diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,55573 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 9258,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.194244604316547e-08,
+      "loss": 1.7433,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4388489208633095e-07,
+      "loss": 1.793,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1582733812949643e-07,
+      "loss": 1.7063,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.877697841726619e-07,
+      "loss": 1.7176,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5971223021582736e-07,
+      "loss": 1.6802,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3165467625899287e-07,
+      "loss": 1.567,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.035971223021583e-07,
+      "loss": 1.4337,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.755395683453238e-07,
+      "loss": 1.479,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.474820143884893e-07,
+      "loss": 1.446,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.194244604316547e-07,
+      "loss": 1.295,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.913669064748202e-07,
+      "loss": 1.2527,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.633093525179857e-07,
+      "loss": 1.1633,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.352517985611512e-07,
+      "loss": 1.0915,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0071942446043167e-06,
+      "loss": 1.0103,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.079136690647482e-06,
+      "loss": 0.9653,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1510791366906476e-06,
+      "loss": 0.8776,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2230215827338131e-06,
+      "loss": 0.858,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2949640287769785e-06,
+      "loss": 0.7871,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.366906474820144e-06,
+      "loss": 0.7278,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4388489208633094e-06,
+      "loss": 0.6461,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.510791366906475e-06,
+      "loss": 0.6554,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5827338129496403e-06,
+      "loss": 0.6359,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.654676258992806e-06,
+      "loss": 0.5749,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7266187050359715e-06,
+      "loss": 0.5846,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7985611510791368e-06,
+      "loss": 0.5231,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8705035971223024e-06,
+      "loss": 0.5442,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.942446043165468e-06,
+      "loss": 0.5123,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0143884892086333e-06,
+      "loss": 0.4823,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0863309352517987e-06,
+      "loss": 0.4743,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.158273381294964e-06,
+      "loss": 0.4697,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.23021582733813e-06,
+      "loss": 0.4427,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.302158273381295e-06,
+      "loss": 0.4522,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3741007194244605e-06,
+      "loss": 0.4473,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4460431654676263e-06,
+      "loss": 0.4268,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5179856115107916e-06,
+      "loss": 0.4334,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.589928057553957e-06,
+      "loss": 0.3896,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6618705035971228e-06,
+      "loss": 0.4326,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.733812949640288e-06,
+      "loss": 0.3845,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.805755395683453e-06,
+      "loss": 0.3675,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.877697841726619e-06,
+      "loss": 0.4069,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.949640287769784e-06,
+      "loss": 0.3621,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.02158273381295e-06,
+      "loss": 0.3737,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0935251798561158e-06,
+      "loss": 0.3554,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1654676258992807e-06,
+      "loss": 0.3597,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.237410071942446e-06,
+      "loss": 0.3633,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.309352517985612e-06,
+      "loss": 0.3358,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.381294964028777e-06,
+      "loss": 0.3427,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.453237410071943e-06,
+      "loss": 0.3451,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.525179856115108e-06,
+      "loss": 0.3291,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5971223021582737e-06,
+      "loss": 0.3362,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.669064748201439e-06,
+      "loss": 0.3497,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.741007194244605e-06,
+      "loss": 0.3246,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.81294964028777e-06,
+      "loss": 0.3313,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.884892086330936e-06,
+      "loss": 0.3376,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.956834532374101e-06,
+      "loss": 0.3602,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.028776978417267e-06,
+      "loss": 0.3118,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.100719424460432e-06,
+      "loss": 0.3204,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.172661870503597e-06,
+      "loss": 0.3235,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.244604316546763e-06,
+      "loss": 0.3274,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.316546762589928e-06,
+      "loss": 0.3052,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.388489208633094e-06,
+      "loss": 0.3009,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.46043165467626e-06,
+      "loss": 0.3275,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5323741007194245e-06,
+      "loss": 0.3141,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.60431654676259e-06,
+      "loss": 0.3032,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.676258992805755e-06,
+      "loss": 0.2833,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.748201438848921e-06,
+      "loss": 0.3178,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.820143884892087e-06,
+      "loss": 0.3169,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.892086330935253e-06,
+      "loss": 0.3227,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9640287769784175e-06,
+      "loss": 0.2952,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.035971223021583e-06,
+      "loss": 0.2839,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.107913669064749e-06,
+      "loss": 0.311,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.179856115107914e-06,
+      "loss": 0.2996,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.251798561151079e-06,
+      "loss": 0.2954,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3237410071942456e-06,
+      "loss": 0.3012,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3956834532374105e-06,
+      "loss": 0.2845,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.467625899280576e-06,
+      "loss": 0.2786,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.539568345323741e-06,
+      "loss": 0.2879,
+      "step": 77
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.611510791366906e-06,
+      "loss": 0.3091,
+      "step": 78
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.683453237410073e-06,
+      "loss": 0.3009,
+      "step": 79
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.755395683453238e-06,
+      "loss": 0.2847,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.8273381294964035e-06,
+      "loss": 0.3022,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.899280575539568e-06,
+      "loss": 0.2913,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.971223021582734e-06,
+      "loss": 0.2878,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0431654676259e-06,
+      "loss": 0.3028,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.115107913669065e-06,
+      "loss": 0.2918,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.1870503597122315e-06,
+      "loss": 0.2774,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.2589928057553964e-06,
+      "loss": 0.2958,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.330935251798561e-06,
+      "loss": 0.2605,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.402877697841727e-06,
+      "loss": 0.2662,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.474820143884892e-06,
+      "loss": 0.2893,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.546762589928059e-06,
+      "loss": 0.2927,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.618705035971224e-06,
+      "loss": 0.265,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.6906474820143886e-06,
+      "loss": 0.2707,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.762589928057554e-06,
+      "loss": 0.289,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.834532374100719e-06,
+      "loss": 0.252,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.906474820143886e-06,
+      "loss": 0.2649,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.978417266187051e-06,
+      "loss": 0.2609,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.050359712230216e-06,
+      "loss": 0.2968,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.122302158273382e-06,
+      "loss": 0.2751,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.194244604316547e-06,
+      "loss": 0.2562,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.266187050359713e-06,
+      "loss": 0.2716,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.338129496402878e-06,
+      "loss": 0.2915,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.410071942446043e-06,
+      "loss": 0.2943,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.48201438848921e-06,
+      "loss": 0.2637,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5539568345323745e-06,
+      "loss": 0.2519,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.62589928057554e-06,
+      "loss": 0.2868,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.697841726618706e-06,
+      "loss": 0.279,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.769784172661872e-06,
+      "loss": 0.2512,
+      "step": 108
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.841726618705036e-06,
+      "loss": 0.2631,
+      "step": 109
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.913669064748202e-06,
+      "loss": 0.2655,
+      "step": 110
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.985611510791367e-06,
+      "loss": 0.2603,
+      "step": 111
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.057553956834533e-06,
+      "loss": 0.2851,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.129496402877699e-06,
+      "loss": 0.2661,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.201438848920865e-06,
+      "loss": 0.2738,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.273381294964029e-06,
+      "loss": 0.2808,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.345323741007195e-06,
+      "loss": 0.2932,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.41726618705036e-06,
+      "loss": 0.2548,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.489208633093526e-06,
+      "loss": 0.2528,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.561151079136692e-06,
+      "loss": 0.2486,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.633093525179856e-06,
+      "loss": 0.2573,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.705035971223022e-06,
+      "loss": 0.2627,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.776978417266188e-06,
+      "loss": 0.291,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.848920863309353e-06,
+      "loss": 0.2597,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.92086330935252e-06,
+      "loss": 0.2454,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.992805755395683e-06,
+      "loss": 0.2472,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.064748201438849e-06,
+      "loss": 0.2711,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.136690647482015e-06,
+      "loss": 0.2693,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.20863309352518e-06,
+      "loss": 0.2538,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.280575539568346e-06,
+      "loss": 0.2507,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.35251798561151e-06,
+      "loss": 0.25,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.424460431654678e-06,
+      "loss": 0.2404,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.496402877697842e-06,
+      "loss": 0.2644,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.568345323741008e-06,
+      "loss": 0.2635,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.640287769784174e-06,
+      "loss": 0.2603,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.712230215827338e-06,
+      "loss": 0.2686,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.784172661870505e-06,
+      "loss": 0.258,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.85611510791367e-06,
+      "loss": 0.2732,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.928057553956835e-06,
+      "loss": 0.269,
+      "step": 138
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 0.2675,
+      "step": 139
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0071942446043167e-05,
+      "loss": 0.2553,
+      "step": 140
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.014388489208633e-05,
+      "loss": 0.2579,
+      "step": 141
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0215827338129498e-05,
+      "loss": 0.221,
+      "step": 142
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0287769784172664e-05,
+      "loss": 0.2571,
+      "step": 143
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0359712230215828e-05,
+      "loss": 0.2863,
+      "step": 144
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0431654676258994e-05,
+      "loss": 0.2533,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0503597122302158e-05,
+      "loss": 0.2449,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0575539568345325e-05,
+      "loss": 0.2274,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0647482014388491e-05,
+      "loss": 0.2711,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0719424460431655e-05,
+      "loss": 0.2533,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0791366906474821e-05,
+      "loss": 0.2437,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0863309352517985e-05,
+      "loss": 0.2807,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0935251798561153e-05,
+      "loss": 0.2479,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1007194244604318e-05,
+      "loss": 0.2536,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1079136690647482e-05,
+      "loss": 0.2562,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1151079136690648e-05,
+      "loss": 0.2377,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1223021582733812e-05,
+      "loss": 0.2573,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.129496402877698e-05,
+      "loss": 0.2481,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1366906474820146e-05,
+      "loss": 0.2373,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.143884892086331e-05,
+      "loss": 0.2553,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1510791366906475e-05,
+      "loss": 0.236,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.158273381294964e-05,
+      "loss": 0.2329,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1654676258992807e-05,
+      "loss": 0.2379,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1726618705035973e-05,
+      "loss": 0.2347,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1798561151079137e-05,
+      "loss": 0.2271,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1870503597122303e-05,
+      "loss": 0.2422,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1942446043165468e-05,
+      "loss": 0.2311,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2014388489208634e-05,
+      "loss": 0.233,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.20863309352518e-05,
+      "loss": 0.2436,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2158273381294966e-05,
+      "loss": 0.2379,
+      "step": 169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.223021582733813e-05,
+      "loss": 0.2442,
+      "step": 170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2302158273381296e-05,
+      "loss": 0.243,
+      "step": 171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2374100719424463e-05,
+      "loss": 0.2279,
+      "step": 172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2446043165467627e-05,
+      "loss": 0.2651,
+      "step": 173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2517985611510793e-05,
+      "loss": 0.2411,
+      "step": 174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2589928057553957e-05,
+      "loss": 0.2569,
+      "step": 175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2661870503597123e-05,
+      "loss": 0.2716,
+      "step": 176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.273381294964029e-05,
+      "loss": 0.2583,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2805755395683454e-05,
+      "loss": 0.2586,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.287769784172662e-05,
+      "loss": 0.2444,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2949640287769784e-05,
+      "loss": 0.2419,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.302158273381295e-05,
+      "loss": 0.2381,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3093525179856117e-05,
+      "loss": 0.2413,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3165467625899282e-05,
+      "loss": 0.2506,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3237410071942447e-05,
+      "loss": 0.2516,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3309352517985611e-05,
+      "loss": 0.2251,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3381294964028777e-05,
+      "loss": 0.2531,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3453237410071945e-05,
+      "loss": 0.2445,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3525179856115109e-05,
+      "loss": 0.2498,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3597122302158274e-05,
+      "loss": 0.2269,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3669064748201439e-05,
+      "loss": 0.249,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3741007194244604e-05,
+      "loss": 0.2218,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3812949640287772e-05,
+      "loss": 0.2539,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3884892086330936e-05,
+      "loss": 0.2338,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3956834532374102e-05,
+      "loss": 0.2492,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4028776978417267e-05,
+      "loss": 0.2576,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4100719424460432e-05,
+      "loss": 0.2212,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4172661870503599e-05,
+      "loss": 0.2438,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4244604316546765e-05,
+      "loss": 0.2486,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4316546762589929e-05,
+      "loss": 0.2583,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4388489208633095e-05,
+      "loss": 0.242,
+      "step": 200
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4460431654676259e-05,
+      "loss": 0.2439,
+      "step": 201
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4532374100719426e-05,
+      "loss": 0.2449,
+      "step": 202
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4604316546762592e-05,
+      "loss": 0.2356,
+      "step": 203
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4676258992805756e-05,
+      "loss": 0.2419,
+      "step": 204
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4748201438848922e-05,
+      "loss": 0.245,
+      "step": 205
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4820143884892086e-05,
+      "loss": 0.2369,
+      "step": 206
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4892086330935253e-05,
+      "loss": 0.2456,
+      "step": 207
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.496402877697842e-05,
+      "loss": 0.23,
+      "step": 208
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5035971223021583e-05,
+      "loss": 0.2551,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5107913669064749e-05,
+      "loss": 0.2429,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5179856115107917e-05,
+      "loss": 0.2524,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.525179856115108e-05,
+      "loss": 0.2218,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5323741007194246e-05,
+      "loss": 0.2654,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5395683453237412e-05,
+      "loss": 0.2386,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5467625899280575e-05,
+      "loss": 0.2351,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5539568345323744e-05,
+      "loss": 0.2307,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.561151079136691e-05,
+      "loss": 0.2272,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5683453237410072e-05,
+      "loss": 0.2245,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5755395683453238e-05,
+      "loss": 0.2443,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5827338129496403e-05,
+      "loss": 0.2152,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.589928057553957e-05,
+      "loss": 0.2394,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5971223021582735e-05,
+      "loss": 0.2201,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.60431654676259e-05,
+      "loss": 0.223,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6115107913669067e-05,
+      "loss": 0.2235,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6187050359712232e-05,
+      "loss": 0.2421,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6258992805755398e-05,
+      "loss": 0.2256,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6330935251798564e-05,
+      "loss": 0.2315,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.640287769784173e-05,
+      "loss": 0.2305,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6474820143884892e-05,
+      "loss": 0.2639,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6546762589928058e-05,
+      "loss": 0.2277,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6618705035971227e-05,
+      "loss": 0.2384,
+      "step": 231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.669064748201439e-05,
+      "loss": 0.243,
+      "step": 232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6762589928057555e-05,
+      "loss": 0.2593,
+      "step": 233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.683453237410072e-05,
+      "loss": 0.2378,
+      "step": 234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6906474820143887e-05,
+      "loss": 0.2194,
+      "step": 235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6978417266187053e-05,
+      "loss": 0.215,
+      "step": 236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7050359712230218e-05,
+      "loss": 0.2453,
+      "step": 237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7122302158273384e-05,
+      "loss": 0.2206,
+      "step": 238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7194244604316546e-05,
+      "loss": 0.2235,
+      "step": 239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7266187050359712e-05,
+      "loss": 0.2504,
+      "step": 240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.733812949640288e-05,
+      "loss": 0.2275,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7410071942446044e-05,
+      "loss": 0.2552,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.748201438848921e-05,
+      "loss": 0.2148,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7553956834532375e-05,
+      "loss": 0.2256,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.762589928057554e-05,
+      "loss": 0.2272,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7697841726618707e-05,
+      "loss": 0.2385,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7769784172661873e-05,
+      "loss": 0.2033,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.784172661870504e-05,
+      "loss": 0.2259,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7913669064748204e-05,
+      "loss": 0.2236,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7985611510791367e-05,
+      "loss": 0.2469,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8057553956834536e-05,
+      "loss": 0.2369,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8129496402877698e-05,
+      "loss": 0.2438,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8201438848920864e-05,
+      "loss": 0.26,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.827338129496403e-05,
+      "loss": 0.2414,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8345323741007196e-05,
+      "loss": 0.2079,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.841726618705036e-05,
+      "loss": 0.2396,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8489208633093527e-05,
+      "loss": 0.2278,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8561151079136693e-05,
+      "loss": 0.2312,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.863309352517986e-05,
+      "loss": 0.2516,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.870503597122302e-05,
+      "loss": 0.2281,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.877697841726619e-05,
+      "loss": 0.2266,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8848920863309356e-05,
+      "loss": 0.2046,
+      "step": 262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.892086330935252e-05,
+      "loss": 0.2414,
+      "step": 263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8992805755395684e-05,
+      "loss": 0.2236,
+      "step": 264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.906474820143885e-05,
+      "loss": 0.2445,
+      "step": 265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9136690647482016e-05,
+      "loss": 0.2182,
+      "step": 266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.920863309352518e-05,
+      "loss": 0.213,
+      "step": 267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9280575539568347e-05,
+      "loss": 0.2452,
+      "step": 268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9352517985611513e-05,
+      "loss": 0.2126,
+      "step": 269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9424460431654675e-05,
+      "loss": 0.235,
+      "step": 270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9496402877697845e-05,
+      "loss": 0.2454,
+      "step": 271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.956834532374101e-05,
+      "loss": 0.2352,
+      "step": 272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640287769784173e-05,
+      "loss": 0.212,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.971223021582734e-05,
+      "loss": 0.2552,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784172661870504e-05,
+      "loss": 0.2598,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985611510791367e-05,
+      "loss": 0.2179,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9928057553956836e-05,
+      "loss": 0.2158,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2188,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999388048404e-05,
+      "loss": 0.2447,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999997552193694e-05,
+      "loss": 0.2373,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994492436087e-05,
+      "loss": 0.2091,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999020877596e-05,
+      "loss": 0.2401,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999984701213838e-05,
+      "loss": 0.2261,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999779697504e-05,
+      "loss": 0.209,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999970014386465e-05,
+      "loss": 0.2334,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999960835123006e-05,
+      "loss": 0.2382,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995043196115e-05,
+      "loss": 0.2457,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999938804902167e-05,
+      "loss": 0.2454,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999925953947482e-05,
+      "loss": 0.2261,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999911879098665e-05,
+      "loss": 0.2739,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999989658035744e-05,
+      "loss": 0.2241,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999880057725685e-05,
+      "loss": 0.2443,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999862311205415e-05,
+      "loss": 0.2224,
+      "step": 293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999843340798802e-05,
+      "loss": 0.2225,
+      "step": 294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982314650817e-05,
+      "loss": 0.2337,
+      "step": 295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999980172833599e-05,
+      "loss": 0.2404,
+      "step": 296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999977908628489e-05,
+      "loss": 0.2099,
+      "step": 297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999975522035763e-05,
+      "loss": 0.2348,
+      "step": 298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999973013055714e-05,
+      "loss": 0.24,
+      "step": 299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999970381688648e-05,
+      "loss": 0.2134,
+      "step": 300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999676279348886e-05,
+      "loss": 0.2494,
+      "step": 301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999647517947714e-05,
+      "loss": 0.2103,
+      "step": 302
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999617532686492e-05,
+      "loss": 0.2448,
+      "step": 303
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999586323568885e-05,
+      "loss": 0.2316,
+      "step": 304
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999955389059872e-05,
+      "loss": 0.2305,
+      "step": 305
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999952023377996e-05,
+      "loss": 0.2326,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999485353116726e-05,
+      "loss": 0.24,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999944924861329e-05,
+      "loss": 0.2399,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999411920274064e-05,
+      "loss": 0.2313,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999373368103622e-05,
+      "loss": 0.223,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999933359210668e-05,
+      "loss": 0.2025,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999929259228811e-05,
+      "loss": 0.2214,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999250368652927e-05,
+      "loss": 0.2272,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999206921206294e-05,
+      "loss": 0.2269,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999916224995354e-05,
+      "loss": 0.2367,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999116354900124e-05,
+      "loss": 0.2243,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999069236051665e-05,
+      "loss": 0.2148,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999902089341393e-05,
+      "loss": 0.235,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998971326992834e-05,
+      "loss": 0.2192,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999892053679445e-05,
+      "loss": 0.2322,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998868522824986e-05,
+      "loss": 0.2303,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999881528509081e-05,
+      "loss": 0.2007,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999876082359844e-05,
+      "loss": 0.2019,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999870513835454e-05,
+      "loss": 0.2232,
+      "step": 324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998648229365927e-05,
+      "loss": 0.2422,
+      "step": 325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999859009663957e-05,
+      "loss": 0.2248,
+      "step": 326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998530740182574e-05,
+      "loss": 0.2088,
+      "step": 327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999847016000221e-05,
+      "loss": 0.2214,
+      "step": 328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999840835610589e-05,
+      "loss": 0.2136,
+      "step": 329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998345328501177e-05,
+      "loss": 0.227,
+      "step": 330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999828107719579e-05,
+      "loss": 0.2368,
+      "step": 331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999821560219759e-05,
+      "loss": 0.2111,
+      "step": 332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999814890351459e-05,
+      "loss": 0.2454,
+      "step": 333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998080981154956e-05,
+      "loss": 0.2355,
+      "step": 334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998011835126995e-05,
+      "loss": 0.1997,
+      "step": 335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997941465439176e-05,
+      "loss": 0.2264,
+      "step": 336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999786987210011e-05,
+      "loss": 0.2056,
+      "step": 337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999779705511856e-05,
+      "loss": 0.2334,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997723014503433e-05,
+      "loss": 0.1979,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997647750263795e-05,
+      "loss": 0.2189,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997571262408856e-05,
+      "loss": 0.2301,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997493550947983e-05,
+      "loss": 0.2156,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997414615890678e-05,
+      "loss": 0.2103,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999733445724661e-05,
+      "loss": 0.2384,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997253075025585e-05,
+      "loss": 0.2145,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999717046923756e-05,
+      "loss": 0.209,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997086639892657e-05,
+      "loss": 0.222,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997001587001125e-05,
+      "loss": 0.2061,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996915310573378e-05,
+      "loss": 0.2133,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996827810619972e-05,
+      "loss": 0.2137,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999673908715162e-05,
+      "loss": 0.233,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999664914017918e-05,
+      "loss": 0.2296,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996557969713662e-05,
+      "loss": 0.1884,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996465575766223e-05,
+      "loss": 0.2017,
+      "step": 354
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999637195834817e-05,
+      "loss": 0.2162,
+      "step": 355
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996277117470958e-05,
+      "loss": 0.1897,
+      "step": 356
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996181053146203e-05,
+      "loss": 0.2096,
+      "step": 357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996083765385656e-05,
+      "loss": 0.2274,
+      "step": 358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995985254201223e-05,
+      "loss": 0.2095,
+      "step": 359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995885519604967e-05,
+      "loss": 0.2122,
+      "step": 360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999578456160909e-05,
+      "loss": 0.2068,
+      "step": 361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999568238022595e-05,
+      "loss": 0.2237,
+      "step": 362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995578975468054e-05,
