,
+      "step": 2055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6257048422639995e-07,
+      "loss": 0.2027,
+      "step": 2056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587263841487454e-07,
+      "loss": 0.2044,
+      "step": 2057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5490239886352584e-07,
+      "loss": 0.2014,
+      "step": 2058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.510985363488051e-07,
+      "loss": 0.2059,
+      "step": 2059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4731480454065823e-07,
+      "loss": 0.1973,
+      "step": 2060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4355121133316847e-07,
+      "loss": 0.1966,
+      "step": 2061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3980776457839593e-07,
+      "loss": 0.2045,
+      "step": 2062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.360844720863765e-07,
+      "loss": 0.2054,
+      "step": 2063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3238134162508984e-07,
+      "loss": 0.2025,
+      "step": 2064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.286983809204558e-07,
+      "loss": 0.2039,
+      "step": 2065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2503559765631577e-07,
+      "loss": 0.2013,
+      "step": 2066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2139299947441026e-07,
+      "loss": 0.2066,
+      "step": 2067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1777059397436693e-07,
+      "loss": 0.2075,
+      "step": 2068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1416838871368925e-07,
+      "loss": 0.1976,
+      "step": 2069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1058639120773225e-07,
+      "loss": 0.2094,
+      "step": 2070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0702460892969356e-07,
+      "loss": 0.2004,
+      "step": 2071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0348304931059556e-07,
+      "loss": 0.1997,
+      "step": 2072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9996171973926567e-07,
+      "loss": 0.2052,
+      "step": 2073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9646062756233273e-07,
+      "loss": 0.2019,
+      "step": 2074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9297978008419604e-07,
+      "loss": 0.2066,
+      "step": 2075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8951918456702087e-07,
+      "loss": 0.2082,
+      "step": 2076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8607884823072196e-07,
+      "loss": 0.1995,
+      "step": 2077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.826587782529444e-07,
+      "loss": 0.203,
+      "step": 2078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7925898176905275e-07,
+      "loss": 0.1974,
+      "step": 2079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.758794658721164e-07,
+      "loss": 0.2101,
+      "step": 2080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725202376128855e-07,
+      "loss": 0.2098,
+      "step": 2081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.691813039997937e-07,
+      "loss": 0.1978,
+      "step": 2082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.658626719989266e-07,
+      "loss": 0.2019,
+      "step": 2083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6256434853401455e-07,
+      "loss": 0.2032,
+      "step": 2084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5928634048642323e-07,
+      "loss": 0.2013,
+      "step": 2085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5602865469512626e-07,
+      "loss": 0.1976,
+      "step": 2086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.527912979567038e-07,
+      "loss": 0.1995,
+      "step": 2087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4957427702532177e-07,
+      "loss": 0.2047,
+      "step": 2088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4637759861271704e-07,
+      "loss": 0.198,
+      "step": 2089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4320126938819023e-07,
+      "loss": 0.2034,
+      "step": 2090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.400452959785815e-07,
+      "loss": 0.2027,
+      "step": 2091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3690968496826505e-07,
+      "loss": 0.2038,
+      "step": 2092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3379444289913344e-07,
+      "loss": 0.1972,
+      "step": 2093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3069957627058104e-07,
+      "loss": 0.2063,
+      "step": 2094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2762509153949285e-07,
+      "loss": 0.2051,
+      "step": 2095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2457099512023129e-07,
+      "loss": 0.2003,
+      "step": 2096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2153729338462272e-07,
+      "loss": 0.2096,
+      "step": 2097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.2012,
+      "step": 2098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1553109923890703e-07,
+      "loss": 0.2073,
+      "step": 2099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1255861935964628e-07,
+      "loss": 0.1983,
+      "step": 2100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0960655922571238e-07,
+      "loss": 0.2074,
+      "step": 2101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.066749249960498e-07,
+      "loss": 0.2138,
+      "step": 2102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.037637227869871e-07,
+      "loss": 0.2002,