+      "loss": 0.2337,
+      "step": 363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995474347348053e-05,
+      "loss": 0.2123,
+      "step": 364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995368495878757e-05,
+      "loss": 0.2439,
+      "step": 365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995261421073117e-05,
+      "loss": 0.2153,
+      "step": 366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995153122944243e-05,
+      "loss": 0.2309,
+      "step": 367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999504360150539e-05,
+      "loss": 0.2098,
+      "step": 368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999493285676996e-05,
+      "loss": 0.2185,
+      "step": 369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99948208887515e-05,
+      "loss": 0.2374,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994707697463723e-05,
+      "loss": 0.2173,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994593282920484e-05,
+      "loss": 0.1991,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999447764513578e-05,
+      "loss": 0.2104,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994360784123766e-05,
+      "loss": 0.1968,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994242699898744e-05,
+      "loss": 0.2115,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994123392475168e-05,
+      "loss": 0.1939,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999400286186764e-05,
+      "loss": 0.2225,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999388110809091e-05,
+      "loss": 0.2091,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999375813115988e-05,
+      "loss": 0.2067,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993633931089605e-05,
+      "loss": 0.2328,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999350850789528e-05,
+      "loss": 0.2216,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993381861592263e-05,
+      "loss": 0.2268,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993253992196043e-05,
+      "loss": 0.2283,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993124899722282e-05,
+      "loss": 0.2204,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992994584186772e-05,
+      "loss": 0.2026,
+      "step": 385
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992863045605465e-05,
+      "loss": 0.1998,
+      "step": 386
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992730283994458e-05,
+      "loss": 0.2018,
+      "step": 387
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999259629937e-05,
+      "loss": 0.2202,
+      "step": 388
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992461091748493e-05,
+      "loss": 0.2001,
+      "step": 389
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992324661146485e-05,
+      "loss": 0.2038,
+      "step": 390
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992187007580666e-05,
+      "loss": 0.2207,
+      "step": 391
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992048131067895e-05,
+      "loss": 0.2181,
+      "step": 392
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991908031625164e-05,
+      "loss": 0.208,
+      "step": 393
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991766709269612e-05,
+      "loss": 0.2241,
+      "step": 394
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991624164018548e-05,
+      "loss": 0.2122,
+      "step": 395
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991480395889412e-05,
+      "loss": 0.2102,
+      "step": 396
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99913354048998e-05,
+      "loss": 0.2195,
+      "step": 397
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991189191067457e-05,
+      "loss": 0.2079,
+      "step": 398
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991041754410282e-05,
+      "loss": 0.2291,
+      "step": 399
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990893094946315e-05,
+      "loss": 0.213,
+      "step": 400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999074321269375e-05,
+      "loss": 0.2235,
+      "step": 401
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990592107670938e-05,
+      "loss": 0.2072,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990439779896368e-05,
+      "loss": 0.1993,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999028622938868e-05,
+      "loss": 0.2139,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990131456166675e-05,
+      "loss": 0.2222,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989975460249292e-05,
+      "loss": 0.2052,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989818241655622e-05,
+      "loss": 0.206,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998965980040491e-05,
+      "loss": 0.2049,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998950013651654e-05,
+      "loss": 0.2209,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989339250010064e-05,
+      "loss": 0.2169,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989177140905166e-05,
+      "loss": 0.2147,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998901380922169e-05,
+      "loss": 0.2207,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988849254979626e-05,
+      "loss": 0.1805,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988683478199112e-05,
+      "loss": 0.1891,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998851647890044e-05,
+      "loss": 0.224,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998834825710404e-05,
+      "loss": 0.217,
+      "step": 416
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988178812830515e-05,
+      "loss": 0.2176,
+      "step": 417
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988008146100594e-05,
+      "loss": 0.2028,
+      "step": 418
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987836256935166e-05,
+      "loss": 0.2268,
+      "step": 419
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987663145355272e-05,
+      "loss": 0.2502,
+      "step": 420
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987488811382098e-05,
+      "loss": 0.2156,
+      "step": 421
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998731325503698e-05,
+      "loss": 0.2053,
+      "step": 422
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99871364763414e-05,
+      "loss": 0.1849,
+      "step": 423
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986958475317003e-05,
+      "loss": 0.2156,
+      "step": 424
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986779251985565e-05,
+      "loss": 0.2063,
+      "step": 425
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998659880636903e-05,
+      "loss": 0.243,
+      "step": 426
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998641713848948e-05,
+      "loss": 0.2108,
+      "step": 427
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986234248369144e-05,
+      "loss": 0.2206,
+      "step": 428
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998605013603041e-05,
+      "loss": 0.1961,
+      "step": 429
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985864801495815e-05,
+      "loss": 0.2118,
+      "step": 430
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998567824478804e-05,
+      "loss": 0.2195,
+      "step": 431
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985490465929913e-05,
+      "loss": 0.1956,
+      "step": 432
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998530146494442e-05,
+      "loss": 0.2285,
+      "step": 433
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99851112418547e-05,
+      "loss": 0.2095,
+      "step": 434
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984919796684016e-05,
+      "loss": 0.2084,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998472712945582e-05,
+      "loss": 0.2043,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998453324019368e-05,
+      "loss": 0.1997,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984338128921327e-05,
+      "loss": 0.2134,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984141795662645e-05,
+      "loss": 0.2275,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998394424044166e-05,
+      "loss": 0.2047,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983745463282552e-05,
+      "loss": 0.1931,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998354546420965e-05,
+      "loss": 0.2009,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983344243247435e-05,
+      "loss": 0.2053,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983141800420525e-05,
+      "loss": 0.21,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982938135753708e-05,
+      "loss": 0.2347,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99827332492719e-05,
+      "loss": 0.2171,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982527141000185e-05,
+      "loss": 0.1927,
+      "step": 447
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982319810963786e-05,
+      "loss": 0.2112,
+      "step": 448
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998211125918808e-05,
+      "loss": 0.2076,
+      "step": 449
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998190148569859e-05,
+      "loss": 0.213,
+      "step": 450
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998169049052099e-05,
+      "loss": 0.2228,
+      "step": 451
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981478273681104e-05,
+      "loss": 0.2056,
+      "step": 452
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981264835204904e-05,
+      "loss": 0.2183,
+      "step": 453
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998105017511852e-05,
+      "loss": 0.2152,
+      "step": 454
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998083429344821e-05,
+      "loss": 0.2094,
+      "step": 455
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998061719022041e-05,
+      "loss": 0.217,
+      "step": 456
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980398865461684e-05,
+      "loss": 0.2132,
+      "step": 457
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980179319198757e-05,
+      "loss": 0.215,
+      "step": 458
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979958551458497e-05,
+      "loss": 0.1981,
+      "step": 459
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997973656226792e-05,
+      "loss": 0.2243,
+      "step": 460
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99795133516542e-05,
+      "loss": 0.1872,
+      "step": 461
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979288919644658e-05,
+      "loss": 0.1926,
+      "step": 462
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979063266266757e-05,
+      "loss": 0.2216,
+      "step": 463
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978836391548116e-05,
+      "loss": 0.1823,
+      "step": 464
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978608295516508e-05,
+      "loss": 0.218,
+      "step": 465
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978378978199843e-05,
+      "loss": 0.2017,
+      "step": 466
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978148439626185e-05,
+      "loss": 0.2121,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997791667982376e-05,
+      "loss": 0.2022,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997768369882092e-05,
+      "loss": 0.19,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997744949664619e-05,
+      "loss": 0.2154,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977214073328236e-05,
+      "loss": 0.2067,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997697742889586e-05,
+      "loss": 0.2003,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976739563378037e-05,
+      "loss": 0.2058,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976500476803866e-05,
+      "loss": 0.1888,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997626016920262e-05,
+      "loss": 0.2045,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997601864060371e-05,
+      "loss": 0.1891,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975775891036693e-05,
+      "loss": 0.1971,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975531920531277e-05,
+      "loss": 0.2048,
+      "step": 478
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975286729117328e-05,
+      "loss": 0.2146,
+      "step": 479
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975040316824848e-05,
+      "loss": 0.1926,
+      "step": 480
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974792683683997e-05,
+      "loss": 0.1915,
+      "step": 481
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974543829725094e-05,
+      "loss": 0.1853,
+      "step": 482
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997429375497858e-05,
+      "loss": 0.2111,
+      "step": 483
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997404245947507e-05,
+      "loss": 0.2177,
+      "step": 484
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997378994324532e-05,
+      "loss": 0.1775,
+      "step": 485
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973536206320237e-05,
+      "loss": 0.1944,
+      "step": 486
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997328124873087e-05,
+      "loss": 0.2258,
+      "step": 487
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997302507050843e-05,
+      "loss": 0.2142,
+      "step": 488
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972767671684265e-05,
+      "loss": 0.204,
+      "step": 489
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972509052289882e-05,
+      "loss": 0.2081,
+      "step": 490
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972249212356932e-05,
+      "loss": 0.1967,
+      "step": 491
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971988151917218e-05,
+      "loss": 0.2218,
+      "step": 492
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997172587100269e-05,
+      "loss": 0.2231,
+      "step": 493
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997146236964545e-05,
+      "loss": 0.2139,
+      "step": 494
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971197647877745e-05,
+      "loss": 0.2063,
+      "step": 495
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997093170573198e-05,
+      "loss": 0.2271,
+      "step": 496
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970664543240698e-05,
+      "loss": 0.1896,
+      "step": 497
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970396160436602e-05,
+      "loss": 0.1914,
+      "step": 498
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970126557352533e-05,
+      "loss": 0.2215,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969855734021497e-05,
+      "loss": 0.2065,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996958369047663e-05,
+      "loss": 0.2247,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996931042675123e-05,
+      "loss": 0.2203,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996903594287875e-05,
+      "loss": 0.2341,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968760238892776e-05,
+      "loss": 0.2034,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968483314827053e-05,
+      "loss": 0.2147,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996820517071548e-05,
+      "loss": 0.1886,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996792580659209e-05,
+      "loss": 0.2078,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967645222491074e-05,
+      "loss": 0.2103,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967363418446785e-05,
+      "loss": 0.2042,
+      "step": 509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99670803944937e-05,
+      "loss": 0.2004,
+      "step": 510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966796150666462e-05,
+      "loss": 0.1777,
+      "step": 511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966510686999865e-05,
+      "loss": 0.2163,
+      "step": 512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966224003528842e-05,
+      "loss": 0.2059,
+      "step": 513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996593610028848e-05,
+      "loss": 0.1917,
+      "step": 514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996564697731402e-05,
+      "loss": 0.1906,
+      "step": 515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965356634640845e-05,
+      "loss": 0.1979,
+      "step": 516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965065072304485e-05,
+      "loss": 0.1791,
+      "step": 517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964772290340633e-05,
+      "loss": 0.2225,
+      "step": 518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996447828878512e-05,
+      "loss": 0.1903,
+      "step": 519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996418306767393e-05,
+      "loss": 0.2092,
+      "step": 520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996388662704319e-05,
+      "loss": 0.1918,
+      "step": 521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963588966929184e-05,
+      "loss": 0.2298,
+      "step": 522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963290087368343e-05,
+      "loss": 0.1921,
+      "step": 523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996298998839725e-05,
+      "loss": 0.2075,
+      "step": 524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996268867005263e-05,
+      "loss": 0.182,
+      "step": 525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962386132371364e-05,
+      "loss": 0.2043,
+      "step": 526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962082375390478e-05,
+      "loss": 0.195,
+      "step": 527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961777399147152e-05,
+      "loss": 0.1951,
+      "step": 528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961471203678707e-05,
+      "loss": 0.1826,
+      "step": 529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961163789022625e-05,
+      "loss": 0.1961,
+      "step": 530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960855155216522e-05,
+      "loss": 0.2164,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996054530229818e-05,
+      "loss": 0.1898,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960234230305516e-05,
+      "loss": 0.1891,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959921939276606e-05,
+      "loss": 0.2048,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995960842924967e-05,
+      "loss": 0.189,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959293700263077e-05,
+      "loss": 0.1923,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958977752355352e-05,
+      "loss": 0.1773,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995866058556516e-05,
+      "loss": 0.1829,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995834219993132e-05,
+      "loss": 0.2084,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958022595492793e-05,
+      "loss": 0.1926,
+      "step": 540
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957701772288707e-05,
+      "loss": 0.2137,
+      "step": 541
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957379730358322e-05,
+      "loss": 0.2052,
+      "step": 542
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957056469741052e-05,
+      "loss": 0.2185,
+      "step": 543
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956731990476462e-05,
+      "loss": 0.2149,
+      "step": 544
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956406292604266e-05,
+      "loss": 0.2088,
+      "step": 545
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956079376164323e-05,
+      "loss": 0.1993,
+      "step": 546
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955751241196646e-05,
+      "loss": 0.2134,
+      "step": 547
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99554218877414e-05,
+      "loss": 0.205,
+      "step": 548
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955091315838888e-05,
+      "loss": 0.195,
+      "step": 549
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995475952552957e-05,
+      "loss": 0.2137,
+      "step": 550
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995442651685406e-05,
+      "loss": 0.1977,
+      "step": 551
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995409228985311e-05,
+      "loss": 0.2042,
+      "step": 552
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953756844567622e-05,
+      "loss": 0.1957,
+      "step": 553
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995342018103866e-05,
+      "loss": 0.188,
+      "step": 554
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953082299307428e-05,
+      "loss": 0.1968,
+      "step": 555
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995274319941527e-05,
+      "loss": 0.192,
+      "step": 556
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952402881403695e-05,
+      "loss": 0.1912,
+      "step": 557
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952061345314355e-05,
+      "loss": 0.1795,
+      "step": 558
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995171859118905e-05,
+      "loss": 0.1782,
+      "step": 559
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995137461906973e-05,
+      "loss": 0.1957,
+      "step": 560
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951029428998493e-05,
+      "loss": 0.2003,
+      "step": 561
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950683021017586e-05,
+      "loss": 0.2089,
+      "step": 562
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995033539516941e-05,
+      "loss": 0.1835,
+      "step": 563
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949986551496507e-05,
+      "loss": 0.1976,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949636490041575e-05,
+      "loss": 0.1972,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949285210847454e-05,
+      "loss": 0.2097,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994893271395714e-05,
+      "loss": 0.1826,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948578999413775e-05,
+      "loss": 0.1978,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948224067260652e-05,
+      "loss": 0.2004,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994786791754121e-05,
+      "loss": 0.1973,
+      "step": 570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947510550299033e-05,
+      "loss": 0.2069,
+      "step": 571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947151965577868e-05,
+      "loss": 0.2076,
+      "step": 572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99467921634216e-05,
+      "loss": 0.1872,
+      "step": 573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994643114387426e-05,
+      "loss": 0.1891,
+      "step": 574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946068906980035e-05,
+      "loss": 0.2059,
+      "step": 575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945705452783266e-05,
+      "loss": 0.1857,
+      "step": 576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994534078132843e-05,
+      "loss": 0.1874,
+      "step": 577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944974892660158e-05,
+      "loss": 0.2166,
+      "step": 578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944607786823238e-05,
+      "loss": 0.1953,
+      "step": 579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944239463862596e-05,
+      "loss": 0.1902,
+      "step": 580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943869923823307e-05,
+      "loss": 0.2066,
+      "step": 581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943499166750606e-05,
+      "loss": 0.1967,
+      "step": 582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943127192689868e-05,
+      "loss": 0.192,
+      "step": 583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942754001686618e-05,
+      "loss": 0.1907,
+      "step": 584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942379593786532e-05,
+      "loss": 0.2001,
+      "step": 585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942003969035434e-05,
+      "loss": 0.194,
+      "step": 586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941627127479295e-05,
+      "loss": 0.1793,
+      "step": 587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994124906916424e-05,
+      "loss": 0.1874,
+      "step": 588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940869794136535e-05,
+      "loss": 0.1977,
+      "step": 589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940489302442604e-05,
+      "loss": 0.2011,
+      "step": 590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940107594129013e-05,
+      "loss": 0.1781,
+      "step": 591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993972466924248e-05,
+      "loss": 0.1783,
+      "step": 592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939340527829872e-05,
+      "loss": 0.19,
+      "step": 593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99389551699382e-05,
+      "loss": 0.2083,
+      "step": 594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938568595614637e-05,
+      "loss": 0.1843,
+      "step": 595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938180804906492e-05,
+      "loss": 0.2269,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993779179786122e-05,
+      "loss": 0.2137,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993740157452644e-05,
+      "loss": 0.1845,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993701013494991e-05,
+      "loss": 0.1953,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936617479179533e-05,
+      "loss": 0.1825,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936223607263375e-05,
+      "loss": 0.2127,
+      "step": 601
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935828519249635e-05,
+      "loss": 0.1888,
+      "step": 602
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993543221518667e-05,
+      "loss": 0.1907,
+      "step": 603
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935034695122987e-05,
+      "loss": 0.1956,
+      "step": 604
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934635959107234e-05,
+      "loss": 0.2153,
+      "step": 605
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934236007188214e-05,
+      "loss": 0.1913,
+      "step": 606
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933834839414877e-05,
+      "loss": 0.1839,
+      "step": 607
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933432455836326e-05,
+      "loss": 0.2,
+      "step": 608
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933028856501804e-05,
+      "loss": 0.1819,
+      "step": 609
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932624041460706e-05,
+      "loss": 0.1971,
+      "step": 610
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932218010762585e-05,
+      "loss": 0.2075,
+      "step": 611
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993181076445713e-05,
+      "loss": 0.1833,
+      "step": 612
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993140230259418e-05,
+      "loss": 0.1965,
+      "step": 613
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930992625223737e-05,
+      "loss": 0.1856,
+      "step": 614
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930581732395935e-05,
+      "loss": 0.1977,
+      "step": 615
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930169624161065e-05,
+      "loss": 0.1966,
+      "step": 616
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992975630056956e-05,
+      "loss": 0.2072,
+      "step": 617
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929341761672017e-05,
+      "loss": 0.1961,
+      "step": 618
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992892600751916e-05,
+      "loss": 0.1904,
+      "step": 619
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928509038161886e-05,
+      "loss": 0.1751,
+      "step": 620
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928090853651218e-05,
+      "loss": 0.1918,
+      "step": 621
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992767145403834e-05,
+      "loss": 0.213,
+      "step": 622
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927250839374585e-05,
+      "loss": 0.1857,
+      "step": 623
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992682900971143e-05,
+      "loss": 0.201,
+      "step": 624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99264059651005e-05,
+      "loss": 0.1693,
+      "step": 625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992598170559358e-05,
+      "loss": 0.1928,
+      "step": 626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925556231242587e-05,
+      "loss": 0.1933,
+      "step": 627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925129542099598e-05,
+      "loss": 0.1894,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992470163821684e-05,
+      "loss": 0.2005,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924272519646675e-05,
+      "loss": 0.1842,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992384218644163e-05,
+      "loss": 0.1938,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923410638654372e-05,
+      "loss": 0.201,
+      "step": 632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922977876337718e-05,
+      "loss": 0.169,
+      "step": 633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992254389954463e-05,
+      "loss": 0.1767,
+      "step": 634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992210870832823e-05,
+      "loss": 0.2112,
+      "step": 635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992167230274178e-05,
+      "loss": 0.2047,
+      "step": 636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921234682838685e-05,
+      "loss": 0.2105,
+      "step": 637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920795848672512e-05,
+      "loss": 0.2113,
+      "step": 638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920355800296966e-05,
+      "loss": 0.1669,
+      "step": 639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919914537765908e-05,
+      "loss": 0.1985,
+      "step": 640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919472061133344e-05,
+      "loss": 0.1709,
+      "step": 641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919028370453427e-05,
+      "loss": 0.1777,
+      "step": 642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918583465780456e-05,
+      "loss": 0.1994,
+      "step": 643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918137347168894e-05,
+      "loss": 0.2031,
+      "step": 644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917690014673335e-05,
+      "loss": 0.1855,
+      "step": 645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917241468348527e-05,
+      "loss": 0.1933,
+      "step": 646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991679170824937e-05,
+      "loss": 0.179,
+      "step": 647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916340734430912e-05,
+      "loss": 0.1972,
+      "step": 648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915888546948345e-05,
+      "loss": 0.1901,
+      "step": 649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915435145857008e-05,
+      "loss": 0.1803,
+      "step": 650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914980531212403e-05,
+      "loss": 0.2031,
+      "step": 651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991452470307017e-05,
+      "loss": 0.1992,
+      "step": 652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914067661486087e-05,
+      "loss": 0.1924,
+      "step": 653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913609406516095e-05,
+      "loss": 0.1912,
+      "step": 654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913149938216286e-05,
+      "loss": 0.1896,
+      "step": 655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912689256642895e-05,
+      "loss": 0.2074,
+      "step": 656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912227361852298e-05,
+      "loss": 0.179,
+      "step": 657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911764253901032e-05,
+      "loss": 0.1834,
+      "step": 658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991129993284577e-05,
+      "loss": 0.2072,
+      "step": 659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910834398743347e-05,
+      "loss": 0.1838,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991036765165074e-05,
+      "loss": 0.1874,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909899691625073e-05,
+      "loss": 0.174,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909430518723612e-05,
+      "loss": 0.199,
+      "step": 663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908960133003793e-05,
+      "loss": 0.1962,
+      "step": 664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990848853452318e-05,
+      "loss": 0.1929,
+      "step": 665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990801572333949e-05,
+      "loss": 0.174,
+      "step": 666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990754169951059e-05,
+      "loss": 0.2118,
+      "step": 667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907066463094503e-05,
+      "loss": 0.2176,
+      "step": 668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906590014149387e-05,
+      "loss": 0.1952,
+      "step": 669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906112352733554e-05,
+      "loss": 0.1969,
+      "step": 670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990563347890547e-05,
+      "loss": 0.1988,
+      "step": 671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990515339272374e-05,
+      "loss": 0.2091,
+      "step": 672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990467209424713e-05,
+      "loss": 0.1918,
+      "step": 673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904189583534534e-05,
+      "loss": 0.1793,
+      "step": 674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903705860645015e-05,
+      "loss": 0.2032,
+      "step": 675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903220925637775e-05,
+      "loss": 0.2079,
+      "step": 676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902734778572165e-05,
+      "loss": 0.1832,
+      "step": 677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990224741950768e-05,
+      "loss": 0.1807,
+      "step": 678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901758848503974e-05,
+      "loss": 0.1895,
+      "step": 679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901269065620842e-05,
+      "loss": 0.193,
+      "step": 680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900778070918227e-05,
+      "loss": 0.2026,
+      "step": 681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900285864456227e-05,
+      "loss": 0.1925,
+      "step": 682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899792446295074e-05,
+      "loss": 0.2084,
+      "step": 683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899297816495166e-05,
+      "loss": 0.2063,
+      "step": 684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989880197511704e-05,
+      "loss": 0.1824,
+      "step": 685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898304922221375e-05,
+      "loss": 0.2102,
+      "step": 686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897806657869014e-05,
+      "loss": 0.163,
+      "step": 687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897307182120936e-05,
+      "loss": 0.1808,
+      "step": 688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989680649503827e-05,
+      "loss": 0.1794,
+      "step": 689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.98963045966823e-05,
+      "loss": 0.1675,
+      "step": 690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895801487114448e-05,
+      "loss": 0.2169,
+      "step": 691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895297166396296e-05,
+      "loss": 0.2004,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894791634589562e-05,
+      "loss": 0.2092,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894284891756124e-05,
+      "loss": 0.198,
+      "step": 694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893776937957997e-05,
+      "loss": 0.2007,
+      "step": 695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989326777325735e-05,
+      "loss": 0.1912,
+      "step": 696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989275739771651e-05,
+      "loss": 0.1926,
+      "step": 697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892245811397924e-05,
+      "loss": 0.1742,
+      "step": 698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891733014364223e-05,
+      "loss": 0.1902,
+      "step": 699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891219006678156e-05,
+      "loss": 0.1665,
+      "step": 700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890703788402635e-05,
+      "loss": 0.1864,
+      "step": 701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890187359600725e-05,
+      "loss": 0.2117,
+      "step": 702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889669720335625e-05,
+      "loss": 0.1952,
+      "step": 703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889150870670688e-05,
+      "loss": 0.2116,
+      "step": 704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888630810669423e-05,
+      "loss": 0.2067,
+      "step": 705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888109540395472e-05,
+      "loss": 0.1753,
+      "step": 706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988758705991264e-05,
+      "loss": 0.2044,
+      "step": 707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988706336928487e-05,
+      "loss": 0.1797,
+      "step": 708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988653846857626e-05,
+      "loss": 0.1994,
+      "step": 709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988601235785105e-05,
+      "loss": 0.2063,
+      "step": 710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988548503717363e-05,
+      "loss": 0.1874,
+      "step": 711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988495650660854e-05,
+      "loss": 0.2046,
+      "step": 712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884426766220465e-05,
+      "loss": 0.1759,
+      "step": 713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883895816074246e-05,
+      "loss": 0.1929,
+      "step": 714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883363656234865e-05,
+      "loss": 0.1954,
+      "step": 715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882830286767447e-05,
+      "loss": 0.1783,
+      "step": 716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882295707737275e-05,
+      "loss": 0.1864,
+      "step": 717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988175991920978e-05,
+      "loss": 0.1924,
+      "step": 718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988122292125053e-05,
+      "loss": 0.176,
+      "step": 719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880684713925254e-05,
+      "loss": 0.1834,
+      "step": 720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988014529729982e-05,
+      "loss": 0.208,
+      "step": 721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879604671440252e-05,
+      "loss": 0.1972,
+      "step": 722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987906283641271e-05,
+      "loss": 0.2087,
+      "step": 723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878519792283514e-05,
+      "loss": 0.1885,
+      "step": 724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987797553911913e-05,
+      "loss": 0.1827,
+      "step": 725
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877430076986166e-05,
+      "loss": 0.1845,
+      "step": 726
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.1787,
+      "step": 727
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987633552608168e-05,
+      "loss": 0.1971,
+      "step": 728
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875786437444123e-05,
+      "loss": 0.1864,
+      "step": 729
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875236140105908e-05,
+      "loss": 0.1969,
+      "step": 730
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874684634134395e-05,
+      "loss": 0.1856,
+      "step": 731
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874131919597076e-05,
+      "loss": 0.185,
+      "step": 732
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873577996561594e-05,
+      "loss": 0.172,
+      "step": 733
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873022865095753e-05,
+      "loss": 0.1976,
+      "step": 734
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872466525267492e-05,
+      "loss": 0.1743,
+      "step": 735
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98719089771449e-05,
+      "loss": 0.187,
+      "step": 736
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871350220796215e-05,
+      "loss": 0.1981,
+      "step": 737
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870790256289827e-05,
+      "loss": 0.1831,
+      "step": 738
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870229083694266e-05,
+      "loss": 0.1887,
+      "step": 739
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986966670307822e-05,
+      "loss": 0.1977,
+      "step": 740
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869103114510514e-05,
+      "loss": 0.1739,
+      "step": 741
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868538318060128e-05,
+      "loss": 0.1924,
+      "step": 742
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867972313796183e-05,
+      "loss": 0.1874,
+      "step": 743
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986740510178796e-05,
+      "loss": 0.1962,
+      "step": 744
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866836682104875e-05,
+      "loss": 0.1857,
+      "step": 745
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866267054816496e-05,
+      "loss": 0.1771,
+      "step": 746
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865696219992547e-05,
+      "loss": 0.1958,
+      "step": 747
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865124177702887e-05,
+      "loss": 0.1856,
+      "step": 748
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864550928017526e-05,
+      "loss": 0.1867,
+      "step": 749
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863976471006632e-05,
+      "loss": 0.1944,
+      "step": 750
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863400806740505e-05,
+      "loss": 0.1982,
+      "step": 751
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862823935289605e-05,
+      "loss": 0.2074,
+      "step": 752
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862245856724536e-05,
+      "loss": 0.1802,
+      "step": 753
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861666571116048e-05,
+      "loss": 0.1828,
+      "step": 754
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986108607853504e-05,
+      "loss": 0.1945,
+      "step": 755
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860504379052558e-05,
+      "loss": 0.1823,
+      "step": 756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859921472739798e-05,
+      "loss": 0.1805,
+      "step": 757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859337359668102e-05,
+      "loss": 0.1706,
+      "step": 758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858752039908956e-05,
+      "loss": 0.1752,
+      "step": 759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858165513534002e-05,
+      "loss": 0.1877,
+      "step": 760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857577780615025e-05,
+      "loss": 0.1871,
+      "step": 761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856988841223955e-05,
+      "loss": 0.1908,
+      "step": 762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856398695432877e-05,
+      "loss": 0.177,
+      "step": 763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855807343314014e-05,
+      "loss": 0.2029,
+      "step": 764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985521478493974e-05,
+      "loss": 0.1761,
+      "step": 765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854621020382586e-05,
+      "loss": 0.1952,
+      "step": 766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854026049715218e-05,
+      "loss": 0.182,
+      "step": 767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853429873010456e-05,
+      "loss": 0.1905,
+      "step": 768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852832490341266e-05,
+      "loss": 0.1568,
+      "step": 769
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852233901780762e-05,
+      "loss": 0.2022,
+      "step": 770
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851634107402207e-05,
+      "loss": 0.1842,
+      "step": 771
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851033107279007e-05,
+      "loss": 0.1794,
+      "step": 772
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850430901484722e-05,
+      "loss": 0.186,
+      "step": 773
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984982749009305e-05,
+      "loss": 0.1767,
+      "step": 774
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984922287317785e-05,
+      "loss": 0.1845,
+      "step": 775
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984861705081312e-05,
+      "loss": 0.1853,
+      "step": 776
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848010023073002e-05,
+      "loss": 0.1978,
+      "step": 777
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847401790031792e-05,
+      "loss": 0.1819,
+      "step": 778
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846792351763936e-05,
+      "loss": 0.2024,
+      "step": 779
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846181708344022e-05,
+      "loss": 0.1896,
+      "step": 780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984556985984678e-05,
+      "loss": 0.1885,
+      "step": 781
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844956806347106e-05,
+      "loss": 0.1723,
+      "step": 782
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984434254792002e-05,
+      "loss": 0.1967,
+      "step": 783
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843727084640707e-05,
+      "loss": 0.1989,
+      "step": 784
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843110416584495e-05,
+      "loss": 0.1819,
+      "step": 785
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842492543826858e-05,
+      "loss": 0.1666,
+      "step": 786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841873466443416e-05,
+      "loss": 0.1875,
+      "step": 787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984125318450994e-05,
+      "loss": 0.18,
+      "step": 788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984063169810234e-05,
+      "loss": 0.167,
+      "step": 789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984000900729669e-05,
+      "loss": 0.1838,
+      "step": 790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839385112169193e-05,
+      "loss": 0.1734,
+      "step": 791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983876001279621e-05,
+      "loss": 0.1918,
+      "step": 792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983813370925425e-05,
+      "loss": 0.1648,
+      "step": 793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983750620161996e-05,
+      "loss": 0.1609,
+      "step": 794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836877489970156e-05,
+      "loss": 0.1997,
+      "step": 795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836247574381766e-05,
+      "loss": 0.1851,
+      "step": 796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98356164549319e-05,
+      "loss": 0.1894,
+      "step": 797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834984131697796e-05,
+      "loss": 0.1888,
+      "step": 798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983435060475684e-05,
+      "loss": 0.1901,
+      "step": 799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833715874186578e-05,
+      "loss": 0.1871,
+      "step": 800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833079940064687e-05,
+      "loss": 0.2028,
+      "step": 801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832442802469004e-05,
+      "loss": 0.1967,
+      "step": 802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831804461477512e-05,
+      "loss": 0.1834,
+      "step": 803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831164917168328e-05,
+      "loss": 0.1953,
+      "step": 804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830524169619735e-05,
+      "loss": 0.1879,
+      "step": 805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829882218910145e-05,
+      "loss": 0.1619,
+      "step": 806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829239065118138e-05,
+      "loss": 0.1854,
+      "step": 807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982859470832242e-05,
+      "loss": 0.1719,
+      "step": 808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982794914860186e-05,
+      "loss": 0.1831,
+      "step": 809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827302386035467e-05,
+      "loss": 0.1724,
+      "step": 810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826654420702395e-05,
+      "loss": 0.202,
+      "step": 811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982600525268195e-05,
+      "loss": 0.1713,
+      "step": 812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825354882053588e-05,
+      "loss": 0.1913,
+      "step": 813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824703308896905e-05,
+      "loss": 0.1804,
+      "step": 814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824050533291648e-05,
+      "loss": 0.1781,
+      "step": 815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982339655531771e-05,
+      "loss": 0.1774,
+      "step": 816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822741375055128e-05,
+      "loss": 0.1687,
+      "step": 817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822084992584098e-05,
+      "loss": 0.1753,
+      "step": 818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821427407984946e-05,
+      "loss": 0.17,
+      "step": 819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820768621338156e-05,
+      "loss": 0.1951,
+      "step": 820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820108632724367e-05,
+      "loss": 0.1883,
+      "step": 821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819447442224342e-05,
+      "loss": 0.1855,
+      "step": 822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818785049919012e-05,
+      "loss": 0.1951,
+      "step": 823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818121455889443e-05,
+      "loss": 0.1886,
+      "step": 824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817456660216857e-05,
+      "loss": 0.1792,
+      "step": 825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816790662982614e-05,
+      "loss": 0.1694,
+      "step": 826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981612346426823e-05,
+      "loss": 0.1818,
+      "step": 827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981545506415536e-05,
+      "loss": 0.1833,
+      "step": 828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981478546272581e-05,
+      "loss": 0.1992,
+      "step": 829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814114660061533e-05,
+      "loss": 0.186,
+      "step": 830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813442656244635e-05,
+      "loss": 0.1563,
+      "step": 831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812769451357352e-05,
+      "loss": 0.1951,
+      "step": 832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812095045482084e-05,
+      "loss": 0.2203,
+      "step": 833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811419438701372e-05,
+      "loss": 0.2029,
+      "step": 834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810742631097904e-05,
+      "loss": 0.1677,
+      "step": 835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810064622754513e-05,
+      "loss": 0.1649,
+      "step": 836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809385413754178e-05,
+      "loss": 0.1733,
+      "step": 837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808705004180032e-05,
+      "loss": 0.1839,
+      "step": 838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980802339411535e-05,
+      "loss": 0.1809,
+      "step": 839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980734058364355e-05,
+      "loss": 0.1924,
+      "step": 840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980665657284821e-05,
+      "loss": 0.1759,
+      "step": 841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980597136181304e-05,
+      "loss": 0.1793,
+      "step": 842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805284950621907e-05,
+      "loss": 0.2002,
+      "step": 843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804597339358812e-05,
+      "loss": 0.18,
+      "step": 844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803908528107923e-05,
+      "loss": 0.1652,
+      "step": 845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803218516953538e-05,
+      "loss": 0.1873,
+      "step": 846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802527305980108e-05,
+      "loss": 0.1705,
+      "step": 847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801834895272233e-05,
+      "loss": 0.1849,
+      "step": 848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801141284914654e-05,
+      "loss": 0.1771,
+      "step": 849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980044647499227e-05,
+      "loss": 0.1913,
+      "step": 850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799750465590107e-05,
+      "loss": 0.1839,
+      "step": 851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799053256793355e-05,
+      "loss": 0.1769,
+      "step": 852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798354848687353e-05,
+      "loss": 0.1911,
+      "step": 853
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797655241357568e-05,
+      "loss": 0.1806,
+      "step": 854
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796954434889634e-05,
+      "loss": 0.1831,
+      "step": 855
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796252429369316e-05,
+      "loss": 0.1768,
+      "step": 856
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795549224882536e-05,
+      "loss": 0.192,
+      "step": 857
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979484482151536e-05,
+      "loss": 0.1998,
+      "step": 858
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794139219354e-05,
+      "loss": 0.1872,
+      "step": 859
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979343241848481e-05,
+      "loss": 0.1753,
+      "step": 860
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792724418994304e-05,
+      "loss": 0.1692,
+      "step": 861
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792015220969128e-05,
+      "loss": 0.163,
+      "step": 862
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791304824496086e-05,
+      "loss": 0.1741,
+      "step": 863
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790593229662115e-05,
+      "loss": 0.1892,
+      "step": 864
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789880436554317e-05,
+      "loss": 0.1775,
+      "step": 865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789166445259925e-05,
+      "loss": 0.1707,
+      "step": 866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978845125586633e-05,
+      "loss": 0.201,
+      "step": 867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787734868461058e-05,
+      "loss": 0.171,
+      "step": 868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787017283131793e-05,
+      "loss": 0.1754,
+      "step": 869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786298499966356e-05,
+      "loss": 0.1794,
+      "step": 870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785578519052722e-05,
+      "loss": 0.1723,
+      "step": 871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784857340479008e-05,
+      "loss": 0.1764,
+      "step": 872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978413496433348e-05,
+      "loss": 0.1729,
+      "step": 873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783411390704553e-05,
+      "loss": 0.1833,
+      "step": 874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978268661968078e-05,
+      "loss": 0.1794,
+      "step": 875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781960651350873e-05,
+      "loss": 0.1728,
+      "step": 876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781233485803675e-05,
+      "loss": 0.1792,
+      "step": 877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780505123128187e-05,
+      "loss": 0.1815,
+      "step": 878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9779775563413555e-05,
+      "loss": 0.19,
+      "step": 879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779044806749072e-05,
+      "loss": 0.1742,
+      "step": 880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977831285322417e-05,
+      "loss": 0.1718,
+      "step": 881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977757970292844e-05,
+      "loss": 0.1664,
+      "step": 882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776845355951603e-05,
+      "loss": 0.1835,
+      "step": 883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776109812383547e-05,
+      "loss": 0.1841,
+      "step": 884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977537307231429e-05,
+      "loss": 0.1745,
+      "step": 885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774635135834e-05,
+      "loss": 0.1984,
+      "step": 886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773896003032994e-05,
+      "loss": 0.1662,
+      "step": 887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773155674001737e-05,
+      "loss": 0.1898,
+      "step": 888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977241414883084e-05,
+      "loss": 0.1806,
+      "step": 889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977167142761105e-05,
+      "loss": 0.199,
+      "step": 890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770927510433277e-05,
+      "loss": 0.1927,
+      "step": 891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770182397388567e-05,
+      "loss": 0.204,
+      "step": 892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769436088568112e-05,
+      "loss": 0.1929,
+      "step": 893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768688584063258e-05,
+      "loss": 0.1724,
+      "step": 894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976793988396549e-05,
+      "loss": 0.1705,
+      "step": 895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767189988366438e-05,
+      "loss": 0.18,
+      "step": 896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976643889735789e-05,
+      "loss": 0.1775,
+      "step": 897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976568661103176e-05,
+      "loss": 0.166,
+      "step": 898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764933129480134e-05,
+      "loss": 0.1623,
+      "step": 899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764178452795227e-05,
+      "loss": 0.1603,
+      "step": 900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763422581069397e-05,
+      "loss": 0.1858,
+      "step": 901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976266551439516e-05,
+      "loss": 0.1657,
+      "step": 902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761907252865176e-05,
+      "loss": 0.1623,
+      "step": 903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9761147796572245e-05,
+      "loss": 0.1924,
+      "step": 904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760387145609323e-05,
+      "loss": 0.1886,
+      "step": 905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759625300069497e-05,
+      "loss": 0.2049,
+      "step": 906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758862260046017e-05,
+      "loss": 0.1591,
+      "step": 907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975809802563227e-05,
+      "loss": 0.1887,
+      "step": 908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975733259692179e-05,
+      "loss": 0.1678,
+      "step": 909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756565974008262e-05,
+      "loss": 0.1719,
+      "step": 910
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755798156985505e-05,
+      "loss": 0.1805,
+      "step": 911
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755029145947496e-05,
+      "loss": 0.1709,
+      "step": 912
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975425894098836e-05,
+      "loss": 0.1775,
+      "step": 913
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753487542202355e-05,
+      "loss": 0.1782,
+      "step": 914
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752714949683897e-05,
+      "loss": 0.1817,
+      "step": 915
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751941163527542e-05,
+      "loss": 0.1831,
+      "step": 916
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751166183827998e-05,
+      "loss": 0.1792,
+      "step": 917
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975039001068011e-05,
+      "loss": 0.1808,
+      "step": 918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749612644178874e-05,
+      "loss": 0.1868,
+      "step": 919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748834084419433e-05,
+      "loss": 0.1663,
+      "step": 920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748054331497077e-05,
+      "loss": 0.1939,
+      "step": 921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974727338550724e-05,
+      "loss": 0.1757,
+      "step": 922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746491246545504e-05,
+      "loss": 0.191,
+      "step": 923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974570791470759e-05,
+      "loss": 0.1961,
+      "step": 924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744923390089373e-05,
+      "loss": 0.1968,
+      "step": 925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744137672786873e-05,
+      "loss": 0.1811,
+      "step": 926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974335076289625e-05,
+      "loss": 0.1975,
+      "step": 927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742562660513815e-05,
+      "loss": 0.1743,
+      "step": 928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974177336573603e-05,
+      "loss": 0.1644,
+      "step": 929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974098287865949e-05,
+      "loss": 0.1698,
+      "step": 930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974019119938095e-05,
+      "loss": 0.176,
+      "step": 931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739398327997294e-05,
+      "loss": 0.1772,
+      "step": 932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973860426460557e-05,
+      "loss": 0.2065,
+      "step": 933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973780900930296e-05,
+      "loss": 0.1738,
+      "step": 934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97370125621868e-05,
+      "loss": 0.1693,
+      "step": 935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736214923354557e-05,
+      "loss": 0.1828,
+      "step": 936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735416092903866e-05,
+      "loss": 0.1671,
+      "step": 937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973461607093249e-05,
+      "loss": 0.1881,
+      "step": 938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973381485753835e-05,
+      "loss": 0.1604,
+      "step": 939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733012452819496e-05,
+      "loss": 0.1692,
+      "step": 940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973220885687414e-05,
+      "loss": 0.1785,
+      "step": 941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973140406980064e-05,
+      "loss": 0.1699,
+      "step": 942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730598091697488e-05,
+      "loss": 0.1776,
+      "step": 943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729790922663328e-05,
+      "loss": 0.1885,
+      "step": 944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972898256279695e-05,
+      "loss": 0.1783,
+      "step": 945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728173012197293e-05,
+      "loss": 0.1844,
+      "step": 946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972736227096343e-05,
+      "loss": 0.1973,
+      "step": 947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97265503391946e-05,
+      "loss": 0.1831,
+      "step": 948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725737216990164e-05,
+      "loss": 0.1768,
+      "step": 949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724922904449646e-05,
+      "loss": 0.1721,
+      "step": 950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724107401672706e-05,
+      "loss": 0.148,
+      "step": 951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723290708759163e-05,
+      "loss": 0.1974,
+      "step": 952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972247282580896e-05,
+      "loss": 0.1793,
+      "step": 953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972165375292221e-05,
+      "loss": 0.1922,
+      "step": 954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972083349019915e-05,
+      "loss": 0.1881,
+      "step": 955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720012037740173e-05,
+      "loss": 0.164,
+      "step": 956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719189395645826e-05,
+      "loss": 0.1828,
+      "step": 957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718365564016785e-05,
+      "loss": 0.1796,
+      "step": 958
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717540542953877e-05,
+      "loss": 0.1878,
+      "step": 959
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971671433255808e-05,
+      "loss": 0.1707,
+      "step": 960
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971588693293051e-05,
+      "loss": 0.1832,
+      "step": 961
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715058344172443e-05,
+      "loss": 0.1713,
+      "step": 962
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714228566385282e-05,
+      "loss": 0.1674,
+      "step": 963
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713397599670584e-05,
+      "loss": 0.1825,
+      "step": 964
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712565444130053e-05,
+      "loss": 0.1799,
+      "step": 965
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971173209986554e-05,
+      "loss": 0.1897,
+      "step": 966
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710897566979033e-05,
+      "loss": 0.1861,
+      "step": 967
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971006184557267e-05,
+      "loss": 0.1912,
+      "step": 968
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970922493574874e-05,
+      "loss": 0.1689,
+      "step": 969
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970838683760967e-05,
+      "loss": 0.1802,
+      "step": 970
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707547551258037e-05,
+      "loss": 0.1728,
+      "step": 971
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706707076796555e-05,
+      "loss": 0.1847,
+      "step": 972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705865414328102e-05,
+      "loss": 0.171,
+      "step": 973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705022563955677e-05,
+      "loss": 0.1707,
+      "step": 974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704178525782444e-05,
+      "loss": 0.182,
+      "step": 975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703333299911705e-05,
+      "loss": 0.1652,
+      "step": 976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702486886446903e-05,
+      "loss": 0.178,
+      "step": 977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701639285491633e-05,
+      "loss": 0.1821,
+      "step": 978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9700790497149636e-05,
+      "loss": 0.1802,
+      "step": 979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699940521524793e-05,
+      "loss": 0.1653,
+      "step": 980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699089358721132e-05,
+      "loss": 0.1801,
+      "step": 981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698237008842826e-05,
+      "loss": 0.1775,
+      "step": 982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96973834719942e-05,
+      "loss": 0.1706,
+      "step": 983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969652874827971e-05,
+      "loss": 0.1829,
+      "step": 984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695672837803976e-05,
+      "loss": 0.1803,
+      "step": 985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694815740671746e-05,
+      "loss": 0.1656,
+      "step": 986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693957456987923e-05,
+      "loss": 0.1839,
+      "step": 987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969309798685755e-05,
+      "loss": 0.1844,
+      "step": 988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692237330385823e-05,
+      "loss": 0.1921,
+      "step": 989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969137548767807e-05,
+      "loss": 0.183,
+      "step": 990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969051245883978e-05,
+      "loss": 0.1891,
+      "step": 991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968964824397658e-05,
+      "loss": 0.1827,
+      "step": 992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688782843194236e-05,
+      "loss": 0.1945,
+      "step": 993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687916256598662e-05,
+      "loss": 0.1765,
+      "step": 994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968704848429593e-05,
+      "loss": 0.1839,
+      "step": 995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686179526392237e-05,
+      "loss": 0.1726,
+      "step": 996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685309382993944e-05,
+      "loss": 0.1987,
+      "step": 997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968443805420754e-05,
+      "loss": 0.1895,
+      "step": 998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683565540139673e-05,
+      "loss": 0.1845,
+      "step": 999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682691840897128e-05,
+      "loss": 0.1698,
+      "step": 1000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681816956586836e-05,
+      "loss": 0.173,
+      "step": 1001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968094088731588e-05,
+      "loss": 0.1721,
+      "step": 1002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680063633191474e-05,
+      "loss": 0.1632,
+      "step": 1003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679185194320988e-05,
+      "loss": 0.1758,
+      "step": 1004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678305570811937e-05,
+      "loss": 0.1737,
+      "step": 1005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677424762771978e-05,
+      "loss": 0.1824,
+      "step": 1006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676542770308915e-05,
+      "loss": 0.1743,
+      "step": 1007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967565959353069e-05,
+      "loss": 0.1859,
+      "step": 1008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96747752325454e-05,
+      "loss": 0.1769,
+      "step": 1009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967388968746128e-05,
+      "loss": 0.1966,
+      "step": 1010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673002958386712e-05,
+      "loss": 0.1975,
+      "step": 1011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672115045430228e-05,
+      "loss": 0.187,
+      "step": 1012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671225948700493e-05,
+      "loss": 0.196,
+      "step": 1013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670335668306325e-05,
+      "loss": 0.1704,
+      "step": 1014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966944420435669e-05,
+      "loss": 0.1682,
+      "step": 1015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966855155696069e-05,
+      "loss": 0.1843,