+      "step": 2103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0087295867222912e-07,
+      "loss": 0.2002,
+      "step": 2104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9800263868283708e-07,
+      "loss": 0.1965,
+      "step": 2105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.951527688072208e-07,
+      "loss": 0.1989,
+      "step": 2106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9232335499112186e-07,
+      "loss": 0.202,
+      "step": 2107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8951440313760838e-07,
+      "loss": 0.1997,
+      "step": 2108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8672591910705363e-07,
+      "loss": 0.2103,
+      "step": 2109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8395790871713393e-07,
+      "loss": 0.2052,
+      "step": 2110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8121037774280314e-07,
+      "loss": 0.199,
+      "step": 2111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.784833319162971e-07,
+      "loss": 0.2034,
+      "step": 2112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7577677692710682e-07,
+      "loss": 0.1968,
+      "step": 2113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7309071842197544e-07,
+      "loss": 0.1994,
+      "step": 2114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7042516200488357e-07,
+      "loss": 0.2005,
+      "step": 2115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6778011323703825e-07,
+      "loss": 0.2068,
+      "step": 2116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6515557763685964e-07,
+      "loss": 0.2007,
+      "step": 2117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6255156067997325e-07,
+      "loss": 0.2078,
+      "step": 2118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5996806779919215e-07,
+      "loss": 0.2046,
+      "step": 2119
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5740510438451374e-07,
+      "loss": 0.2038,
+      "step": 2120
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5486267578310177e-07,
+      "loss": 0.2019,
+      "step": 2121
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5234078729927992e-07,
+      "loss": 0.2066,
+      "step": 2122
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4983944419451613e-07,
+      "loss": 0.2109,
+      "step": 2123
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.473586516874159e-07,
+      "loss": 0.2025,
+      "step": 2124
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4489841495370804e-07,
+      "loss": 0.2009,
+      "step": 2125
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4245873912623777e-07,
+      "loss": 0.2007,
+      "step": 2126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.400396292949513e-07,
+      "loss": 0.2015,
+      "step": 2127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.37641090506887e-07,
+      "loss": 0.2054,
+      "step": 2128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.352631277661709e-07,
+      "loss": 0.2045,
+      "step": 2129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.329057460339933e-07,
+      "loss": 0.2011,
+      "step": 2130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3056895022861005e-07,
+      "loss": 0.1998,
+      "step": 2131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2825274522532795e-07,
+      "loss": 0.2014,
+      "step": 2132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2595713585649482e-07,
+      "loss": 0.2055,
+      "step": 2133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2368212691148852e-07,
+      "loss": 0.201,
+      "step": 2134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.214277231367078e-07,
+      "loss": 0.2068,
+      "step": 2135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1919392923556261e-07,
+      "loss": 0.2077,
+      "step": 2136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1698074986846608e-07,
+      "loss": 0.2069,
+      "step": 2137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1478818965281912e-07,
+      "loss": 0.1979,
+      "step": 2138
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1261625316300596e-07,
+      "loss": 0.2048,
+      "step": 2139
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1046494493038517e-07,
+      "loss": 0.2,
+      "step": 2140
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.083342694432754e-07,
+      "loss": 0.1932,
+      "step": 2141
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0622423114694968e-07,
+      "loss": 0.2049,
+      "step": 2142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.041348344436277e-07,
+      "loss": 0.2085,
+      "step": 2143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0206608369245807e-07,
+      "loss": 0.1991,
+      "step": 2144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0001798320952161e-07,
+      "loss": 0.1975,
+      "step": 2145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.79905372678136e-08,
+      "loss": 0.2042,
+      "step": 2146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.598375009723603e-08,
+      "loss": 0.2064,
+      "step": 2147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.399762588459205e-08,
+      "loss": 0.2019,
+      "step": 2148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.20321687735748e-08,
+      "loss": 0.1957,
+      "step": 2149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.008738286475748e-08,
+      "loss": 0.1997,
+      "step": 2150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.816327221558896e-08,