+      "step": 1016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966765772622758e-05,
+      "loss": 0.1657,
+      "step": 1017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966676271226675e-05,
+      "loss": 0.1891,
+      "step": 1018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966586651518775e-05,
+      "loss": 0.1879,
+      "step": 1019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664969135100262e-05,
+      "loss": 0.1833,
+      "step": 1020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664070572114115e-05,
+      "loss": 0.164,
+      "step": 1021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663170826339285e-05,
+      "loss": 0.1896,
+      "step": 1022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966226989788589e-05,
+      "loss": 0.1867,
+      "step": 1023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.96613677868642e-05,
+      "loss": 0.1666,
+      "step": 1024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966046449338462e-05,
+      "loss": 0.1719,
+      "step": 1025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965956001755771e-05,
+      "loss": 0.1643,
+      "step": 1026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658654359494165e-05,
+      "loss": 0.1904,
+      "step": 1027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965774751930483e-05,
+      "loss": 0.1637,
+      "step": 1028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965683949710069e-05,
+      "loss": 0.1692,
+      "step": 1029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655930292992882e-05,
+      "loss": 0.1731,
+      "step": 1030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965501990709268e-05,
+      "loss": 0.1942,
+      "step": 1031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654108339511518e-05,
+      "loss": 0.1795,
+      "step": 1032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653195590360947e-05,
+      "loss": 0.1899,
+      "step": 1033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652281659752688e-05,
+      "loss": 0.1857,
+      "step": 1034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651366547798596e-05,
+      "loss": 0.1766,
+      "step": 1035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650450254610668e-05,
+      "loss": 0.2082,
+      "step": 1036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649532780301055e-05,
+      "loss": 0.1902,
+      "step": 1037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648614124982044e-05,
+      "loss": 0.1752,
+      "step": 1038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964769428876607e-05,
+      "loss": 0.1668,
+      "step": 1039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646773271765712e-05,
+      "loss": 0.1865,
+      "step": 1040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645851074093692e-05,
+      "loss": 0.1934,
+      "step": 1041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644927695862882e-05,
+      "loss": 0.1717,
+      "step": 1042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644003137186288e-05,
+      "loss": 0.1773,
+      "step": 1043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643077398177075e-05,
+      "loss": 0.1873,
+      "step": 1044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642150478948542e-05,
+      "loss": 0.1816,
+      "step": 1045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964122237961413e-05,
+      "loss": 0.1842,
+      "step": 1046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640293100287434e-05,
+      "loss": 0.1911,
+      "step": 1047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963936264108219e-05,
+      "loss": 0.163,
+      "step": 1048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638431002112273e-05,
+      "loss": 0.1785,
+      "step": 1049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637498183491708e-05,
+      "loss": 0.1871,
+      "step": 1050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636564185334664e-05,
+      "loss": 0.1647,
+      "step": 1051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635629007755456e-05,
+      "loss": 0.1917,
+      "step": 1052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963469265086853e-05,
+      "loss": 0.1799,
+      "step": 1053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96337551147885e-05,
+      "loss": 0.1603,
+      "step": 1054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632816399630107e-05,
+      "loss": 0.1937,
+      "step": 1055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631876505508235e-05,
+      "loss": 0.1617,
+      "step": 1056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630935432537928e-05,
+      "loss": 0.194,
+      "step": 1057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629993180834356e-05,
+      "loss": 0.173,
+      "step": 1058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629049750512844e-05,
+      "loss": 0.1821,
+      "step": 1059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962810514168886e-05,
+      "loss": 0.1652,
+      "step": 1060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627159354478015e-05,
+      "loss": 0.1816,
+      "step": 1061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962621238899606e-05,
+      "loss": 0.183,
+      "step": 1062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625264245358897e-05,
+      "loss": 0.1792,
+      "step": 1063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624314923682574e-05,
+      "loss": 0.1724,
+      "step": 1064
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962336442408327e-05,
+      "loss": 0.1854,
+      "step": 1065
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622412746677327e-05,
+      "loss": 0.1731,
+      "step": 1066
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621459891581215e-05,
+      "loss": 0.1731,
+      "step": 1067
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620505858911554e-05,
+      "loss": 0.1667,
+      "step": 1068
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619550648785107e-05,
+      "loss": 0.2023,
+      "step": 1069
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618594261318788e-05,
+      "loss": 0.177,
+      "step": 1070
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961763669662965e-05,
+      "loss": 0.1647,
+      "step": 1071
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961667795483488e-05,
+      "loss": 0.1825,
+      "step": 1072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615718036051828e-05,
+      "loss": 0.1725,
+      "step": 1073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614756940397974e-05,
+      "loss": 0.1694,
+      "step": 1074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961379466799095e-05,
+      "loss": 0.1872,
+      "step": 1075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612831218948528e-05,
+      "loss": 0.1779,
+      "step": 1076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611866593388624e-05,
+      "loss": 0.1981,
+      "step": 1077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96109007914293e-05,
+      "loss": 0.1762,
+      "step": 1078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609933813188757e-05,
+      "loss": 0.1623,
+      "step": 1079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960896565878535e-05,
+      "loss": 0.1655,
+      "step": 1080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607996328337565e-05,
+      "loss": 0.1994,
+      "step": 1081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607025821964044e-05,
+      "loss": 0.1674,
+      "step": 1082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606054139783563e-05,
+      "loss": 0.1727,
+      "step": 1083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605081281915053e-05,
+      "loss": 0.1856,
+      "step": 1084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604107248477576e-05,
+      "loss": 0.1754,
+      "step": 1085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960313203959035e-05,
+      "loss": 0.1488,
+      "step": 1086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602155655372722e-05,
+      "loss": 0.1856,
+      "step": 1087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601178095944202e-05,
+      "loss": 0.1869,
+      "step": 1088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960019936142443e-05,
+      "loss": 0.1595,
+      "step": 1089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959921945193319e-05,
+      "loss": 0.1667,
+      "step": 1090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598238367590417e-05,
+      "loss": 0.1565,
+      "step": 1091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959725610851619e-05,
+      "loss": 0.1851,
+      "step": 1092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959627267483072e-05,
+      "loss": 0.1733,
+      "step": 1093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959528806665438e-05,
+      "loss": 0.1597,
+      "step": 1094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594302284107663e-05,
+      "loss": 0.1716,
+      "step": 1095
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593315327311232e-05,
+      "loss": 0.1678,
+      "step": 1096
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959232719638587e-05,
+      "loss": 0.1774,
+      "step": 1097
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591337891452525e-05,
+      "loss": 0.1766,
+      "step": 1098
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959034741263227e-05,
+      "loss": 0.1777,
+      "step": 1099
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589355760046332e-05,
+      "loss": 0.1773,
+      "step": 1100
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588362933816084e-05,
+      "loss": 0.1807,
+      "step": 1101
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587368934063035e-05,
+      "loss": 0.18,
+      "step": 1102
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958637376090884e-05,
+      "loss": 0.159,
+      "step": 1103
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585377414475303e-05,
+      "loss": 0.1844,
+      "step": 1104
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584379894884364e-05,
+      "loss": 0.18,
+      "step": 1105
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583381202258106e-05,
+      "loss": 0.1825,
+      "step": 1106
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582381336718767e-05,
+      "loss": 0.1706,
+      "step": 1107
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958138029838872e-05,
+      "loss": 0.1655,
+      "step": 1108
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958037808739047e-05,
+      "loss": 0.1722,
+      "step": 1109
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579374703846693e-05,
+      "loss": 0.1681,
+      "step": 1110
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957837014788019e-05,
+      "loss": 0.1619,
+      "step": 1111
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577364419613903e-05,
+      "loss": 0.1904,
+      "step": 1112
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576357519170925e-05,
+      "loss": 0.1747,
+      "step": 1113
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575349446674496e-05,
+      "loss": 0.1638,
+      "step": 1114
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574340202247994e-05,
+      "loss": 0.1525,
+      "step": 1115
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573329786014934e-05,
+      "loss": 0.1675,
+      "step": 1116
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957231819809899e-05,
+      "loss": 0.1788,
+      "step": 1117
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957130543862396e-05,
+      "loss": 0.1636,
+      "step": 1118
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570291507713802e-05,
+      "loss": 0.1753,
+      "step": 1119
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569276405492616e-05,
+      "loss": 0.1743,
+      "step": 1120
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956826013208463e-05,
+      "loss": 0.1746,
+      "step": 1121
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567242687614238e-05,
+      "loss": 0.1681,
+      "step": 1122
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566224072205955e-05,
+      "loss": 0.1846,
+      "step": 1123
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565204285984454e-05,
+      "loss": 0.1823,
+      "step": 1124
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564183329074548e-05,
+      "loss": 0.1791,
+      "step": 1125
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956316120160119e-05,
+      "loss": 0.1862,
+      "step": 1126
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562137903689478e-05,
+      "loss": 0.1703,
+      "step": 1127
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956111343546466e-05,
+      "loss": 0.1766,
+      "step": 1128
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956008779705211e-05,
+      "loss": 0.1825,
+      "step": 1129
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559060988577365e-05,
+      "loss": 0.1733,
+      "step": 1130
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558033010166095e-05,
+      "loss": 0.1913,
+      "step": 1131
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557003861944116e-05,
+      "loss": 0.1868,
+      "step": 1132
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955597354403738e-05,
+      "loss": 0.18,
+      "step": 1133
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554942056571992e-05,
+      "loss": 0.1548,
+      "step": 1134
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553909399674194e-05,
+      "loss": 0.1711,
+      "step": 1135
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552875573470374e-05,
+      "loss": 0.1632,
+      "step": 1136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551840578087065e-05,
+      "loss": 0.18,
+      "step": 1137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550804413650935e-05,
+      "loss": 0.1796,
+      "step": 1138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954976708028881e-05,
+      "loss": 0.1849,
+      "step": 1139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548728578127635e-05,
+      "loss": 0.1694,
+      "step": 1140
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547688907294527e-05,
+      "loss": 0.1678,
+      "step": 1141
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546648067916726e-05,
+      "loss": 0.184,
+      "step": 1142
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545606060121617e-05,
+      "loss": 0.1747,
+      "step": 1143
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544562884036735e-05,
+      "loss": 0.1686,
+      "step": 1144
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543518539789756e-05,
+      "loss": 0.1692,
+      "step": 1145
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542473027508494e-05,
+      "loss": 0.1752,
+      "step": 1146
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541426347320917e-05,
+      "loss": 0.1597,
+      "step": 1147
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540378499355118e-05,
+      "loss": 0.1789,
+      "step": 1148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953932948373935e-05,
+      "loss": 0.1693,
+      "step": 1149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538279300602e-05,
+      "loss": 0.1845,
+      "step": 1150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537227950071602e-05,
+      "loss": 0.1757,
+      "step": 1151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536175432276833e-05,
+      "loss": 0.1814,
+      "step": 1152
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535121747346507e-05,
+      "loss": 0.157,
+      "step": 1153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534066895409584e-05,
+      "loss": 0.1692,
+      "step": 1154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533010876595175e-05,
+      "loss": 0.1763,
+      "step": 1155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531953691032518e-05,
+      "loss": 0.1577,
+      "step": 1156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9530895338851e-05,
+      "loss": 0.1765,
+      "step": 1157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529835820180166e-05,
+      "loss": 0.1776,
+      "step": 1158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528775135149684e-05,
+      "loss": 0.1865,
+      "step": 1159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952771328388937e-05,
+      "loss": 0.1822,
+      "step": 1160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526650266529183e-05,
+      "loss": 0.1768,
+      "step": 1161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525586083199232e-05,
+      "loss": 0.1847,
+      "step": 1162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524520734029754e-05,
+      "loss": 0.164,
+      "step": 1163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952345421915115e-05,
+      "loss": 0.1788,
+      "step": 1164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522386538693937e-05,
+      "loss": 0.1971,
+      "step": 1165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95213176927888e-05,
+      "loss": 0.1669,
+      "step": 1166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952024768156655e-05,
+      "loss": 0.1861,
+      "step": 1167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519176505158146e-05,
+      "loss": 0.1758,
+      "step": 1168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951810416369469e-05,
+      "loss": 0.1832,
+      "step": 1169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517030657307425e-05,
+      "loss": 0.1697,
+      "step": 1170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515955986127742e-05,
+      "loss": 0.1518,
+      "step": 1171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951488015028717e-05,
+      "loss": 0.1624,
+      "step": 1172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513803149917375e-05,
+      "loss": 0.158,
+      "step": 1173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512724985150176e-05,
+      "loss": 0.1782,
+      "step": 1174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951164565611753e-05,
+      "loss": 0.183,
+      "step": 1175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.1513,
+      "step": 1176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509483505784437e-05,
+      "loss": 0.1857,
+      "step": 1177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9508400684748615e-05,
+      "loss": 0.156,
+      "step": 1178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507316699976595e-05,
+      "loss": 0.1768,
+      "step": 1179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950623155160105e-05,
+      "loss": 0.196,
+      "step": 1180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505145239754794e-05,
+      "loss": 0.1801,
+      "step": 1181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950405776457077e-05,
+      "loss": 0.1789,
+      "step": 1182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502969126182086e-05,
+      "loss": 0.1877,
+      "step": 1183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950187932472198e-05,
+      "loss": 0.1791,
+      "step": 1184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500788360323826e-05,
+      "loss": 0.167,
+      "step": 1185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949969623312115e-05,
+      "loss": 0.1482,
+      "step": 1186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498602943247625e-05,
+      "loss": 0.1886,
+      "step": 1187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9497508490837053e-05,
+      "loss": 0.1807,
+      "step": 1188
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949641287602338e-05,
+      "loss": 0.1695,
+      "step": 1189
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495316098940706e-05,
+      "loss": 0.1797,
+      "step": 1190
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949421815972326e-05,
+      "loss": 0.1849,
+      "step": 1191
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493119058505426e-05,
+      "loss": 0.1559,
+      "step": 1192
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949201879542172e-05,
+      "loss": 0.1847,
+      "step": 1193
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490917370606804e-05,
+      "loss": 0.1719,
+      "step": 1194
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948981478419548e-05,
+      "loss": 0.1674,
+      "step": 1195
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488711036322693e-05,
+      "loss": 0.1868,
+      "step": 1196
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487606127123534e-05,
+      "loss": 0.1703,
+      "step": 1197
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948650005673323e-05,
+      "loss": 0.1695,
+      "step": 1198
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485392825287158e-05,
+      "loss": 0.1706,
+      "step": 1199
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484284432920832e-05,
+      "loss": 0.1709,
+      "step": 1200
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483174879769902e-05,
+      "loss": 0.1668,
+      "step": 1201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482064165970173e-05,
+      "loss": 0.1658,
+      "step": 1202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480952291657585e-05,
+      "loss": 0.161,
+      "step": 1203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479839256968217e-05,
+      "loss": 0.1461,
+      "step": 1204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478725062038298e-05,
+      "loss": 0.1686,
+      "step": 1205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947760970700419e-05,
+      "loss": 0.1714,
+      "step": 1206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476493192002406e-05,
+      "loss": 0.1698,
+      "step": 1207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475375517169595e-05,
+      "loss": 0.1732,
+      "step": 1208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947425668264255e-05,
+      "loss": 0.1617,
+      "step": 1209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473136688558202e-05,
+      "loss": 0.1668,
+      "step": 1210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472015535053633e-05,
+      "loss": 0.1521,
+      "step": 1211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470893222266058e-05,
+      "loss": 0.1545,
+      "step": 1212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946976975033284e-05,
+      "loss": 0.1764,
+      "step": 1213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468645119391474e-05,
+      "loss": 0.1845,
+      "step": 1214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467519329579613e-05,
+      "loss": 0.1811,
+      "step": 1215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946639238103504e-05,
+      "loss": 0.1741,
+      "step": 1216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465264273895676e-05,
+      "loss": 0.1769,
+      "step": 1217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94641350082996e-05,
+      "loss": 0.1575,
+      "step": 1218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946300458438502e-05,
+      "loss": 0.1697,
+      "step": 1219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461873002290282e-05,
+      "loss": 0.1611,
+      "step": 1220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946074026215389e-05,
+      "loss": 0.1763,
+      "step": 1221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945960636411448e-05,
+      "loss": 0.169,
+      "step": 1222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458471308310822e-05,
+      "loss": 0.1783,
+      "step": 1223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457335094881846e-05,
+      "loss": 0.1717,
+      "step": 1224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456197723966602e-05,
+      "loss": 0.1781,
+      "step": 1225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455059195704303e-05,
+      "loss": 0.1708,
+      "step": 1226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453919510234293e-05,
+      "loss": 0.1636,
+      "step": 1227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452778667696053e-05,
+      "loss": 0.1659,
+      "step": 1228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451636668229215e-05,
+      "loss": 0.1668,
+      "step": 1229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450493511973548e-05,
+      "loss": 0.1792,
+      "step": 1230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449349199068963e-05,
+      "loss": 0.1671,
+      "step": 1231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9448203729655512e-05,
+      "loss": 0.1749,
+      "step": 1232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944705710387339e-05,
+      "loss": 0.1827,
+      "step": 1233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445909321862936e-05,
+      "loss": 0.1662,
+      "step": 1234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944476038376462e-05,
+      "loss": 0.1565,
+      "step": 1235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443610289719067e-05,
+      "loss": 0.1715,
+      "step": 1236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944245903986704e-05,
+      "loss": 0.1575,
+      "step": 1237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944130663434943e-05,
+      "loss": 0.1595,
+      "step": 1238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440153073307292e-05,
+      "loss": 0.1717,
+      "step": 1239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94389983568818e-05,
+      "loss": 0.1835,
+      "step": 1240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943784248521429e-05,
+      "loss": 0.1705,
+      "step": 1241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436685458446224e-05,
+      "loss": 0.17,
+      "step": 1242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435527276719214e-05,
+      "loss": 0.1656,
+      "step": 1243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434367940175005e-05,
+      "loss": 0.1671,
+      "step": 1244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433207448955497e-05,
+      "loss": 0.1613,
+      "step": 1245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432045803202712e-05,
+      "loss": 0.1738,
+      "step": 1246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430883003058833e-05,
+      "loss": 0.171,
+      "step": 1247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429719048666173e-05,
+      "loss": 0.1826,
+      "step": 1248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428553940167188e-05,
+      "loss": 0.1691,
+      "step": 1249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427387677704472e-05,
+      "loss": 0.1777,
+      "step": 1250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426220261420773e-05,
+      "loss": 0.1703,
+      "step": 1251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425051691458966e-05,
+      "loss": 0.1666,
+      "step": 1252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423881967962073e-05,
+      "loss": 0.1938,
+      "step": 1253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942271109107326e-05,
+      "loss": 0.1627,
+      "step": 1254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421539060935825e-05,
+      "loss": 0.1861,
+      "step": 1255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420365877693218e-05,
+      "loss": 0.1805,
+      "step": 1256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941919154148902e-05,
+      "loss": 0.1923,
+      "step": 1257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418016052466965e-05,
+      "loss": 0.1753,
+      "step": 1258
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416839410770922e-05,
+      "loss": 0.1901,
+      "step": 1259
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415661616544892e-05,
+      "loss": 0.1874,
+      "step": 1260
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414482669933035e-05,
+      "loss": 0.1858,
+      "step": 1261
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413302571079633e-05,
+      "loss": 0.1596,
+      "step": 1262
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941212132012913e-05,
+      "loss": 0.1846,
+      "step": 1263
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941093891722609e-05,
+      "loss": 0.1669,
+      "step": 1264
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409755362515233e-05,
+      "loss": 0.1742,
+      "step": 1265
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940857065614141e-05,
+      "loss": 0.1807,
+      "step": 1266
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9407384798249626e-05,
+      "loss": 0.1819,
+      "step": 1267
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9406197788985013e-05,
+      "loss": 0.1756,
+      "step": 1268
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405009628492846e-05,
+      "loss": 0.176,
+      "step": 1269
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940382031691855e-05,
+      "loss": 0.1755,
+      "step": 1270
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402629854407683e-05,
+      "loss": 0.1827,
+      "step": 1271
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401438241105944e-05,
+      "loss": 0.1637,
+      "step": 1272
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940024547715918e-05,
+      "loss": 0.1618,
+      "step": 1273
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939905156271337e-05,
+      "loss": 0.1496,
+      "step": 1274
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939785649791464e-05,
+      "loss": 0.1773,
+      "step": 1275
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396660282909252e-05,
+      "loss": 0.1486,
+      "step": 1276
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939546291784361e-05,
+      "loss": 0.1457,
+      "step": 1277
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394264402864265e-05,
+      "loss": 0.1749,
+      "step": 1278
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93930647381179e-05,
+      "loss": 0.1968,
+      "step": 1279
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391863923751346e-05,
+      "loss": 0.17,
+      "step": 1280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390661959911564e-05,
+      "loss": 0.1612,
+      "step": 1281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389458846745667e-05,
+      "loss": 0.1622,
+      "step": 1282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9388254584400905e-05,
+      "loss": 0.1602,
+      "step": 1283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387049173024667e-05,
+      "loss": 0.1698,
+      "step": 1284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9385842612764483e-05,
+      "loss": 0.153,
+      "step": 1285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384634903768024e-05,
+      "loss": 0.1729,
+      "step": 1286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938342604618311e-05,
+      "loss": 0.1773,
+      "step": 1287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938221604015768e-05,
+      "loss": 0.1673,
+      "step": 1288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381004885839833e-05,
+      "loss": 0.1742,
+      "step": 1289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379792583377807e-05,
+      "loss": 0.1773,
+      "step": 1290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937857913291997e-05,
+      "loss": 0.1485,
+      "step": 1291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377364534614836e-05,
+      "loss": 0.1668,
+      "step": 1292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937614878861107e-05,
+      "loss": 0.1684,
+      "step": 1293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374931895057453e-05,
+      "loss": 0.1708,
+      "step": 1294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373713854102934e-05,
+      "loss": 0.1727,
+      "step": 1295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372494665896583e-05,
+      "loss": 0.1656,
+      "step": 1296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371274330587616e-05,
+      "loss": 0.1533,
+      "step": 1297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370052848325392e-05,
+      "loss": 0.1793,
+      "step": 1298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936883021925941e-05,
+      "loss": 0.171,
+      "step": 1299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367606443539303e-05,
+      "loss": 0.1768,
+      "step": 1300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366381521314857e-05,
+      "loss": 0.1639,
+      "step": 1301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365155452735986e-05,
+      "loss": 0.1558,
+      "step": 1302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363928237952744e-05,
+      "loss": 0.1651,
+      "step": 1303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362699877115343e-05,
+      "loss": 0.213,
+      "step": 1304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936147037037411e-05,
+      "loss": 0.2017,
+      "step": 1305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936023971787953e-05,
+      "loss": 0.1819,
+      "step": 1306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359007919782223e-05,
+      "loss": 0.1691,
+      "step": 1307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935777497623295e-05,
+      "loss": 0.1695,
+      "step": 1308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356540887382615e-05,
+      "loss": 0.1689,
+      "step": 1309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355305653382243e-05,
+      "loss": 0.1585,
+      "step": 1310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354069274383032e-05,
+      "loss": 0.1638,
+      "step": 1311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9352831750536294e-05,
+      "loss": 0.1697,
+      "step": 1312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9351593081993494e-05,
+      "loss": 0.1662,
+      "step": 1313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350353268906234e-05,
+      "loss": 0.1733,
+      "step": 1314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934911231142625e-05,
+      "loss": 0.1496,
+      "step": 1315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934787020970542e-05,
+      "loss": 0.1421,
+      "step": 1316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346626963895776e-05,
+      "loss": 0.1619,
+      "step": 1317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345382574149473e-05,
+      "loss": 0.1768,
+      "step": 1318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344137040618816e-05,
+      "loss": 0.1598,
+      "step": 1319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934289036345624e-05,
+      "loss": 0.1785,
+      "step": 1320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341642542814328e-05,
+      "loss": 0.1675,
+      "step": 1321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340393578845805e-05,
+      "loss": 0.1787,
+      "step": 1322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933914347170353e-05,
+      "loss": 0.1733,
+      "step": 1323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933789222154051e-05,
+      "loss": 0.1666,
+      "step": 1324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336639828509872e-05,
+      "loss": 0.1687,
+      "step": 1325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335386292764907e-05,
+      "loss": 0.1735,
+      "step": 1326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933413161445903e-05,
+      "loss": 0.1879,
+      "step": 1327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933287579374581e-05,
+      "loss": 0.1797,
+      "step": 1328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331618830778943e-05,
+      "loss": 0.158,
+      "step": 1329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330360725712263e-05,
+      "loss": 0.1771,
+      "step": 1330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329101478699756e-05,
+      "loss": 0.1656,
+      "step": 1331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327841089895543e-05,
+      "loss": 0.1667,
+      "step": 1332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326579559453882e-05,
+      "loss": 0.1668,
+      "step": 1333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325316887529165e-05,
+      "loss": 0.1519,
+      "step": 1334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324053074275945e-05,
+      "loss": 0.177,
+      "step": 1335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932278811984889e-05,
+      "loss": 0.1616,
+      "step": 1336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321522024402817e-05,
+      "loss": 0.1634,
+      "step": 1337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932025478809269e-05,
+      "loss": 0.1724,
+      "step": 1338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318986411073605e-05,
+      "loss": 0.1696,
+      "step": 1339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93177168935008e-05,
+      "loss": 0.1649,
+      "step": 1340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316446235529643e-05,
+      "loss": 0.1592,
+      "step": 1341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315174437315663e-05,
+      "loss": 0.1551,
+      "step": 1342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313901499014507e-05,
+      "loss": 0.1787,
+      "step": 1343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312627420781973e-05,
+      "loss": 0.1838,
+      "step": 1344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311352202773997e-05,
+      "loss": 0.1805,
+      "step": 1345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931007584514665e-05,
+      "loss": 0.1554,
+      "step": 1346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9308798348056153e-05,
+      "loss": 0.1689,
+      "step": 1347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930751971165885e-05,
+      "loss": 0.1678,
+      "step": 1348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930623993611124e-05,
+      "loss": 0.1804,
+      "step": 1349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930495902156995e-05,
+      "loss": 0.1606,
+      "step": 1350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303676968191758e-05,
+      "loss": 0.1728,
+      "step": 1351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302393776133573e-05,
+      "loss": 0.175,
+      "step": 1352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930110944555244e-05,
+      "loss": 0.1719,
+      "step": 1353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299823976605557e-05,
+      "loss": 0.1614,
+      "step": 1354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298537369450245e-05,
+      "loss": 0.1757,
+      "step": 1355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929724962424398e-05,
+      "loss": 0.1716,
+      "step": 1356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929596074114436e-05,
+      "loss": 0.2021,
+      "step": 1357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929467072030914e-05,
+      "loss": 0.187,
+      "step": 1358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293379561896202e-05,
+      "loss": 0.1706,
+      "step": 1359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929208726606357e-05,
+      "loss": 0.1726,
+      "step": 1360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290793832969418e-05,
+      "loss": 0.1774,
+      "step": 1361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928949926277204e-05,
+      "loss": 0.1714,
+      "step": 1362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288203555629877e-05,
+      "loss": 0.1674,
+      "step": 1363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928690671170152e-05,
+      "loss": 0.1833,
+      "step": 1364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285608731145683e-05,
+      "loss": 0.1648,
+      "step": 1365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284309614121234e-05,
+      "loss": 0.1654,
+      "step": 1366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928300936078716e-05,
+      "loss": 0.1768,
+      "step": 1367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281707971302612e-05,
+      "loss": 0.1859,
+      "step": 1368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928040544582686e-05,
+      "loss": 0.1542,
+      "step": 1369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279101784519326e-05,
+      "loss": 0.1602,
+      "step": 1370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277796987539558e-05,
+      "loss": 0.1495,
+      "step": 1371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276491055047258e-05,
+      "loss": 0.1726,
+      "step": 1372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275183987202255e-05,
+      "loss": 0.1724,
+      "step": 1373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273875784164522e-05,
+      "loss": 0.1625,
+      "step": 1374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272566446094174e-05,
+      "loss": 0.1563,
+      "step": 1375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271255973151454e-05,
+      "loss": 0.1662,
+      "step": 1376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269944365496757e-05,
+      "loss": 0.1662,
+      "step": 1377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926863162329061e-05,
+      "loss": 0.1698,
+      "step": 1378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267317746693683e-05,
+      "loss": 0.1719,
+      "step": 1379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266002735866775e-05,
+      "loss": 0.1704,
+      "step": 1380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264686590970832e-05,
+      "loss": 0.1606,
+      "step": 1381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263369312166943e-05,
+      "loss": 0.1963,
+      "step": 1382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262050899616325e-05,
+      "loss": 0.1605,
+      "step": 1383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926073135348034e-05,
+      "loss": 0.1765,
+      "step": 1384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925941067392049e-05,
+      "loss": 0.1688,
+      "step": 1385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258088861098412e-05,
+      "loss": 0.1701,
+      "step": 1386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925676591517588e-05,
+      "loss": 0.1684,
+      "step": 1387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255441836314813e-05,
+      "loss": 0.1546,
+      "step": 1388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254116624677267e-05,
+      "loss": 0.1612,
+      "step": 1389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252790280425428e-05,
+      "loss": 0.1532,
+      "step": 1390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251462803721638e-05,
+      "loss": 0.1604,
+      "step": 1391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250134194728362e-05,
+      "loss": 0.1692,
+      "step": 1392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248804453608202e-05,
+      "loss": 0.1773,
+      "step": 1393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247473580523918e-05,
+      "loss": 0.1691,
+      "step": 1394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924614157563839e-05,
+      "loss": 0.186,
+      "step": 1395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9244808439114644e-05,
+      "loss": 0.1697,
+      "step": 1396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243474171115836e-05,
+      "loss": 0.1729,
+      "step": 1397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924213877180528e-05,
+      "loss": 0.1698,
+      "step": 1398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240802241346405e-05,
+      "loss": 0.1854,
+      "step": 1399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239464579902797e-05,
+      "loss": 0.1778,
+      "step": 1400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238125787638163e-05,
+      "loss": 0.1843,
+      "step": 1401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923678586471637e-05,
+      "loss": 0.1471,
+      "step": 1402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235444811301404e-05,
+      "loss": 0.1709,
+      "step": 1403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92341026275574e-05,
+      "loss": 0.1613,
+      "step": 1404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232759313648627e-05,
+      "loss": 0.1582,
+      "step": 1405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231414869739497e-05,
+      "loss": 0.1695,
+      "step": 1406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230069295994548e-05,
+      "loss": 0.1643,
+      "step": 1407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228722592578478e-05,
+      "loss": 0.165,
+      "step": 1408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92273747596561e-05,
+      "loss": 0.1629,
+      "step": 1409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226025797392375e-05,
+      "loss": 0.1751,
+      "step": 1410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224675705952414e-05,
+      "loss": 0.158,
+      "step": 1411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223324485501445e-05,
+      "loss": 0.167,
+      "step": 1412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922197213620485e-05,
+      "loss": 0.1463,
+      "step": 1413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220618658228137e-05,
+      "loss": 0.1873,
+      "step": 1414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219264051736968e-05,
+      "loss": 0.1619,
+      "step": 1415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217908316897126e-05,
+      "loss": 0.1674,
+      "step": 1416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216551453874542e-05,
+      "loss": 0.1679,
+      "step": 1417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9215193462835285e-05,
+      "loss": 0.1652,
+      "step": 1418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213834343945557e-05,
+      "loss": 0.1595,
+      "step": 1419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92124740973717e-05,
+      "loss": 0.1618,
+      "step": 1420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92111127232802e-05,
+      "loss": 0.1588,
+      "step": 1421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920975022183767e-05,
+      "loss": 0.182,
+      "step": 1422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208386593210874e-05,
+      "loss": 0.1527,
+      "step": 1423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92070218375667e-05,
+      "loss": 0.1516,
+      "step": 1424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205655955072188e-05,
+      "loss": 0.1858,
+      "step": 1425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92042889458945e-05,
+      "loss": 0.1769,
+      "step": 1426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920292081020095e-05,
+      "loss": 0.1585,
+      "step": 1427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201551548158986e-05,
+      "loss": 0.1759,
+      "step": 1428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920018115993619e-05,
+      "loss": 0.1677,
+      "step": 1429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198809645700285e-05,
+      "loss": 0.1594,
+      "step": 1430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919743700561913e-05,
+      "loss": 0.1816,
+      "step": 1431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196063239860726e-05,
+      "loss": 0.1597,
+      "step": 1432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194688348593202e-05,
+      "loss": 0.1796,
+      "step": 1433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919331233198484e-05,
+      "loss": 0.1865,
+      "step": 1434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191935190204044e-05,
+      "loss": 0.1633,
+      "step": 1435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190556923419366e-05,
+      "loss": 0.1631,
+      "step": 1436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918917753179949e-05,
+      "loss": 0.1851,
+      "step": 1437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187797015513244e-05,
+      "loss": 0.1666,
+      "step": 1438
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918641537472959e-05,
+      "loss": 0.19,
+      "step": 1439
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185032609617623e-05,
+      "loss": 0.1672,
+      "step": 1440
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183648720346585e-05,
+      "loss": 0.1684,
+      "step": 1441
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182263707085845e-05,
+      "loss": 0.1597,
+      "step": 1442
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918087757000492e-05,
+      "loss": 0.1555,
+      "step": 1443
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179490309273458e-05,
+      "loss": 0.1701,
+      "step": 1444
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9178101925061248e-05,
+      "loss": 0.1659,
+      "step": 1445
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176712417538215e-05,
+      "loss": 0.1672,
+      "step": 1446
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175321786874414e-05,
+      "loss": 0.1641,
+      "step": 1447
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173930033240055e-05,
+      "loss": 0.1582,
+      "step": 1448
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172537156805468e-05,
+      "loss": 0.1604,
+      "step": 1449
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917114315774113e-05,
+      "loss": 0.1845,
+      "step": 1450
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169748036217653e-05,
+      "loss": 0.1636,
+      "step": 1451
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168351792405787e-05,
+      "loss": 0.1718,
+      "step": 1452
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916695442647642e-05,
+      "loss": 0.1697,
+      "step": 1453
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165555938600572e-05,
+      "loss": 0.1749,
+      "step": 1454
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164156328949407e-05,
+      "loss": 0.174,
+      "step": 1455
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162755597694223e-05,
+      "loss": 0.1637,
+      "step": 1456
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161353745006454e-05,
+      "loss": 0.1549,
+      "step": 1457
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915995077105768e-05,
+      "loss": 0.146,
+      "step": 1458
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158546676019606e-05,
+      "loss": 0.155,
+      "step": 1459
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157141460064077e-05,
+      "loss": 0.1771,
+      "step": 1460
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155735123363084e-05,
+      "loss": 0.1652,
+      "step": 1461
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154327666088744e-05,
+      "loss": 0.1822,
+      "step": 1462
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915291908841332e-05,
+      "loss": 0.1664,
+      "step": 1463
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151509390509208e-05,
+      "loss": 0.1611,
+      "step": 1464
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915009857254894e-05,
+      "loss": 0.1432,
+      "step": 1465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148686634705186e-05,
+      "loss": 0.1622,
+      "step": 1466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147273577150752e-05,
+      "loss": 0.1567,
+      "step": 1467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145859400058584e-05,
+      "loss": 0.1659,
+      "step": 1468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9144444103601768e-05,
+      "loss": 0.1667,
+      "step": 1469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143027687953516e-05,
+      "loss": 0.1665,
+      "step": 1470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914161015328719e-05,
+      "loss": 0.1547,
+      "step": 1471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140191499776275e-05,
+      "loss": 0.1614,
+      "step": 1472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138771727594405e-05,
+      "loss": 0.1617,
+      "step": 1473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137350836915346e-05,
+      "loss": 0.1644,
+      "step": 1474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135928827913002e-05,
+      "loss": 0.1687,
+      "step": 1475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913450570076141e-05,
+      "loss": 0.1866,
+      "step": 1476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913308145563475e-05,
+      "loss": 0.1727,
+      "step": 1477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131656092707337e-05,
+      "loss": 0.1644,
+      "step": 1478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130229612153616e-05,
+      "loss": 0.1705,
+      "step": 1479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128802014148182e-05,
+      "loss": 0.1656,
+      "step": 1480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912737329886575e-05,
+      "loss": 0.1606,
+      "step": 1481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912594346648119e-05,
+      "loss": 0.1527,
+      "step": 1482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124512517169497e-05,
+      "loss": 0.1563,
+      "step": 1483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123080451105802e-05,
+      "loss": 0.1615,
+      "step": 1484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121647268465377e-05,
+      "loss": 0.167,
+      "step": 1485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912021296942363e-05,
+      "loss": 0.1601,
+      "step": 1486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118777554156108e-05,
+      "loss": 0.171,
+      "step": 1487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911734102283849e-05,
+      "loss": 0.1675,
+      "step": 1488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115903375646596e-05,
+      "loss": 0.1853,
+      "step": 1489
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9114464612756373e-05,
+      "loss": 0.1671,
+      "step": 1490
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113024734343918e-05,
+      "loss": 0.1615,
+      "step": 1491
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111583740585455e-05,
+      "loss": 0.1657,
+      "step": 1492
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110141631657352e-05,
+      "loss": 0.1674,
+      "step": 1493
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108698407736104e-05,
+      "loss": 0.1466,
+      "step": 1494
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107254068998352e-05,
+      "loss": 0.1574,
+      "step": 1495
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105808615620864e-05,
+      "loss": 0.1502,
+      "step": 1496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9104362047780555e-05,
+      "loss": 0.1578,
+      "step": 1497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9102914365654465e-05,
+      "loss": 0.168,
+      "step": 1498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101465569419784e-05,
+      "loss": 0.1628,
+      "step": 1499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100015659253825e-05,
+      "loss": 0.177,
+      "step": 1500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098564635334042e-05,
+      "loss": 0.1678,
+      "step": 1501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097112497838034e-05,
+      "loss": 0.1528,
+      "step": 1502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095659246943517e-05,
+      "loss": 0.1738,
+      "step": 1503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094204882828365e-05,
+      "loss": 0.18,
+      "step": 1504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092749405670572e-05,
+      "loss": 0.1822,
+      "step": 1505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091292815648275e-05,
+      "loss": 0.1879,
+      "step": 1506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089835112939752e-05,
+      "loss": 0.1797,
+      "step": 1507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088376297723407e-05,
+      "loss": 0.1754,
+      "step": 1508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086916370177785e-05,
+      "loss": 0.1686,
+      "step": 1509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085455330481565e-05,
+      "loss": 0.1596,
+      "step": 1510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083993178813568e-05,
+      "loss": 0.1736,
+      "step": 1511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082529915352746e-05,
+      "loss": 0.1572,
+      "step": 1512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081065540278187e-05,
+      "loss": 0.1719,
+      "step": 1513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9079600053769116e-05,
+      "loss": 0.1607,
+      "step": 1514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90781334560049e-05,
+      "loss": 0.1468,
+      "step": 1515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907666574716503e-05,
+      "loss": 0.1862,
+      "step": 1516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907519692742914e-05,
+      "loss": 0.1681,
+      "step": 1517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073726996977004e-05,
+      "loss": 0.1697,
+      "step": 1518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907225595598852e-05,
+      "loss": 0.1527,
+      "step": 1519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070783804643736e-05,
+      "loss": 0.1669,
+      "step": 1520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906931054312283e-05,
+      "loss": 0.1695,
+      "step": 1521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067836171606106e-05,
+      "loss": 0.1674,
+      "step": 1522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066360690274018e-05,
+      "loss": 0.1537,
+      "step": 1523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064884099307153e-05,
+      "loss": 0.194,
+      "step": 1524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063406398886227e-05,
+      "loss": 0.1644,
+      "step": 1525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90619275891921e-05,
+      "loss": 0.1712,
+      "step": 1526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906044767040576e-05,
+      "loss": 0.1586,
+      "step": 1527
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905896664270834e-05,
+      "loss": 0.182,
+      "step": 1528
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.90574845062811e-05,
+      "loss": 0.1802,
+      "step": 1529
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905600126130544e-05,
+      "loss": 0.1708,
+      "step": 1530
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054516907962896e-05,
+      "loss": 0.1689,
+      "step": 1531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053031446435132e-05,
+      "loss": 0.1534,
+      "step": 1532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051544876903963e-05,
+      "loss": 0.166,
+      "step": 1533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9050057199551325e-05,
+      "loss": 0.1579,
+      "step": 1534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048568414559295e-05,
+      "loss": 0.1678,
+      "step": 1535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904707852211009e-05,
+      "loss": 0.1545,
+      "step": 1536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904558752238606e-05,
+      "loss": 0.1701,
+      "step": 1537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904409541556968e-05,
+      "loss": 0.1559,
+      "step": 1538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904260220184358e-05,
+      "loss": 0.1648,
+      "step": 1539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041107881390507e-05,
+      "loss": 0.1456,
+      "step": 1540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903961245439335e-05,
+      "loss": 0.1536,
+      "step": 1541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038115921035144e-05,
+      "loss": 0.1766,
+      "step": 1542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036618281499043e-05,
+      "loss": 0.1547,
+      "step": 1543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903511953596835e-05,
+      "loss": 0.1649,
+      "step": 1544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033619684626488e-05,
+      "loss": 0.1698,
+      "step": 1545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032118727657026e-05,
+      "loss": 0.149,
+      "step": 1546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9030616665243675e-05,
+      "loss": 0.1562,
+      "step": 1547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029113497570266e-05,
+      "loss": 0.1654,
+      "step": 1548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027609224820774e-05,
+      "loss": 0.1773,
+      "step": 1549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026103847179303e-05,
+      "loss": 0.1654,
+      "step": 1550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024597364830103e-05,
+      "loss": 0.1811,
+      "step": 1551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023089777957553e-05,
+      "loss": 0.1496,
+      "step": 1552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902158108674616e-05,
+      "loss": 0.1589,
+      "step": 1553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902007129138058e-05,
+      "loss": 0.1667,
+      "step": 1554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018560392045596e-05,
+      "loss": 0.1576,
+      "step": 1555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017048388926124e-05,
+      "loss": 0.1547,
+      "step": 1556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015535282207224e-05,
+      "loss": 0.1647,
+      "step": 1557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901402107207408e-05,
+      "loss": 0.1601,
+      "step": 1558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901250575871202e-05,
+      "loss": 0.161,
+      "step": 1559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010989342306503e-05,
+      "loss": 0.172,
+      "step": 1560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009471823043123e-05,
+      "loss": 0.1698,
+      "step": 1561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900795320110761e-05,
+      "loss": 0.1669,
+      "step": 1562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900643347668583e-05,
+      "loss": 0.1592,
+      "step": 1563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004912649963784e-05,
+      "loss": 0.1615,
+      "step": 1564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9003390721127603e-05,
+      "loss": 0.1594,
+      "step": 1565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9001867690363553e-05,
+      "loss": 0.1715,
+      "step": 1566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000343557858046e-05,
+      "loss": 0.1589,
+      "step": 1567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998818323797617e-05,
+      "loss": 0.1591,
+      "step": 1568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899729198836894e-05,
+      "loss": 0.1809,
+      "step": 1569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995764551758824e-05,
+      "loss": 0.1702,
+      "step": 1570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994236014154212e-05,
+      "loss": 0.1729,
+      "step": 1571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992706375742184e-05,
+      "loss": 0.148,
+      "step": 1572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991175636709954e-05,
+      "loss": 0.1713,
+      "step": 1573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989643797244862e-05,
+      "loss": 0.1753,
+      "step": 1574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89881108575344e-05,
+      "loss": 0.1605,
+      "step": 1575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898657681776618e-05,
+      "loss": 0.1463,
+      "step": 1576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985041678127955e-05,
+      "loss": 0.1736,
+      "step": 1577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898350543880761e-05,
+      "loss": 0.1579,
+      "step": 1578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981968099993162e-05,
+      "loss": 0.1814,
+      "step": 1579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980429661872776e-05,
+      "loss": 0.175,
+      "step": 1580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978890124634733e-05,
+      "loss": 0.1884,
+      "step": 1581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977349488467464e-05,
+      "loss": 0.1761,
+      "step": 1582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975807753559526e-05,
+      "loss": 0.1721,
+      "step": 1583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897426492009961e-05,
+      "loss": 0.1671,
+      "step": 1584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972720988276542e-05,
+      "loss": 0.1606,
+      "step": 1585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971175958279294e-05,
+      "loss": 0.1762,
+      "step": 1586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8969629830296953e-05,
+      "loss": 0.1671,
+      "step": 1587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968082604518754e-05,
+      "loss": 0.1633,
+      "step": 1588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966534281134062e-05,
+      "loss": 0.1506,
+      "step": 1589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8964984860332376e-05,
+      "loss": 0.1768,
+      "step": 1590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896343434230333e-05,
+      "loss": 0.1676,
+      "step": 1591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961882727236694e-05,
+      "loss": 0.1502,
+      "step": 1592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960330015322373e-05,
+      "loss": 0.1589,
+      "step": 1593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958776206750398e-05,
+      "loss": 0.1602,
+      "step": 1594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957221301710943e-05,
+      "loss": 0.1716,
+      "step": 1595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955665300394315e-05,
+      "loss": 0.1693,
+      "step": 1596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954108202990952e-05,
+      "loss": 0.1752,
+      "step": 1597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952550009691424e-05,
+      "loss": 0.1552,
+      "step": 1598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950990720686445e-05,
+      "loss": 0.1881,
+      "step": 1599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949430336166852e-05,
+      "loss": 0.1587,
+      "step": 1600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947868856323626e-05,
+      "loss": 0.1605,
+      "step": 1601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946306281347873e-05,
+      "loss": 0.1665,
+      "step": 1602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944742611430835e-05,
+      "loss": 0.1489,
+      "step": 1603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943177846763898e-05,
+      "loss": 0.143,
+      "step": 1604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941611987538568e-05,
+      "loss": 0.1668,
+      "step": 1605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894004503394649e-05,
+      "loss": 0.1637,
+      "step": 1606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938476986179447e-05,
+      "loss": 0.1759,
+      "step": 1607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893690784442935e-05,
+      "loss": 0.1654,
+      "step": 1608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935337608888254e-05,
+      "loss": 0.1703,
+      "step": 1609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893376627974833e-05,
+      "loss": 0.1501,
+      "step": 1610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932193857201905e-05,
+      "loss": 0.1779,
+      "step": 1611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893062034144142e-05,
+      "loss": 0.1644,
+      "step": 1612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929045732659458e-05,
+      "loss": 0.1685,
+      "step": 1613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927470031048737e-05,
+      "loss": 0.1799,
+      "step": 1614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8925893236802113e-05,
+      "loss": 0.1612,
+      "step": 1615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924315350112563e-05,
+      "loss": 0.1597,
+      "step": 1616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922736371173208e-05,
+      "loss": 0.1603,
+      "step": 1617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921156300177303e-05,
+      "loss": 0.1608,
+      "step": 1618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891957513731823e-05,
+      "loss": 0.1417,
+      "step": 1619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8917992882789507e-05,
+      "loss": 0.1759,
+      "step": 1620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916409536784788e-05,
+      "loss": 0.1602,
+      "step": 1621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891482509949786e-05,
+      "loss": 0.1837,
+      "step": 1622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891323957112264e-05,
+      "loss": 0.1602,
+      "step": 1623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911652951853186e-05,
+      "loss": 0.1617,
+      "step": 1624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.1568,
+      "step": 1625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908476441408447e-05,
+      "loss": 0.1715,
+      "step": 1626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906886550621935e-05,
+      "loss": 0.1625,
+      "step": 1627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8905295569718734e-05,
+      "loss": 0.1627,
+      "step": 1628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890370349889357e-05,
+      "loss": 0.1575,
+      "step": 1629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902110338341287e-05,
+      "loss": 0.1737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8900516088256883e-05,
+      "loss": 0.1664,
+      "step": 1631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898920748835473e-05,
+      "loss": 0.1458,
+      "step": 1632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897324320272307e-05,
+      "loss": 0.1808,
+      "step": 1633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889572680276278e-05,
+      "loss": 0.1655,
+      "step": 1634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889412819650241e-05,
+      "loss": 0.1783,
+      "step": 1635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889252850168685e-05,
+      "loss": 0.1711,
+      "step": 1636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890927718511886e-05,
+      "loss": 0.1638,
+      "step": 1637
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8889325847173444e-05,
+      "loss": 0.1517,
+      "step": 1638
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888772288786757e-05,
+      "loss": 0.1726,
+      "step": 1639
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886118840790458e-05,
+      "loss": 0.1536,
+      "step": 1640
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888451370613842e-05,
+      "loss": 0.1756,
+      "step": 1641
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882907484107913e-05,
+      "loss": 0.1832,
+      "step": 1642
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881300174895524e-05,
+      "loss": 0.1609,
+      "step": 1643
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879691778697975e-05,
+      "loss": 0.1643,
+      "step": 1644
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8878082295712113e-05,
+      "loss": 0.167,
+      "step": 1645
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887647172613492e-05,
+      "loss": 0.163,
+      "step": 1646
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874860070163522e-05,
+      "loss": 0.1423,
+      "step": 1647
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873247327995168e-05,
+      "loss": 0.1645,
+      "step": 1648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887163349982724e-05,
+      "loss": 0.1648,
+      "step": 1649
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870018585857256e-05,