+      "loss": 0.2042,
+      "step": 2151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.625984084038253e-08,
+      "loss": 0.2065,
+      "step": 2152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.437709271030603e-08,
+      "loss": 0.1959,
+      "step": 2153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.251503175337739e-08,
+      "loss": 0.2024,
+      "step": 2154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.067366185445236e-08,
+      "loss": 0.2091,
+      "step": 2155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.885298685522235e-08,
+      "loss": 0.2009,
+      "step": 2156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.705301055419667e-08,
+      "loss": 0.2032,
+      "step": 2157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.52737367067069e-08,
+      "loss": 0.1962,
+      "step": 2158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.351516902488698e-08,
+      "loss": 0.1981,
+      "step": 2159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.177731117767539e-08,
+      "loss": 0.2018,
+      "step": 2160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.006016679080075e-08,
+      "loss": 0.2084,
+      "step": 2161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.836373944677954e-08,
+      "loss": 0.2003,
+      "step": 2162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.668803268490066e-08,
+      "loss": 0.2023,
+      "step": 2163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.503305000122861e-08,
+      "loss": 0.2021,
+      "step": 2164
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.339879484858924e-08,
+      "loss": 0.2002,
+      "step": 2165
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.17852706365607e-08,
+      "loss": 0.199,
+      "step": 2166
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.019248073147577e-08,
+      "loss": 0.2097,
+      "step": 2167
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 0.2063,
+      "step": 2168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.706911709115193e-08,
+      "loss": 0.2036,
+      "step": 2169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5538549872251603e-08,
+      "loss": 0.2055,
+      "step": 2170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4028729992958715e-08,
+      "loss": 0.2002,
+      "step": 2171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.253966060324356e-08,
+      "loss": 0.1969,
+      "step": 2172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1071344809781045e-08,
+      "loss": 0.2017,
+      "step": 2173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.962378567594961e-08,
+      "loss": 0.2001,
+      "step": 2174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.819698622182456e-08,
+      "loss": 0.2022,
+      "step": 2175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.679094942416473e-08,
+      "loss": 0.2066,
+      "step": 2176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.540567821641584e-08,
+      "loss": 0.202,
+      "step": 2177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.404117548869824e-08,
+      "loss": 0.1889,
+      "step": 2178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.269744408780252e-08,
+      "loss": 0.1992,
+      "step": 2179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.137448681718392e-08,
+      "loss": 0.2115,
+      "step": 2180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.007230643695681e-08,
+      "loss": 0.2027,
+      "step": 2181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.879090566388688e-08,
+      "loss": 0.1996,
+      "step": 2182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.753028717138785e-08,
+      "loss": 0.2042,
+      "step": 2183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.629045358951366e-08,
+      "loss": 0.205,
+      "step": 2184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.50714075049563e-08,
+      "loss": 0.2048,
+      "step": 2185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3873151461037975e-08,
+      "loss": 0.2046,
+      "step": 2186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2695687957703393e-08,
+      "loss": 0.2064,
+      "step": 2187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1539019451521935e-08,
+      "loss": 0.1961,
+      "step": 2188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.040314835567326e-08,
+      "loss": 0.1969,
+      "step": 2189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9288077039951734e-08,
+      "loss": 0.1917,
+      "step": 2190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8193807830750875e-08,
+      "loss": 0.2027,
+      "step": 2191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7120343011071138e-08,
+      "loss": 0.1921,
+      "step": 2192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.606768482050215e-08,
+      "loss": 0.2108,
+      "step": 2193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.503583545522714e-08,
+      "loss": 0.2084,
+      "step": 2194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4024797068017414e-08,
+      "loss": 0.1981,
+      "step": 2195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3034571768221213e-08,
+      "loss": 0.2093,
+      "step": 2196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2065161621767084e-08,
+      "loss": 0.2025,
+      "step": 2197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1116568651156076e-08,
+      "loss": 0.1984,
+      "step": 2198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0188794835458437e-08,