+      "loss": 0.1597,
+      "step": 1650
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8868402586282863e-05,
+      "loss": 0.1874,
+      "step": 1651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866785501301846e-05,
+      "loss": 0.186,
+      "step": 1652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865167331112125e-05,
+      "loss": 0.1835,
+      "step": 1653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886354807591174e-05,
+      "loss": 0.1554,
+      "step": 1654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861927735898877e-05,
+      "loss": 0.1534,
+      "step": 1655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886030631127185e-05,
+      "loss": 0.1491,
+      "step": 1656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858683802229105e-05,
+      "loss": 0.1608,
+      "step": 1657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857060208969225e-05,
+      "loss": 0.1529,
+      "step": 1658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855435531690914e-05,
+      "loss": 0.1558,
+      "step": 1659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885380977059302e-05,
+      "loss": 0.1645,
+      "step": 1660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852182925874526e-05,
+      "loss": 0.1583,
+      "step": 1661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850554997734535e-05,
+      "loss": 0.1547,
+      "step": 1662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884892598637229e-05,
+      "loss": 0.1587,
+      "step": 1663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847295891987172e-05,
+      "loss": 0.1565,
+      "step": 1664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845664714778683e-05,
+      "loss": 0.1481,
+      "step": 1665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844032454946467e-05,
+      "loss": 0.1762,
+      "step": 1666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88423991126903e-05,
+      "loss": 0.1654,
+      "step": 1667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840764688210074e-05,
+      "loss": 0.1618,
+      "step": 1668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883912918170584e-05,
+      "loss": 0.1746,
+      "step": 1669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837492593377762e-05,
+      "loss": 0.1658,
+      "step": 1670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8835854923426145e-05,
+      "loss": 0.1673,
+      "step": 1671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834216172051427e-05,
+      "loss": 0.1587,
+      "step": 1672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832576339454164e-05,
+      "loss": 0.1485,
+      "step": 1673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883093542583507e-05,
+      "loss": 0.1617,
+      "step": 1674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829293431394965e-05,
+      "loss": 0.1619,
+      "step": 1675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827650356334817e-05,
+      "loss": 0.1803,
+      "step": 1676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882600620085573e-05,
+      "loss": 0.1709,
+      "step": 1677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882436096515892e-05,
+      "loss": 0.1579,
+      "step": 1678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822714649445758e-05,
+      "loss": 0.1655,
+      "step": 1679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8821067253917735e-05,
+      "loss": 0.1521,
+      "step": 1680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819418778776468e-05,
+      "loss": 0.1614,
+      "step": 1681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881776922422373e-05,
+      "loss": 0.1612,
+      "step": 1682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8816118590461392e-05,
+      "loss": 0.168,
+      "step": 1683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8814466877691492e-05,
+      "loss": 0.1517,
+      "step": 1684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812814086116175e-05,
+      "loss": 0.1684,
+      "step": 1685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881116021593773e-05,
+      "loss": 0.1653,
+      "step": 1686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809505267358567e-05,
+      "loss": 0.1717,
+      "step": 1687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807849240581247e-05,
+      "loss": 0.1634,
+      "step": 1688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806192135808446e-05,
+      "loss": 0.146,
+      "step": 1689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804533953242977e-05,
+      "loss": 0.171,
+      "step": 1690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802874693087787e-05,
+      "loss": 0.1744,
+      "step": 1691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801214355545956e-05,
+      "loss": 0.1602,
+      "step": 1692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799552940820688e-05,
+      "loss": 0.1543,
+      "step": 1693
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8797890449115325e-05,
+      "loss": 0.1806,
+      "step": 1694
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796226880633344e-05,
+      "loss": 0.1519,
+      "step": 1695
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794562235578346e-05,
+      "loss": 0.1772,
+      "step": 1696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792896514154068e-05,
+      "loss": 0.1919,
+      "step": 1697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791229716564374e-05,
+      "loss": 0.1768,
+      "step": 1698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878956184301327e-05,
+      "loss": 0.1528,
+      "step": 1699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787892893704895e-05,
+      "loss": 0.1576,
+      "step": 1700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786222868843493e-05,
+      "loss": 0.1676,
+      "step": 1701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784551768633474e-05,
+      "loss": 0.1748,
+      "step": 1702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782879593279358e-05,
+      "loss": 0.172,
+      "step": 1703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781206342985804e-05,
+      "loss": 0.1618,
+      "step": 1704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779532017957606e-05,
+      "loss": 0.18,
+      "step": 1705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777856618399678e-05,
+      "loss": 0.1467,
+      "step": 1706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776180144517075e-05,
+      "loss": 0.1406,
+      "step": 1707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774502596514984e-05,
+      "loss": 0.169,
+      "step": 1708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877282397459872e-05,
+      "loss": 0.1674,
+      "step": 1709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771144278973728e-05,
+      "loss": 0.1459,
+      "step": 1710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769463509845586e-05,
+      "loss": 0.1429,
+      "step": 1711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876778166742001e-05,
+      "loss": 0.1701,
+      "step": 1712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766098751902834e-05,
+      "loss": 0.1621,
+      "step": 1713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876441476350003e-05,
+      "loss": 0.1659,
+      "step": 1714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876272970241771e-05,
+      "loss": 0.1514,
+      "step": 1715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87610435688621e-05,
+      "loss": 0.1832,
+      "step": 1716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759356363039573e-05,
+      "loss": 0.1541,
+      "step": 1717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757668085156628e-05,
+      "loss": 0.155,
+      "step": 1718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875597873541988e-05,
+      "loss": 0.169,
+      "step": 1719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754288314036108e-05,
+      "loss": 0.1562,
+      "step": 1720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875259682121219e-05,
+      "loss": 0.1534,
+      "step": 1721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750904257155155e-05,
+      "loss": 0.1916,
+      "step": 1722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749210622072154e-05,
+      "loss": 0.14,
+      "step": 1723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747515916170474e-05,
+      "loss": 0.1597,
+      "step": 1724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745820139657526e-05,
+      "loss": 0.1642,
+      "step": 1725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874412329274086e-05,
+      "loss": 0.1633,
+      "step": 1726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742425375628152e-05,
+      "loss": 0.1463,
+      "step": 1727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874072638852721e-05,
+      "loss": 0.1524,
+      "step": 1728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8739026331645975e-05,
+      "loss": 0.1658,
+      "step": 1729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737325205192518e-05,
+      "loss": 0.16,
+      "step": 1730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735623009375042e-05,
+      "loss": 0.1608,
+      "step": 1731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873391974440187e-05,
+      "loss": 0.1544,
+      "step": 1732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873221541048148e-05,
+      "loss": 0.1734,
+      "step": 1733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730510007822455e-05,
+      "loss": 0.1538,
+      "step": 1734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728803536633523e-05,
+      "loss": 0.1476,
+      "step": 1735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8727095997123542e-05,
+      "loss": 0.1588,
+      "step": 1736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725387389501494e-05,
+      "loss": 0.1505,
+      "step": 1737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723677713976494e-05,
+      "loss": 0.1509,
+      "step": 1738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721966970757798e-05,
+      "loss": 0.1691,
+      "step": 1739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8720255160054778e-05,
+      "loss": 0.1585,
+      "step": 1740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718542282076946e-05,
+      "loss": 0.155,
+      "step": 1741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716828337033942e-05,
+      "loss": 0.1809,
+      "step": 1742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715113325135533e-05,
+      "loss": 0.1706,
+      "step": 1743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871339724659162e-05,
+      "loss": 0.1798,
+      "step": 1744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871168010161224e-05,
+      "loss": 0.1565,
+      "step": 1745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870996189040755e-05,
+      "loss": 0.1602,
+      "step": 1746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708242613187845e-05,
+      "loss": 0.1463,
+      "step": 1747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706522270163545e-05,
+      "loss": 0.1521,
+      "step": 1748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.87048008615452e-05,
+      "loss": 0.1725,
+      "step": 1749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703078387543506e-05,
+      "loss": 0.1584,
+      "step": 1750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701354848369265e-05,
+      "loss": 0.1637,
+      "step": 1751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869963024423343e-05,
+      "loss": 0.1527,
+      "step": 1752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697904575347072e-05,
+      "loss": 0.1447,
+      "step": 1753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869617784192139e-05,
+      "loss": 0.1659,
+      "step": 1754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694450044167732e-05,
+      "loss": 0.139,
+      "step": 1755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692721182297556e-05,
+      "loss": 0.1675,
+      "step": 1756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869099125652246e-05,
+      "loss": 0.1731,
+      "step": 1757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868926026705417e-05,
+      "loss": 0.1596,
+      "step": 1758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868752821410454e-05,
+      "loss": 0.1591,
+      "step": 1759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685795097885556e-05,
+      "loss": 0.1521,
+      "step": 1760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868406091860934e-05,
+      "loss": 0.1607,
+      "step": 1761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8682325676488136e-05,
+      "loss": 0.1751,
+      "step": 1762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868058937173432e-05,
+      "loss": 0.1683,
+      "step": 1763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86788520045604e-05,
+      "loss": 0.1622,
+      "step": 1764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867711357517901e-05,
+      "loss": 0.1739,
+      "step": 1765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675374083802922e-05,
+      "loss": 0.1606,
+      "step": 1766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673633530645027e-05,
+      "loss": 0.1662,
+      "step": 1767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867189191591836e-05,
+      "loss": 0.1716,
+      "step": 1768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867014923983607e-05,
+      "loss": 0.1691,
+      "step": 1769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668405502611447e-05,
+      "loss": 0.1456,
+      "step": 1770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666660704457908e-05,
+      "loss": 0.1575,
+      "step": 1771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664914845588996e-05,
+      "loss": 0.1774,
+      "step": 1772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8663167926218392e-05,
+      "loss": 0.1586,
+      "step": 1773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86614199465599e-05,
+      "loss": 0.166,
+      "step": 1774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659670906827458e-05,
+      "loss": 0.143,
+      "step": 1775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8657920807235122e-05,
+      "loss": 0.1612,
+      "step": 1776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656169647997103e-05,
+      "loss": 0.1469,
+      "step": 1777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865441742932771e-05,
+      "loss": 0.1408,
+      "step": 1778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865266415144141e-05,
+      "loss": 0.1608,
+      "step": 1779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865090981455278e-05,
+      "loss": 0.1699,
+      "step": 1780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649154418876537e-05,
+      "loss": 0.1519,
+      "step": 1781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864739796462752e-05,
+      "loss": 0.1551,
+      "step": 1782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645640452020707e-05,
+      "loss": 0.1644,
+      "step": 1783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86438818812712e-05,
+      "loss": 0.1501,
+      "step": 1784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864212225259423e-05,
+      "loss": 0.1776,
+      "step": 1785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864036156620516e-05,
+      "loss": 0.1546,
+      "step": 1786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638599822319476e-05,
+      "loss": 0.1555,
+      "step": 1787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636837021152803e-05,
+      "loss": 0.1573,
+      "step": 1788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863507316292089e-05,
+      "loss": 0.1579,
+      "step": 1789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633308247839616e-05,
+      "loss": 0.1635,
+      "step": 1790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863154227612499e-05,
+      "loss": 0.1658,
+      "step": 1791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629775247993148e-05,
+      "loss": 0.1483,
+      "step": 1792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862800716366036e-05,
+      "loss": 0.1669,
+      "step": 1793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862623802334302e-05,
+      "loss": 0.1575,
+      "step": 1794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624467827257656e-05,
+      "loss": 0.1669,
+      "step": 1795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8622696575620917e-05,
+      "loss": 0.1555,
+      "step": 1796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620924268649593e-05,
+      "loss": 0.1697,
+      "step": 1797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86191509065606e-05,
+      "loss": 0.1737,
+      "step": 1798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861737648957097e-05,
+      "loss": 0.1498,
+      "step": 1799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615601017897882e-05,
+      "loss": 0.1568,
+      "step": 1800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613824491758635e-05,
+      "loss": 0.1713,
+      "step": 1801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861204691137066e-05,
+      "loss": 0.1603,
+      "step": 1802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861026827695151e-05,
+      "loss": 0.1622,
+      "step": 1803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608488588718883e-05,
+      "loss": 0.1625,
+      "step": 1804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8606707846890586e-05,
+      "loss": 0.1811,
+      "step": 1805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604926051684568e-05,
+      "loss": 0.1708,
+      "step": 1806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603143203318904e-05,
+      "loss": 0.1958,
+      "step": 1807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601359302011795e-05,
+      "loss": 0.1641,
+      "step": 1808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599574347981578e-05,
+      "loss": 0.1551,
+      "step": 1809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8597788341446707e-05,
+      "loss": 0.1604,
+      "step": 1810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859600128262578e-05,
+      "loss": 0.156,
+      "step": 1811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594213171737507e-05,
+      "loss": 0.1826,
+      "step": 1812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8592424009000746e-05,
+      "loss": 0.1549,
+      "step": 1813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590633794634464e-05,
+      "loss": 0.1546,
+      "step": 1814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8588842528857772e-05,
+      "loss": 0.1598,
+      "step": 1815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85870502118899e-05,
+      "loss": 0.1639,
+      "step": 1816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585256843950207e-05,
+      "loss": 0.1474,
+      "step": 1817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583462425258192e-05,
+      "loss": 0.1695,
+      "step": 1818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858166695603347e-05,
+      "loss": 0.1567,
+      "step": 1819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857987043649579e-05,
+      "loss": 0.1759,
+      "step": 1820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8578072866865025e-05,
+      "loss": 0.169,
+      "step": 1821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857627424736119e-05,
+      "loss": 0.1552,
+      "step": 1822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8574474578204404e-05,
+      "loss": 0.1527,
+      "step": 1823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572673859614943e-05,
+      "loss": 0.161,
+      "step": 1824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570872091813186e-05,
+      "loss": 0.1698,
+      "step": 1825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856906927501966e-05,
+      "loss": 0.1682,
+      "step": 1826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567265409455014e-05,
+      "loss": 0.156,
+      "step": 1827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565460495340017e-05,
+      "loss": 0.1715,
+      "step": 1828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563654532895573e-05,
+      "loss": 0.1697,
+      "step": 1829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856184752234272e-05,
+      "loss": 0.1544,
+      "step": 1830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856003946390261e-05,
+      "loss": 0.1573,
+      "step": 1831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855823035779654e-05,
+      "loss": 0.1582,
+      "step": 1832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556420204245924e-05,
+      "loss": 0.165,
+      "step": 1833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855460900347231e-05,
+      "loss": 0.1897,
+      "step": 1834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552796755697366e-05,
+      "loss": 0.1644,
+      "step": 1835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85509834611429e-05,
+      "loss": 0.1616,
+      "step": 1836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549169120030837e-05,
+      "loss": 0.1559,
+      "step": 1837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547353732583234e-05,
+      "loss": 0.1659,
+      "step": 1838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854553729902228e-05,
+      "loss": 0.1645,
+      "step": 1839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543719819570288e-05,
+      "loss": 0.1531,
+      "step": 1840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541901294449702e-05,
+      "loss": 0.1703,
+      "step": 1841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8540081723883085e-05,
+      "loss": 0.1762,
+      "step": 1842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538261108093142e-05,
+      "loss": 0.1585,
+      "step": 1843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536439447302698e-05,
+      "loss": 0.1555,
+      "step": 1844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534616741734705e-05,
+      "loss": 0.147,
+      "step": 1845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853279299161224e-05,
+      "loss": 0.1697,
+      "step": 1846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530968197158518e-05,
+      "loss": 0.1479,
+      "step": 1847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529142358596877e-05,
+      "loss": 0.1674,
+      "step": 1848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527315476150775e-05,
+      "loss": 0.1704,
+      "step": 1849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525487550043814e-05,
+      "loss": 0.1476,
+      "step": 1850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852365858049971e-05,
+      "loss": 0.1582,
+      "step": 1851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521828567742308e-05,
+      "loss": 0.1607,
+      "step": 1852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851999751199559e-05,
+      "loss": 0.1528,
+      "step": 1853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8518165413483656e-05,
+      "loss": 0.1701,
+      "step": 1854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516332272430738e-05,
+      "loss": 0.1353,
+      "step": 1855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514498089061193e-05,
+      "loss": 0.1502,
+      "step": 1856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512662863599508e-05,
+      "loss": 0.164,
+      "step": 1857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85108265962703e-05,
+      "loss": 0.1464,
+      "step": 1858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508989287298304e-05,
+      "loss": 0.1534,
+      "step": 1859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85071509369084e-05,
+      "loss": 0.149,
+      "step": 1860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850531154532557e-05,
+      "loss": 0.1526,
+      "step": 1861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503471112774946e-05,
+      "loss": 0.1569,
+      "step": 1862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501629639481776e-05,
+      "loss": 0.1552,
+      "step": 1863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8499787125671446e-05,
+      "loss": 0.1774,
+      "step": 1864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497943571569453e-05,
+      "loss": 0.167,
+      "step": 1865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496098977401433e-05,
+      "loss": 0.1515,
+      "step": 1866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494253343393145e-05,
+      "loss": 0.1713,
+      "step": 1867
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849240666977048e-05,
+      "loss": 0.143,
+      "step": 1868
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490558956759453e-05,
+      "loss": 0.1481,
+      "step": 1869
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84887102045862e-05,
+      "loss": 0.1433,
+      "step": 1870
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486860413477e-05,
+      "loss": 0.1528,
+      "step": 1871
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485009583658245e-05,
+      "loss": 0.1643,
+      "step": 1872
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483157715356456e-05,
+      "loss": 0.1847,
+      "step": 1873
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848130480879829e-05,
+      "loss": 0.1597,
+      "step": 1874
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479450864210516e-05,
+      "loss": 0.1516,
+      "step": 1875
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477595881820042e-05,
+      "loss": 0.1717,
+      "step": 1876
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475739861853906e-05,
+      "loss": 0.1717,
+      "step": 1877
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847388280453926e-05,
+      "loss": 0.1582,
+      "step": 1878
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472024710103394e-05,
+      "loss": 0.1686,
+      "step": 1879
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847016557877372e-05,
+      "loss": 0.1537,
+      "step": 1880
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846830541077777e-05,
+      "loss": 0.1629,
+      "step": 1881
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466444206343223e-05,
+      "loss": 0.1476,
+      "step": 1882
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464581965697866e-05,
+      "loss": 0.1532,
+      "step": 1883
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846271868906962e-05,
+      "loss": 0.1539,
+      "step": 1884
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846085437668653e-05,
+      "loss": 0.1632,
+      "step": 1885
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8458989028776776e-05,
+      "loss": 0.1669,
+      "step": 1886
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845712264556865e-05,
+      "loss": 0.1682,
+      "step": 1887
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455255227290588e-05,
+      "loss": 0.1566,
+      "step": 1888
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453386774171136e-05,
+      "loss": 0.1779,
+      "step": 1889
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451517286438978e-05,
+      "loss": 0.1541,
+      "step": 1890
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844964676432292e-05,
+      "loss": 0.1511,
+      "step": 1891
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84477752080519e-05,
+      "loss": 0.1655,
+      "step": 1892
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844590261785497e-05,
+      "loss": 0.1414,
+      "step": 1893
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844402899396133e-05,
+      "loss": 0.178,
+      "step": 1894
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442154336600284e-05,
+      "loss": 0.1678,
+      "step": 1895
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8440278646001268e-05,
+      "loss": 0.1488,
+      "step": 1896
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438401922393857e-05,
+      "loss": 0.1583,
+      "step": 1897
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8436524166007742e-05,
+      "loss": 0.1711,
+      "step": 1898
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843464537707274e-05,
+      "loss": 0.1562,
+      "step": 1899
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432765555818797e-05,
+      "loss": 0.1651,
+      "step": 1900
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430884702475987e-05,
+      "loss": 0.1688,
+      "step": 1901
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429002817274506e-05,
+      "loss": 0.1627,
+      "step": 1902
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842711990044468e-05,
+      "loss": 0.158,
+      "step": 1903
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425235952216957e-05,
+      "loss": 0.157,
+      "step": 1904
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8423350972821918e-05,
+      "loss": 0.1478,
+      "step": 1905
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842146496249026e-05,
+      "loss": 0.1572,
+      "step": 1906
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841957792145282e-05,
+      "loss": 0.1466,
+      "step": 1907
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417689849940552e-05,
+      "loss": 0.146,
+      "step": 1908
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415800748184535e-05,
+      "loss": 0.157,
+      "step": 1909
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413910616415977e-05,
+      "loss": 0.1608,
+      "step": 1910
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841201945486621e-05,
+      "loss": 0.1405,
+      "step": 1911
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410127263766702e-05,
+      "loss": 0.1756,
+      "step": 1912
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840823404334903e-05,
+      "loss": 0.1585,
+      "step": 1913
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406339793844906e-05,
+      "loss": 0.1617,
+      "step": 1914
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8404444515486176e-05,
+      "loss": 0.1681,
+      "step": 1915
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402548208504797e-05,
+      "loss": 0.1386,
+      "step": 1916
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840065087313286e-05,
+      "loss": 0.1533,
+      "step": 1917
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839875250960258e-05,
+      "loss": 0.1456,
+      "step": 1918
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83968531181463e-05,
+      "loss": 0.167,
+      "step": 1919
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394952698996485e-05,
+      "loss": 0.1464,
+      "step": 1920
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839305125238573e-05,
+      "loss": 0.1813,
+      "step": 1921
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839114877854675e-05,
+      "loss": 0.1577,
+      "step": 1922
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8389245277712396e-05,
+      "loss": 0.1551,
+      "step": 1923
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8387340750115633e-05,
+      "loss": 0.1691,
+      "step": 1924
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838543519598956e-05,
+      "loss": 0.1653,
+      "step": 1925
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383528615567396e-05,
+      "loss": 0.1571,
+      "step": 1926
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8381621009082488e-05,
+      "loss": 0.1641,
+      "step": 1927
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379712376768304e-05,
+      "loss": 0.1516,
+      "step": 1928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377802718858452e-05,
+      "loss": 0.1628,
+      "step": 1929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837589203558665e-05,
+      "loss": 0.1734,
+      "step": 1930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373980327186743e-05,
+      "loss": 0.1584,
+      "step": 1931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837206759389272e-05,
+      "loss": 0.1638,
+      "step": 1932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8370153835938663e-05,
+      "loss": 0.1539,
+      "step": 1933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8368239053558806e-05,
+      "loss": 0.1503,
+      "step": 1934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366323246987506e-05,
+      "loss": 0.1411,
+      "step": 1935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836440641645923e-05,
+      "loss": 0.1721,
+      "step": 1936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362488562208577e-05,
+      "loss": 0.1516,
+      "step": 1937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360569684470286e-05,
+      "loss": 0.1496,
+      "step": 1938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83586497834792e-05,
+      "loss": 0.1443,
+      "step": 1939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356728859470298e-05,
+      "loss": 0.1539,
+      "step": 1940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354806912678684e-05,
+      "loss": 0.1589,
+      "step": 1941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8352883943339582e-05,
+      "loss": 0.1474,
+      "step": 1942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835095995168835e-05,
+      "loss": 0.1598,
+      "step": 1943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834903493796046e-05,
+      "loss": 0.153,
+      "step": 1944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347108902391525e-05,
+      "loss": 0.1495,
+      "step": 1945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834518184521726e-05,
+      "loss": 0.1635,
+      "step": 1946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834325376667353e-05,
+      "loss": 0.1525,
+      "step": 1947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8341324666996307e-05,
+      "loss": 0.1627,
+      "step": 1948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339394546421692e-05,
+      "loss": 0.15,
+      "step": 1949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833746340518592e-05,
+      "loss": 0.156,
+      "step": 1950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8335531243525336e-05,
+      "loss": 0.152,
+      "step": 1951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8333598061676424e-05,
+      "loss": 0.161,
+      "step": 1952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331663859875784e-05,
+      "loss": 0.1704,
+      "step": 1953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329728638360148e-05,
+      "loss": 0.1769,
+      "step": 1954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832779239736636e-05,
+      "loss": 0.1448,
+      "step": 1955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325855137131402e-05,
+      "loss": 0.1563,
+      "step": 1956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832391685789238e-05,
+      "loss": 0.16,
+      "step": 1957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321977559886514e-05,
+      "loss": 0.1517,
+      "step": 1958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320037243351155e-05,
+      "loss": 0.166,
+      "step": 1959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318095908523785e-05,
+      "loss": 0.1427,
+      "step": 1960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316153555641996e-05,
+      "loss": 0.149,
+      "step": 1961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314210184943527e-05,
+      "loss": 0.1735,
+      "step": 1962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312265796666215e-05,
+      "loss": 0.1601,
+      "step": 1963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831032039104804e-05,
+      "loss": 0.1634,
+      "step": 1964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308373968327095e-05,
+      "loss": 0.1578,
+      "step": 1965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830642652874161e-05,
+      "loss": 0.1531,
+      "step": 1966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830447807252993e-05,
+      "loss": 0.1551,
+      "step": 1967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8302528599930526e-05,
+      "loss": 0.1351,
+      "step": 1968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300578111182e-05,
+      "loss": 0.1543,
+      "step": 1969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298626606523067e-05,
+      "loss": 0.1714,
+      "step": 1970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8296674086192575e-05,
+      "loss": 0.147,
+      "step": 1971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829472055042949e-05,
+      "loss": 0.1543,
+      "step": 1972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829276599947291e-05,
+      "loss": 0.1542,
+      "step": 1973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290810433562052e-05,
+      "loss": 0.1565,
+      "step": 1974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828885385293626e-05,
+      "loss": 0.1757,
+      "step": 1975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8286896257834998e-05,
+      "loss": 0.1734,
+      "step": 1976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8284937648497857e-05,
+      "loss": 0.1394,
+      "step": 1977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8282978025164553e-05,
+      "loss": 0.1527,
+      "step": 1978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281017388074924e-05,
+      "loss": 0.1557,
+      "step": 1979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827905573746893e-05,
+      "loss": 0.1678,
+      "step": 1980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277093073586667e-05,
+      "loss": 0.1488,
+      "step": 1981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8275129396668334e-05,
+      "loss": 0.1568,
+      "step": 1982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827316470695428e-05,
+      "loss": 0.1584,
+      "step": 1983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827119900468495e-05,
+      "loss": 0.1605,
+      "step": 1984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8269232290100933e-05,
+      "loss": 0.1709,
+      "step": 1985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826726456344294e-05,
+      "loss": 0.1584,
+      "step": 1986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8265295824951794e-05,
+      "loss": 0.168,
+      "step": 1987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8263326074868456e-05,
+      "loss": 0.1584,
+      "step": 1988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261355313434e-05,
+      "loss": 0.1759,
+      "step": 1989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825938354088963e-05,
+      "loss": 0.1586,
+      "step": 1990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257410757476672e-05,
+      "loss": 0.1613,
+      "step": 1991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255436963436575e-05,
+      "loss": 0.1526,
+      "step": 1992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253462159010913e-05,
+      "loss": 0.163,
+      "step": 1993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251486344441382e-05,
+      "loss": 0.1534,
+      "step": 1994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8249509519969803e-05,
+      "loss": 0.1595,
+      "step": 1995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824753168583812e-05,
+      "loss": 0.1583,
+      "step": 1996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245552842288403e-05,
+      "loss": 0.1647,
+      "step": 1997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824357298956284e-05,
+      "loss": 0.149,
+      "step": 1998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8241592127903747e-05,
+      "loss": 0.1446,
+      "step": 1999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823961025755356e-05,
+      "loss": 0.1491,
+      "step": 2000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237627378754845e-05,
+      "loss": 0.1538,
+      "step": 2001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235643491750286e-05,
+      "loss": 0.1624,
+      "step": 2002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823365859678269e-05,
+      "loss": 0.1705,
+      "step": 2003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823167269409499e-05,
+      "loss": 0.1493,
+      "step": 2004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822968578393024e-05,
+      "loss": 0.1612,
+      "step": 2005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227697866531622e-05,
+      "loss": 0.1603,
+      "step": 2006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8225708942142434e-05,
+      "loss": 0.1551,
+      "step": 2007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223719011006104e-05,
+      "loss": 0.159,
+      "step": 2008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221728073366174e-05,
+      "loss": 0.1676,
+      "step": 2009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8219736129466326e-05,
+      "loss": 0.1662,
+      "step": 2010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217743179550345e-05,
+      "loss": 0.1557,
+      "step": 2011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215749223862154e-05,
+      "loss": 0.1538,
+      "step": 2012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8213754262645793e-05,
+      "loss": 0.1753,
+      "step": 2013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211758296145425e-05,
+      "loss": 0.1557,
+      "step": 2014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209761324605336e-05,
+      "loss": 0.1437,
+      "step": 2015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8207763348269943e-05,
+      "loss": 0.1535,
+      "step": 2016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820576436738377e-05,
+      "loss": 0.1572,
+      "step": 2017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203764382191476e-05,
+      "loss": 0.146,
+      "step": 2018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820176339293784e-05,
+      "loss": 0.1591,
+      "step": 2019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199761399867766e-05,
+      "loss": 0.1397,
+      "step": 2020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197758403226273e-05,
+      "loss": 0.16,
+      "step": 2021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195754403258516e-05,
+      "loss": 0.1699,
+      "step": 2022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193749400209758e-05,
+      "loss": 0.1423,
+      "step": 2023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191743394325396e-05,
+      "loss": 0.1566,
+      "step": 2024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8189736385850946e-05,
+      "loss": 0.1692,
+      "step": 2025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187728375032042e-05,
+      "loss": 0.1553,
+      "step": 2026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818571936211445e-05,
+      "loss": 0.1586,
+      "step": 2027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8183709347344052e-05,
+      "loss": 0.157,
+      "step": 2028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181698330966856e-05,
+      "loss": 0.1745,
+      "step": 2029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179686313228986e-05,
+      "loss": 0.1538,
+      "step": 2030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81776732943767e-05,
+      "loss": 0.1549,
+      "step": 2031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175659274656366e-05,
+      "loss": 0.1666,
+      "step": 2032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173644254314485e-05,
+      "loss": 0.1502,
+      "step": 2033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171628233597674e-05,
+      "loss": 0.1744,
+      "step": 2034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169611212752674e-05,
+      "loss": 0.1624,
+      "step": 2035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816759319202635e-05,
+      "loss": 0.1486,
+      "step": 2036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816557417166569e-05,
+      "loss": 0.1588,
+      "step": 2037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163554151917796e-05,
+      "loss": 0.1624,
+      "step": 2038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161533133029908e-05,
+      "loss": 0.1494,
+      "step": 2039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8159511115249367e-05,
+      "loss": 0.159,
+      "step": 2040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157488098823663e-05,
+      "loss": 0.169,
+      "step": 2041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8155464084000384e-05,
+      "loss": 0.1575,
+      "step": 2042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815343907102725e-05,
+      "loss": 0.1557,
+      "step": 2043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815141306015211e-05,
+      "loss": 0.1684,
+      "step": 2044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8149386051622915e-05,
+      "loss": 0.1492,
+      "step": 2045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8147358045687765e-05,
+      "loss": 0.1713,
+      "step": 2046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145329042594864e-05,
+      "loss": 0.1588,
+      "step": 2047
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814329904259254e-05,
+      "loss": 0.1529,
+      "step": 2048
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141268045929245e-05,
+      "loss": 0.1539,
+      "step": 2049
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139236052853557e-05,
+      "loss": 0.1495,
+      "step": 2050
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137203063614168e-05,
+      "loss": 0.1471,
+      "step": 2051
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135169078459897e-05,
+      "loss": 0.1693,
+      "step": 2052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133134097639687e-05,
+      "loss": 0.1621,
+      "step": 2053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81310981214026e-05,
+      "loss": 0.1641,
+      "step": 2054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129061149997814e-05,
+      "loss": 0.1637,
+      "step": 2055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127023183674643e-05,
+      "loss": 0.1628,
+      "step": 2056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812498422268251e-05,
+      "loss": 0.1516,
+      "step": 2057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812294426727096e-05,
+      "loss": 0.1688,
+      "step": 2058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812090331768967e-05,
+      "loss": 0.1619,
+      "step": 2059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811886137418843e-05,
+      "loss": 0.1543,
+      "step": 2060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116818437017154e-05,
+      "loss": 0.1558,
+      "step": 2061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811477450642588e-05,
+      "loss": 0.1791,
+      "step": 2062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112729582664762e-05,
+      "loss": 0.1744,
+      "step": 2063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110683665984082e-05,
+      "loss": 0.1709,
+      "step": 2064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810863675663424e-05,
+      "loss": 0.1769,
+      "step": 2065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8106588854865753e-05,
+      "loss": 0.1717,
+      "step": 2066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810453996092927e-05,
+      "loss": 0.168,
+      "step": 2067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102490075075553e-05,
+      "loss": 0.1512,
+      "step": 2068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810043919755549e-05,
+      "loss": 0.1833,
+      "step": 2069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098387328620087e-05,
+      "loss": 0.1612,
+      "step": 2070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809633446852047e-05,
+      "loss": 0.1516,
+      "step": 2071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80942806175079e-05,
+      "loss": 0.1493,
+      "step": 2072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8092225775833734e-05,
+      "loss": 0.1601,
+      "step": 2073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.1507,
+      "step": 2074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088113121506733e-05,
+      "loss": 0.1542,
+      "step": 2075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086055309357245e-05,
+      "loss": 0.1582,
+      "step": 2076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083996507552866e-05,
+      "loss": 0.1446,
+      "step": 2077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081936716345574e-05,
+      "loss": 0.169,
+      "step": 2078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079875935987468e-05,
+      "loss": 0.1422,
+      "step": 2079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8077814166730765e-05,
+      "loss": 0.1591,
+      "step": 2080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075751408827808e-05,
+      "loss": 0.1631,
+      "step": 2081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073687662531056e-05,
+      "loss": 0.1662,
+      "step": 2082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8071622928093094e-05,
+      "loss": 0.1542,
+      "step": 2083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069557205766627e-05,
+      "loss": 0.1527,
+      "step": 2084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806749049580448e-05,
+      "loss": 0.1576,
+      "step": 2085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065422798459587e-05,
+      "loss": 0.1535,
+      "step": 2086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806335411398503e-05,
+      "loss": 0.1576,
+      "step": 2087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061284442633986e-05,
+      "loss": 0.1922,
+      "step": 2088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8059213784659766e-05,
+      "loss": 0.1588,
+      "step": 2089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8057142140315797e-05,
+      "loss": 0.1397,
+      "step": 2090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805506950985563e-05,
+      "loss": 0.1593,
+      "step": 2091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052995893532936e-05,
+      "loss": 0.16,
+      "step": 2092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80509212916015e-05,
+      "loss": 0.1619,
+      "step": 2093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804884570431524e-05,
+      "loss": 0.15,
+      "step": 2094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046769131928187e-05,
+      "loss": 0.1284,
+      "step": 2095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044691574694483e-05,
+      "loss": 0.1571,
+      "step": 2096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042613032868413e-05,
+      "loss": 0.1509,
+      "step": 2097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804053350670437e-05,
+      "loss": 0.1446,
+      "step": 2098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803845299645686e-05,
+      "loss": 0.145,
+      "step": 2099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036371502380523e-05,
+      "loss": 0.1454,
+      "step": 2100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803428902473011e-05,
+      "loss": 0.1611,
+      "step": 2101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80322055637605e-05,
+      "loss": 0.1563,
+      "step": 2102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803012111972669e-05,
+      "loss": 0.1649,
+      "step": 2103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028035692883786e-05,
+      "loss": 0.1654,
+      "step": 2104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025949283487034e-05,
+      "loss": 0.1635,
+      "step": 2105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802386189179179e-05,
+      "loss": 0.158,
+      "step": 2106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021773518053523e-05,
+      "loss": 0.1596,
+      "step": 2107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019684162527834e-05,
+      "loss": 0.1532,
+      "step": 2108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017593825470442e-05,
+      "loss": 0.1417,
+      "step": 2109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801550250713718e-05,
+      "loss": 0.1623,
+      "step": 2110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013410207784012e-05,
+      "loss": 0.1534,
+      "step": 2111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011316927667006e-05,
+      "loss": 0.1563,
+      "step": 2112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8009222667042367e-05,
+      "loss": 0.1522,
+      "step": 2113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8007127426166406e-05,
+      "loss": 0.1669,
+      "step": 2114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800503120529556e-05,
+      "loss": 0.16,
+      "step": 2115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8002934004686396e-05,
+      "loss": 0.1568,
+      "step": 2116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800083582459558e-05,
+      "loss": 0.1744,
+      "step": 2117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998736665279914e-05,
+      "loss": 0.1763,
+      "step": 2118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996636526996315e-05,
+      "loss": 0.1709,
+      "step": 2119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799453541000182e-05,
+      "loss": 0.15,
+      "step": 2120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799243331455358e-05,
+      "loss": 0.1525,
+      "step": 2121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990330240908874e-05,
+      "loss": 0.1693,
+      "step": 2122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988226189325104e-05,
+      "loss": 0.1633,
+      "step": 2123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798612116005978e-05,
+      "loss": 0.1582,
+      "step": 2124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798401515337053e-05,
+      "loss": 0.1661,
+      "step": 2125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981908169515125e-05,
+      "loss": 0.143,
+      "step": 2126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797980020875143e-05,
+      "loss": 0.1588,
+      "step": 2127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977691271337438e-05,
+      "loss": 0.1506,
+      "step": 2128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797558135753126e-05,
+      "loss": 0.1588,
+      "step": 2129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797347046759114e-05,
+      "loss": 0.1392,
+      "step": 2130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7971358601775422e-05,
+      "loss": 0.1535,
+      "step": 2131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7969245760342578e-05,
+      "loss": 0.1483,
+      "step": 2132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7967131943551207e-05,
+      "loss": 0.1592,
+      "step": 2133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796501715166001e-05,
+      "loss": 0.1729,
+      "step": 2134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7962901384927817e-05,
+      "loss": 0.1694,
+      "step": 2135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960784643613586e-05,
+      "loss": 0.1582,
+      "step": 2136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795866692797638e-05,
+      "loss": 0.1695,
+      "step": 2137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956548238275387e-05,
+      "loss": 0.1456,
+      "step": 2138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795442857476992e-05,
+      "loss": 0.1566,
+      "step": 2139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952307937719395e-05,
+      "loss": 0.1439,
+      "step": 2140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950186327383362e-05,
+      "loss": 0.1595,
+      "step": 2141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794806374402149e-05,
+      "loss": 0.1536,
+      "step": 2142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945940187893554e-05,
+      "loss": 0.1599,
+      "step": 2143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794381565925947e-05,
+      "loss": 0.1517,
+      "step": 2144
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941690158379244e-05,
+      "loss": 0.1639,
+      "step": 2145
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939563685513026e-05,
+      "loss": 0.1469,
+      "step": 2146
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937436240921077e-05,
+      "loss": 0.1523,
+      "step": 2147
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935307824863767e-05,
+      "loss": 0.1658,
+      "step": 2148
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79331784376016e-05,
+      "loss": 0.1683,
+      "step": 2149
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931048079395196e-05,
+      "loss": 0.1319,
+      "step": 2150
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792891675050528e-05,
+      "loss": 0.1497,
+      "step": 2151
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7926784451192715e-05,
+      "loss": 0.1503,
+      "step": 2152
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792465118171847e-05,
+      "loss": 0.153,
+      "step": 2153
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922516942343636e-05,
+      "loss": 0.1643,
+      "step": 2154
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7920381733329426e-05,
+      "loss": 0.1498,
+      "step": 2155
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918245554937166e-05,
+      "loss": 0.1577,
+      "step": 2156
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7916108407428304e-05,
+      "loss": 0.1456,
+      "step": 2157
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791397029106441e-05,
+      "loss": 0.1485,
+      "step": 2158
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791183120610716e-05,
+      "loss": 0.1529,
+      "step": 2159
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790969115281837e-05,
+      "loss": 0.1566,
+      "step": 2160
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907550131459945e-05,
+      "loss": 0.1552,
+      "step": 2161
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905408142293944e-05,
+      "loss": 0.1374,
+      "step": 2162
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790326518558251e-05,
+      "loss": 0.1782,
+      "step": 2163
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790112126158793e-05,
+      "loss": 0.1684,
+      "step": 2164
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7898976370572595e-05,
+      "loss": 0.161,
+      "step": 2165
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7896830512799025e-05,
+      "loss": 0.1624,
+      "step": 2166
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789468368852984e-05,
+      "loss": 0.1621,
+      "step": 2167
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892535898027804e-05,
+      "loss": 0.165,
+      "step": 2168
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890387141555775e-05,
+      "loss": 0.1466,
+      "step": 2169
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788823741937675e-05,
+      "loss": 0.1441,
+      "step": 2170
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886086731753826e-05,
+      "loss": 0.1555,
+      "step": 2171
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788393507895023e-05,
+      "loss": 0.1513,
+      "step": 2172
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78817824612293e-05,
+      "loss": 0.1655,
+      "step": 2173
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879628878854502e-05,
+      "loss": 0.1424,
+      "step": 2174
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787747433208941e-05,
+      "loss": 0.1569,
+      "step": 2175
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7875318821197715e-05,
+      "loss": 0.1549,
+      "step": 2176
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873162346443238e-05,
+      "loss": 0.1573,
+      "step": 2177
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787100490808991e-05,
+      "loss": 0.1638,
+      "step": 2178
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868846506401775e-05,
+      "loss": 0.1559,
+      "step": 2179
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786668714164301e-05,
+      "loss": 0.1406,
+      "step": 2180
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864526814077886e-05,
+      "loss": 0.1452,
+      "step": 2181
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862365523970818e-05,
+      "loss": 0.1583,
+      "step": 2182
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7860203271586326e-05,
+      "loss": 0.16,
+      "step": 2183
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7858040057189042e-05,
+      "loss": 0.1794,
+      "step": 2184
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785587588104373e-05,
+      "loss": 0.1655,
+      "step": 2185
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785371074341526e-05,
+      "loss": 0.1623,
+      "step": 2186
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7851544644568627e-05,
+      "loss": 0.1741,
+      "step": 2187
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849377584768934e-05,
+      "loss": 0.1703,
+      "step": 2188
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847209564281415e-05,
+      "loss": 0.1781,
+      "step": 2189
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784504058337141e-05,
+      "loss": 0.156,
+      "step": 2190
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784287064230439e-05,
+      "loss": 0.1394,
+      "step": 2191
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784069974134592e-05,
+      "loss": 0.1644,
+      "step": 2192
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783852788076171e-05,
+      "loss": 0.1541,
+      "step": 2193
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783635506081757e-05,
+      "loss": 0.1456,
+      "step": 2194
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783418128177943e-05,
+      "loss": 0.1704,
+      "step": 2195
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783200654391334e-05,
+      "loss": 0.1465,
+      "step": 2196
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829830847485475e-05,
+      "loss": 0.154,
+      "step": 2197
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782765419276211e-05,
+      "loss": 0.14,
+      "step": 2198
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825476580009648e-05,
+      "loss": 0.15,
+      "step": 2199
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823298009494613e-05,
+      "loss": 0.1583,
+      "step": 2200
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782111848148363e-05,
+      "loss": 0.1795,
+      "step": 2201
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818937996243467e-05,
+      "loss": 0.1539,
+      "step": 2202
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816756554040983e-05,
+      "loss": 0.1634,
+      "step": 2203
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781457415514317e-05,
+      "loss": 0.1474,
+      "step": 2204
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812390799817127e-05,
+      "loss": 0.1643,
+      "step": 2205
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7810206488330087e-05,
+      "loss": 0.1454,
+      "step": 2206
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808021220949375e-05,
+      "loss": 0.162,
+      "step": 2207
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805834997942456e-05,
+      "loss": 0.1645,
+      "step": 2208
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7803647819576898e-05,
+      "loss": 0.1649,
+      "step": 2209
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801459686120398e-05,
+      "loss": 0.1475,
+      "step": 2210
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779927059784075e-05,
+      "loss": 0.1584,
+      "step": 2211
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797080555005885e-05,
+      "loss": 0.1705,
+      "step": 2212
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794889557883842e-05,
+      "loss": 0.1448,
+      "step": 2213
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792697606742777e-05,
+      "loss": 0.1524,
+      "step": 2214
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790504701850964e-05,
+      "loss": 0.1471,
+      "step": 2215
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788310843476796e-05,
+      "loss": 0.1574,
+      "step": 2216
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786116031888774e-05,
+      "loss": 0.1687,
+      "step": 2217
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783920267355527e-05,
+      "loss": 0.1784,
+      "step": 2218
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781723550145794e-05,
+      "loss": 0.1601,
+      "step": 2219
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777952588052843e-05,
+      "loss": 0.1494,
+      "step": 2220
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777732725877241e-05,
+      "loss": 0.1622,
+      "step": 2221
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7775127685146824e-05,
+      "loss": 0.157,
+      "step": 2222
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772927159920882e-05,
+      "loss": 0.1364,
+      "step": 2223
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7770725683363896e-05,
+      "loss": 0.1516,
+      "step": 2224
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776852325574532e-05,
+      "loss": 0.1582,
+      "step": 2225
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77663198773347e-05,
+      "loss": 0.166,
+      "step": 2226
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776411554840171e-05,
+      "loss": 0.1558,
+      "step": 2227
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761910269216138e-05,
+      "loss": 0.1722,
+      "step": 2228
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775970404004789e-05,
+      "loss": 0.1521,
+      "step": 2229
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757496861166993e-05,
+      "loss": 0.174,
+      "step": 2230
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755288732843575e-05,
+      "loss": 0.1475,
+      "step": 2231
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7753079655347887e-05,
+      "loss": 0.1454,
+      "step": 2232
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7750869628950313e-05,
+      "loss": 0.1472,
+      "step": 2233
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748658653921328e-05,
+      "loss": 0.1502,
+      "step": 2234
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746446730531537e-05,
+      "loss": 0.1509,
+      "step": 2235
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774423385905166e-05,
+      "loss": 0.1672,
+      "step": 2236
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7742020039752525e-05,
+      "loss": 0.1495,
+      "step": 2237
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739805272905087e-05,
+      "loss": 0.1589,
+      "step": 2238
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773758955878041e-05,
+      "loss": 0.1338,
+      "step": 2239
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735372897649678e-05,
+      "loss": 0.1534,
+      "step": 2240
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7733155289784186e-05,
+      "loss": 0.1616,
+      "step": 2241
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773093673545535e-05,
+      "loss": 0.1621,
+      "step": 2242
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728717234934696e-05,
+      "loss": 0.158,
+      "step": 2243
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772649678849388e-05,
+      "loss": 0.144,
+      "step": 2244
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724275396404647e-05,
+      "loss": 0.1465,
+      "step": 2245
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7722053058938886e-05,
+      "loss": 0.1627,
+      "step": 2246
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7719829776368585e-05,
+      "loss": 0.1629,
+      "step": 2247
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717605548965856e-05,
+      "loss": 0.1519,
+      "step": 2248
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715380377002915e-05,
+      "loss": 0.143,
+      "step": 2249
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7713154260752112e-05,
+      "loss": 0.132,
+      "step": 2250
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710927200485893e-05,
+      "loss": 0.1535,
+      "step": 2251
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708699196476833e-05,
+      "loss": 0.1575,
+      "step": 2252
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770647024899762e-05,
+      "loss": 0.1756,
+      "step": 2253
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704240358321047e-05,
+      "loss": 0.1569,
+      "step": 2254
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770200952472004e-05,
+      "loss": 0.1522,
+      "step": 2255
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769977774846763e-05,
+      "loss": 0.1584,
+      "step": 2256
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697545029836964e-05,
+      "loss": 0.142,
+      "step": 2257
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76953113691013e-05,
+      "loss": 0.1606,
+      "step": 2258
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693076766534022e-05,
+      "loss": 0.1429,
+      "step": 2259
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7690841222408625e-05,
+      "loss": 0.146,
+      "step": 2260
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688604736998715e-05,
+      "loss": 0.1526,
+      "step": 2261
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7686367310578014e-05,
+      "loss": 0.1589,
+      "step": 2262
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684128943420366e-05,
+      "loss": 0.1719,
+      "step": 2263
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681889635799723e-05,
+      "loss": 0.1737,
+      "step": 2264
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679649387990157e-05,
+      "loss": 0.1441,
+      "step": 2265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767740820026585e-05,
+      "loss": 0.1474,
+      "step": 2266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.76751660729011e-05,
+      "loss": 0.1652,
+      "step": 2267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7672923006170324e-05,
+      "loss": 0.1431,
+      "step": 2268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7670679000348055e-05,
+      "loss": 0.1858,
+      "step": 2269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766843405570893e-05,
+      "loss": 0.1611,
+      "step": 2270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666188172527717e-05,
+      "loss": 0.1537,
+      "step": 2271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766394135107928e-05,
+      "loss": 0.1529,
+      "step": 2272
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766169359163862e-05,
+      "loss": 0.1734,
+      "step": 2273
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7659444894480835e-05,
+      "loss": 0.1351,
+      "step": 2274
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7657195259881144e-05,
+      "loss": 0.177,
+      "step": 2275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765494468811488e-05,
+      "loss": 0.1499,
+      "step": 2276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765269317945749e-05,
+      "loss": 0.1586,
+      "step": 2277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765044073418454e-05,
+      "loss": 0.1539,
+      "step": 2278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7648187352571707e-05,
+      "loss": 0.1512,
+      "step": 2279
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764593303489478e-05,
+      "loss": 0.1544,
+      "step": 2280
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643677781429672e-05,
+      "loss": 0.1493,
+      "step": 2281
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641421592452395e-05,
+      "loss": 0.1347,