+      "loss": 0.2002,
+      "step": 2199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9281842110308035e-08,
+      "loss": 0.199,
+      "step": 2200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8395712367897943e-08,
+      "loss": 0.2051,
+      "step": 2201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7530407456980425e-08,
+      "loss": 0.204,
+      "step": 2202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6685929182860274e-08,
+      "loss": 0.2048,
+      "step": 2203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.586227930738815e-08,
+      "loss": 0.2046,
+      "step": 2204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.505945954896393e-08,
+      "loss": 0.2033,
+      "step": 2205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.427747158252779e-08,
+      "loss": 0.2007,
+      "step": 2206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3516317039555804e-08,
+      "loss": 0.1979,
+      "step": 2207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2775997508061023e-08,
+      "loss": 0.202,
+      "step": 2208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2056514532590158e-08,
+      "loss": 0.1949,
+      "step": 2209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.135786961421248e-08,
+      "loss": 0.2055,
+      "step": 2210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0680064210527585e-08,
+      "loss": 0.1988,
+      "step": 2211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0023099735654295e-08,
+      "loss": 0.2024,
+      "step": 2212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.38697756023288e-09,
+      "loss": 0.1931,
+      "step": 2213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.771699011416169e-09,
+      "loss": 0.2043,
+      "step": 2214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.17726537287511e-09,
+      "loss": 0.1907,
+      "step": 2215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.603677884787663e-09,
+      "loss": 0.2026,
+      "step": 2216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.050937743841024e-09,
+      "loss": 0.2028,
+      "step": 2217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5190461032305085e-09,
+      "loss": 0.203,
+      "step": 2218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.008004072650675e-09,
+      "loss": 0.1983,
+      "step": 2219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.517812718301985e-09,
+      "loss": 0.2053,
+      "step": 2220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.048473062878589e-09,
+      "loss": 0.2038,
+      "step": 2221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.599986085573882e-09,
+      "loss": 0.1924,
+      "step": 2222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.172352722077167e-09,
+      "loss": 0.205,
+      "step": 2223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.765573864568106e-09,
+      "loss": 0.2089,
+      "step": 2224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3796503617167244e-09,
+      "loss": 0.1965,
+      "step": 2225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.014583018686734e-09,
+      "loss": 0.1982,
+      "step": 2226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6703725971222173e-09,
+      "loss": 0.2011,
+      "step": 2227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.347019815158724e-09,
+      "loss": 0.2005,
+      "step": 2228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.044525347413284e-09,
+      "loss": 0.2035,
+      "step": 2229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7628898249866243e-09,
+      "loss": 0.2047,
+      "step": 2230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5021138354609498e-09,
+      "loss": 0.2035,
+      "step": 2231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2621979228988334e-09,
+      "loss": 0.2044,
+      "step": 2232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431425878421053e-09,
+      "loss": 0.2089,
+      "step": 2233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.449482873096326e-10,
+      "loss": 0.2036,
+      "step": 2234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.676154347995401e-10,
+      "loss": 0.2029,
+      "step": 2235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.111444002836585e-10,
+      "loss": 0.1966,
+      "step": 2236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7553551021085597e-10,
+      "loss": 0.1954,
+      "step": 2237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.607890475070374e-10,
+      "loss": 0.1903,
+      "step": 2238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6690525156737303e-10,
+      "loss": 0.1984,
+      "step": 2239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.388431826629074e-11,
+      "loss": 0.2016,
+      "step": 2240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.172639994637351e-11,
+      "loss": 0.1974,
+      "step": 2241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0431605427241309e-11,
+      "loss": 0.2047,
+      "step": 2242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.2056,
+      "step": 2243
+    },
+    {
+      "epoch": 1.0,
+      "step": 2243,
+      "total_flos": 853631281004544.0,
+      "train_loss": 0.24281185524595458,
+      "train_runtime": 26730.9352,
+      "train_samples_per_second": 10.737,
+      "train_steps_per_second": 0.084
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2243,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 853631281004544.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}