+      "step": 2282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639164468239092e-05,
+      "loss": 0.1738,
+      "step": 2283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763690640906601e-05,
+      "loss": 0.1604,
+      "step": 2284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634647415209517e-05,
+      "loss": 0.1526,
+      "step": 2285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632387486946087e-05,
+      "loss": 0.1542,
+      "step": 2286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630126624552317e-05,
+      "loss": 0.158,
+      "step": 2287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7627864828304916e-05,
+      "loss": 0.173,
+      "step": 2288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7625602098480697e-05,
+      "loss": 0.1543,
+      "step": 2289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623338435356607e-05,
+      "loss": 0.1587,
+      "step": 2290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7621073839209693e-05,
+      "loss": 0.1544,
+      "step": 2291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7618808310317117e-05,
+      "loss": 0.1743,
+      "step": 2292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616541848956158e-05,
+      "loss": 0.1499,
+      "step": 2293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7614274455404215e-05,
+      "loss": 0.1669,
+      "step": 2294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612006129938788e-05,
+      "loss": 0.1484,
+      "step": 2295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609736872837498e-05,
+      "loss": 0.1414,
+      "step": 2296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760746668437809e-05,
+      "loss": 0.182,
+      "step": 2297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76051955648384e-05,
+      "loss": 0.1574,
+      "step": 2298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76029235144964e-05,
+      "loss": 0.1737,
+      "step": 2299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760065053363016e-05,
+      "loss": 0.1547,
+      "step": 2300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598376622517878e-05,
+      "loss": 0.1603,
+      "step": 2301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7596101781437854e-05,
+      "loss": 0.1518,
+      "step": 2302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593826010668507e-05,
+      "loss": 0.1324,
+      "step": 2303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7591549310488373e-05,
+      "loss": 0.1438,
+      "step": 2304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589271681176092e-05,
+      "loss": 0.1538,
+      "step": 2305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758699312301043e-05,
+      "loss": 0.1502,
+      "step": 2306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7584713636270255e-05,
+      "loss": 0.1586,
+      "step": 2307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7582433221234557e-05,
+      "loss": 0.1456,
+      "step": 2308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7580151878182436e-05,
+      "loss": 0.1637,
+      "step": 2309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577869607393108e-05,
+      "loss": 0.1716,
+      "step": 2310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75755864091459e-05,
+      "loss": 0.1637,
+      "step": 2311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7573302283720254e-05,
+      "loss": 0.1626,
+      "step": 2312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571017231395718e-05,
+      "loss": 0.1533,
+      "step": 2313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756873125245197e-05,
+      "loss": 0.1562,
+      "step": 2314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566444347168792e-05,
+      "loss": 0.1597,
+      "step": 2315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564156515826067e-05,
+      "loss": 0.1615,
+      "step": 2316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561867758703815e-05,
+      "loss": 0.1576,
+      "step": 2317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559578076082156e-05,
+      "loss": 0.1645,
+      "step": 2318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755728746824132e-05,
+      "loss": 0.1687,
+      "step": 2319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554995935461656e-05,
+      "loss": 0.1486,
+      "step": 2320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755270347802363e-05,
+      "loss": 0.18,
+      "step": 2321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7550410096207816e-05,
+      "loss": 0.1539,
+      "step": 2322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548115790294896e-05,
+      "loss": 0.1589,
+      "step": 2323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545820560565678e-05,
+      "loss": 0.147,
+      "step": 2324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7543524407301073e-05,
+      "loss": 0.1501,
+      "step": 2325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541227330782105e-05,
+      "loss": 0.1505,
+      "step": 2326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7538929331289915e-05,
+      "loss": 0.1547,
+      "step": 2327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753663040910576e-05,
+      "loss": 0.1541,
+      "step": 2328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534330564511e-05,
+      "loss": 0.1408,
+      "step": 2329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7532029797787126e-05,
+      "loss": 0.1463,
+      "step": 2330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529728109215712e-05,
+      "loss": 0.1668,
+      "step": 2331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752742549907847e-05,
+      "loss": 0.1556,
+      "step": 2332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7525121967657222e-05,
+      "loss": 0.1441,
+      "step": 2333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7522817515233893e-05,
+      "loss": 0.1481,
+      "step": 2334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7520512142090528e-05,
+      "loss": 0.1483,
+      "step": 2335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7518205848509276e-05,
+      "loss": 0.1514,
+      "step": 2336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7515898634772414e-05,
+      "loss": 0.1537,
+      "step": 2337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751359050116232e-05,
+      "loss": 0.1648,
+      "step": 2338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7511281447961486e-05,
+      "loss": 0.1386,
+      "step": 2339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508971475452514e-05,
+      "loss": 0.1517,
+      "step": 2340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506660583918127e-05,
+      "loss": 0.1558,
+      "step": 2341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7504348773641157e-05,
+      "loss": 0.1445,
+      "step": 2342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750203604490454e-05,
+      "loss": 0.1575,
+      "step": 2343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749972239799134e-05,
+      "loss": 0.1526,
+      "step": 2344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749740783318472e-05,
+      "loss": 0.153,
+      "step": 2345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7495092350767967e-05,
+      "loss": 0.1459,
+      "step": 2346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492775951024464e-05,
+      "loss": 0.1803,
+      "step": 2347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490458634237718e-05,
+      "loss": 0.1622,
+      "step": 2348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748814040069135e-05,
+      "loss": 0.1701,
+      "step": 2349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748582125066909e-05,
+      "loss": 0.1478,
+      "step": 2350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483501184454776e-05,
+      "loss": 0.1464,
+      "step": 2351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481180202332363e-05,
+      "loss": 0.1507,
+      "step": 2352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7478858304585914e-05,
+      "loss": 0.1569,
+      "step": 2353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476535491499613e-05,
+      "loss": 0.1567,
+      "step": 2354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474211763357743e-05,
+      "loss": 0.1523,
+      "step": 2355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471887120444716e-05,
+      "loss": 0.1635,
+      "step": 2356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7469561563045033e-05,
+      "loss": 0.1554,
+      "step": 2357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467235091443326e-05,
+      "loss": 0.1566,
+      "step": 2358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746490770592433e-05,
+      "loss": 0.1514,
+      "step": 2359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7462579406772903e-05,
+      "loss": 0.1479,
+      "step": 2360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7460250194274e-05,
+      "loss": 0.146,
+      "step": 2361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457920068712687e-05,
+      "loss": 0.1744,
+      "step": 2362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745558903037416e-05,
+      "loss": 0.1626,
+      "step": 2363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453257079543712e-05,
+      "loss": 0.1497,
+      "step": 2364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745092421650675e-05,
+      "loss": 0.1394,
+      "step": 2365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448590441548792e-05,
+      "loss": 0.1582,
+      "step": 2366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446255754955475e-05,
+      "loss": 0.1497,
+      "step": 2367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443920157012538e-05,
+      "loss": 0.1373,
+      "step": 2368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744158364800584e-05,
+      "loss": 0.1526,
+      "step": 2369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743924622822134e-05,
+      "loss": 0.1612,
+      "step": 2370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7436907897945118e-05,
+      "loss": 0.1519,
+      "step": 2371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434568657463362e-05,
+      "loss": 0.1693,
+      "step": 2372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7432228507062378e-05,
+      "loss": 0.156,
+      "step": 2373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429887447028575e-05,
+      "loss": 0.1469,
+      "step": 2374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427545477648474e-05,
+      "loss": 0.1555,
+      "step": 2375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742520259920871e-05,
+      "loss": 0.1592,
+      "step": 2376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422858811996033e-05,
+      "loss": 0.1838,
+      "step": 2377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7420514116297294e-05,
+      "loss": 0.1472,
+      "step": 2378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418168512399464e-05,
+      "loss": 0.1457,
+      "step": 2379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741582200058962e-05,
+      "loss": 0.1555,
+      "step": 2380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741347458115495e-05,
+      "loss": 0.1579,
+      "step": 2381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411126254382765e-05,
+      "loss": 0.1629,
+      "step": 2382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408777020560473e-05,
+      "loss": 0.1609,
+      "step": 2383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7406426879975594e-05,
+      "loss": 0.1548,
+      "step": 2384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404075832915764e-05,
+      "loss": 0.1489,
+      "step": 2385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740172387966873e-05,
+      "loss": 0.1656,
+      "step": 2386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7399371020522347e-05,
+      "loss": 0.1486,
+      "step": 2387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397017255764583e-05,
+      "loss": 0.1494,
+      "step": 2388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394662585683514e-05,
+      "loss": 0.1447,
+      "step": 2389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7392307010567334e-05,
+      "loss": 0.1533,
+      "step": 2390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7389950530704337e-05,
+      "loss": 0.1511,
+      "step": 2391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387593146382933e-05,
+      "loss": 0.1578,
+      "step": 2392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385234857891646e-05,
+      "loss": 0.1762,
+      "step": 2393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738287566551911e-05,
+      "loss": 0.1482,
+      "step": 2394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738051556955406e-05,
+      "loss": 0.151,
+      "step": 2395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378154570285355e-05,
+      "loss": 0.1537,
+      "step": 2396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375792668001958e-05,
+      "loss": 0.1608,
+      "step": 2397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737342986299294e-05,
+      "loss": 0.1304,
+      "step": 2398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371066155547488e-05,
+      "loss": 0.1551,
+      "step": 2399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368701545954894e-05,
+      "loss": 0.1502,
+      "step": 2400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366336034504573e-05,
+      "loss": 0.1674,
+      "step": 2401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363969621486024e-05,
+      "loss": 0.1532,
+      "step": 2402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361602307188884e-05,
+      "loss": 0.1605,
+      "step": 2403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735923409190289e-05,
+      "loss": 0.1598,
+      "step": 2404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356864975917885e-05,
+      "loss": 0.1567,
+      "step": 2405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735449495952383e-05,
+      "loss": 0.1531,
+      "step": 2406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7352124043010783e-05,
+      "loss": 0.1513,
+      "step": 2407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734975222666893e-05,
+      "loss": 0.1709,
+      "step": 2408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347379510788557e-05,
+      "loss": 0.1596,
+      "step": 2409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734500589566006e-05,
+      "loss": 0.1616,
+      "step": 2410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7342631381573944e-05,
+      "loss": 0.1555,
+      "step": 2411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734025596882083e-05,
+      "loss": 0.1587,
+      "step": 2412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337879657691442e-05,
+      "loss": 0.1596,
+      "step": 2413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335502448476623e-05,
+      "loss": 0.1586,
+      "step": 2414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333124341467317e-05,
+      "loss": 0.1596,
+      "step": 2415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330745336954585e-05,
+      "loss": 0.1443,
+      "step": 2416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7328365435229587e-05,
+      "loss": 0.1493,
+      "step": 2417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325984636583606e-05,
+      "loss": 0.1637,
+      "step": 2418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323602941308023e-05,
+      "loss": 0.1369,
+      "step": 2419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732122034969434e-05,
+      "loss": 0.154,
+      "step": 2420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318836862034164e-05,
+      "loss": 0.1529,
+      "step": 2421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316452478619204e-05,
+      "loss": 0.1455,
+      "step": 2422
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7314067199741293e-05,
+      "loss": 0.1648,
+      "step": 2423
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311681025692356e-05,
+      "loss": 0.1257,
+      "step": 2424
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7309293956764454e-05,
+      "loss": 0.158,
+      "step": 2425
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306905993249722e-05,
+      "loss": 0.1542,
+      "step": 2426
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304517135440436e-05,
+      "loss": 0.1529,
+      "step": 2427
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7302127383628966e-05,
+      "loss": 0.1381,
+      "step": 2428
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7299736738107795e-05,
+      "loss": 0.1676,
+      "step": 2429
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7297345199169512e-05,
+      "loss": 0.1521,
+      "step": 2430
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294952767106822e-05,
+      "loss": 0.1715,
+      "step": 2431
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7292559442212533e-05,
+      "loss": 0.1452,
+      "step": 2432
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7290165224779566e-05,
+      "loss": 0.1475,
+      "step": 2433
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728777011510095e-05,
+      "loss": 0.1512,
+      "step": 2434
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285374113469822e-05,
+      "loss": 0.1405,
+      "step": 2435
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7282977220179432e-05,
+      "loss": 0.1656,
+      "step": 2436
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280579435523136e-05,
+      "loss": 0.166,
+      "step": 2437
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278180759794397e-05,
+      "loss": 0.1684,
+      "step": 2438
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275781193286792e-05,
+      "loss": 0.1429,
+      "step": 2439
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273380736294e-05,
+      "loss": 0.1536,
+      "step": 2440
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270979389109826e-05,
+      "loss": 0.1596,
+      "step": 2441
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7268577152028156e-05,
+      "loss": 0.1462,
+      "step": 2442
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726617402534301e-05,
+      "loss": 0.147,
+      "step": 2443
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7263770009348503e-05,
+      "loss": 0.1608,
+      "step": 2444
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261365104338868e-05,
+      "loss": 0.1567,
+      "step": 2445
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7258959310608442e-05,
+      "loss": 0.1658,
+      "step": 2446
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256552628451666e-05,
+      "loss": 0.152,
+      "step": 2447
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254145058163097e-05,
+      "loss": 0.1519,
+      "step": 2448
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251736600037403e-05,
+      "loss": 0.1534,
+      "step": 2449
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249327254369342e-05,
+      "loss": 0.1503,
+      "step": 2450
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7246917021453812e-05,
+      "loss": 0.1387,
+      "step": 2451
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724450590158579e-05,
+      "loss": 0.1743,
+      "step": 2452
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724209389506038e-05,
+      "loss": 0.1489,
+      "step": 2453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7239681002172783e-05,
+      "loss": 0.1547,
+      "step": 2454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723726722321832e-05,
+      "loss": 0.152,
+      "step": 2455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234852558492412e-05,
+      "loss": 0.1406,
+      "step": 2456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7232437008290588e-05,
+      "loss": 0.1542,
+      "step": 2457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723002057290849e-05,
+      "loss": 0.1466,
+      "step": 2458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7227603252641863e-05,
+      "loss": 0.1782,
+      "step": 2459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722518504778657e-05,
+      "loss": 0.1542,
+      "step": 2460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222765958638572e-05,
+      "loss": 0.1645,
+      "step": 2461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220345985493942e-05,
+      "loss": 0.1464,
+      "step": 2462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721792512864886e-05,
+      "loss": 0.1633,
+      "step": 2463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215503388399622e-05,
+      "loss": 0.1292,
+      "step": 2464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213080765042616e-05,
+      "loss": 0.1628,
+      "step": 2465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210657258874353e-05,
+      "loss": 0.1447,
+      "step": 2466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7208232870191448e-05,
+      "loss": 0.1607,
+      "step": 2467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720580759929062e-05,
+      "loss": 0.1675,
+      "step": 2468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203381446468698e-05,
+      "loss": 0.1562,
+      "step": 2469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200954412022624e-05,
+      "loss": 0.1494,
+      "step": 2470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7198526496249437e-05,
+      "loss": 0.1652,
+      "step": 2471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196097699446297e-05,
+      "loss": 0.1701,
+      "step": 2472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193668021910462e-05,
+      "loss": 0.1572,
+      "step": 2473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191237463939302e-05,
+      "loss": 0.1483,
+      "step": 2474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718880602583029e-05,
+      "loss": 0.1525,
+      "step": 2475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186373707881014e-05,
+      "loss": 0.1606,
+      "step": 2476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183940510389165e-05,
+      "loss": 0.1514,
+      "step": 2477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181506433652545e-05,
+      "loss": 0.1473,
+      "step": 2478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717907147796906e-05,
+      "loss": 0.1502,
+      "step": 2479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7176635643636725e-05,
+      "loss": 0.1434,
+      "step": 2480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717419893095366e-05,
+      "loss": 0.1485,
+      "step": 2481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7171761340218102e-05,
+      "loss": 0.1368,
+      "step": 2482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716932287172838e-05,
+      "loss": 0.1469,
+      "step": 2483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7166883525782946e-05,
+      "loss": 0.1562,
+      "step": 2484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7164443302680348e-05,
+      "loss": 0.1433,
+      "step": 2485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7162002202719245e-05,
+      "loss": 0.1354,
+      "step": 2486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159560226198412e-05,
+      "loss": 0.1432,
+      "step": 2487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7157117373416713e-05,
+      "loss": 0.1573,
+      "step": 2488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154673644673134e-05,
+      "loss": 0.1629,
+      "step": 2489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715222904026676e-05,
+      "loss": 0.1483,
+      "step": 2490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7149783560496792e-05,
+      "loss": 0.1507,
+      "step": 2491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147337205662536e-05,
+      "loss": 0.1477,
+      "step": 2492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7144889976063393e-05,
+      "loss": 0.1428,
+      "step": 2493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7142441871998887e-05,
+      "loss": 0.1524,
+      "step": 2494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713999289376864e-05,
+      "loss": 0.1475,
+      "step": 2495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137543041672382e-05,
+      "loss": 0.163,
+      "step": 2496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135092316009954e-05,
+      "loss": 0.1593,
+      "step": 2497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71326407170813e-05,
+      "loss": 0.1612,
+      "step": 2498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713018824518647e-05,
+      "loss": 0.147,
+      "step": 2499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7127734900625625e-05,
+      "loss": 0.1621,
+      "step": 2500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712528068369903e-05,
+      "loss": 0.1564,
+      "step": 2501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122825594707057e-05,
+      "loss": 0.1449,
+      "step": 2502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7120369633950186e-05,
+      "loss": 0.1394,
+      "step": 2503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117912801729e-05,
+      "loss": 0.1403,
+      "step": 2504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115455098344197e-05,
+      "loss": 0.1609,
+      "step": 2505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711299652409657e-05,
+      "loss": 0.1375,
+      "step": 2506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711053707928703e-05,
+      "loss": 0.1476,
+      "step": 2507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7108076764216585e-05,
+      "loss": 0.1466,
+      "step": 2508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105615579186356e-05,
+      "loss": 0.1671,
+      "step": 2509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710315352449757e-05,
+      "loss": 0.163,
+      "step": 2510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100690600451556e-05,
+      "loss": 0.1467,
+      "step": 2511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709822680734975e-05,
+      "loss": 0.142,
+      "step": 2512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095762145493703e-05,
+      "loss": 0.1644,
+      "step": 2513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093296615185058e-05,
+      "loss": 0.1645,
+      "step": 2514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7090830216725578e-05,
+      "loss": 0.152,
+      "step": 2515
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708836295041712e-05,
+      "loss": 0.1513,
+      "step": 2516
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708589481656166e-05,
+      "loss": 0.1625,
+      "step": 2517
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083425815461273e-05,
+      "loss": 0.1337,
+      "step": 2518
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080955947418137e-05,
+      "loss": 0.1706,
+      "step": 2519
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7078485212734542e-05,
+      "loss": 0.1628,
+      "step": 2520
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076013611712886e-05,
+      "loss": 0.1685,
+      "step": 2521
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707354114465566e-05,
+      "loss": 0.1803,
+      "step": 2522
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.1507,
+      "step": 2523
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068593613645045e-05,
+      "loss": 0.1394,
+      "step": 2524
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706611855029719e-05,
+      "loss": 0.1485,
+      "step": 2525
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7063642622124824e-05,
+      "loss": 0.1428,
+      "step": 2526
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061165829430985e-05,
+      "loss": 0.1402,
+      "step": 2527
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7058688172518802e-05,
+      "loss": 0.1415,
+      "step": 2528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056209651691526e-05,
+      "loss": 0.1486,
+      "step": 2529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053730267252493e-05,
+      "loss": 0.1562,
+      "step": 2530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051250019505163e-05,
+      "loss": 0.1402,
+      "step": 2531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704876890875309e-05,
+      "loss": 0.1551,
+      "step": 2532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046286935299945e-05,
+      "loss": 0.1393,
+      "step": 2533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043804099449486e-05,
+      "loss": 0.1385,
+      "step": 2534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041320401505598e-05,
+      "loss": 0.146,
+      "step": 2535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703883584177226e-05,
+      "loss": 0.1677,
+      "step": 2536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703635042055355e-05,
+      "loss": 0.1627,
+      "step": 2537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703386413815367e-05,
+      "loss": 0.1454,
+      "step": 2538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031376994876916e-05,
+      "loss": 0.1548,
+      "step": 2539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702888899102768e-05,
+      "loss": 0.1348,
+      "step": 2540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702640012691048e-05,
+      "loss": 0.1544,
+      "step": 2541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023910402829925e-05,
+      "loss": 0.1594,
+      "step": 2542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021419819090734e-05,
+      "loss": 0.1463,
+      "step": 2543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018928375997726e-05,
+      "loss": 0.154,
+      "step": 2544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701643607385584e-05,
+      "loss": 0.1447,
+      "step": 2545
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.70139429129701e-05,
+      "loss": 0.1565,
+      "step": 2546
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011448893645647e-05,
+      "loss": 0.1564,
+      "step": 2547
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7008954016187725e-05,
+      "loss": 0.1395,
+      "step": 2548
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006458280901684e-05,
+      "loss": 0.1562,
+      "step": 2549
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7003961688092973e-05,
+      "loss": 0.1615,
+      "step": 2550
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7001464238067163e-05,
+      "loss": 0.1602,
+      "step": 2551
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6998965931129904e-05,
+      "loss": 0.1533,
+      "step": 2552
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6996466767586972e-05,
+      "loss": 0.161,
+      "step": 2553
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993966747744237e-05,
+      "loss": 0.1592,
+      "step": 2554
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6991465871907678e-05,
+      "loss": 0.156,
+      "step": 2555
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988964140383383e-05,
+      "loss": 0.1381,
+      "step": 2556
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986461553477536e-05,
+      "loss": 0.1515,
+      "step": 2557
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983958111496428e-05,
+      "loss": 0.1362,
+      "step": 2558
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981453814746455e-05,
+      "loss": 0.1506,
+      "step": 2559
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697894866353412e-05,
+      "loss": 0.142,
+      "step": 2560
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976442658166034e-05,
+      "loss": 0.1534,
+      "step": 2561
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69739357989489e-05,
+      "loss": 0.1446,
+      "step": 2562
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697142808618954e-05,
+      "loss": 0.1527,
+      "step": 2563
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696891952019487e-05,
+      "loss": 0.1483,
+      "step": 2564
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6966410101271915e-05,
+      "loss": 0.1491,
+      "step": 2565
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6963899829727802e-05,
+      "loss": 0.138,
+      "step": 2566
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961388705869767e-05,
+      "loss": 0.149,
+      "step": 2567
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958876730005146e-05,
+      "loss": 0.1745,
+      "step": 2568
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6956363902441385e-05,
+      "loss": 0.1408,
+      "step": 2569
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953850223486018e-05,
+      "loss": 0.1439,
+      "step": 2570
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951335693446704e-05,
+      "loss": 0.1314,
+      "step": 2571
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6948820312631195e-05,
+      "loss": 0.1724,
+      "step": 2572
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946304081347352e-05,
+      "loss": 0.1466,
+      "step": 2573
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943786999903133e-05,
+      "loss": 0.1543,
+      "step": 2574
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6941269068606608e-05,
+      "loss": 0.1393,
+      "step": 2575
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938750287765943e-05,
+      "loss": 0.1414,
+      "step": 2576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6936230657689416e-05,
+      "loss": 0.1651,
+      "step": 2577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6933710178685406e-05,
+      "loss": 0.1638,
+      "step": 2578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931188851062393e-05,
+      "loss": 0.1761,
+      "step": 2579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928666675128962e-05,
+      "loss": 0.1554,
+      "step": 2580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692614365119381e-05,
+      "loss": 0.1417,
+      "step": 2581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692361977956572e-05,
+      "loss": 0.1557,
+      "step": 2582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6921095060553595e-05,
+      "loss": 0.1563,
+      "step": 2583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6918569494466436e-05,
+      "loss": 0.1397,
+      "step": 2584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691604308161335e-05,
+      "loss": 0.1527,
+      "step": 2585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691351582230354e-05,
+      "loss": 0.1587,
+      "step": 2586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6910987716846327e-05,
+      "loss": 0.152,
+      "step": 2587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908458765551116e-05,
+      "loss": 0.1596,
+      "step": 2588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905928968727435e-05,
+      "loss": 0.1592,
+      "step": 2589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69033983266849e-05,
+      "loss": 0.1656,
+      "step": 2590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900866839733237e-05,
+      "loss": 0.136,
+      "step": 2591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6898334508182282e-05,
+      "loss": 0.1485,
+      "step": 2592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6895801332341967e-05,
+      "loss": 0.157,
+      "step": 2593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893267312522323e-05,
+      "loss": 0.1513,
+      "step": 2594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689073244903349e-05,
+      "loss": 0.1649,
+      "step": 2595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888196742185713e-05,
+      "loss": 0.1506,
+      "step": 2596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688566019228934e-05,
+      "loss": 0.1391,
+      "step": 2597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883122799654814e-05,
+      "loss": 0.1646,
+      "step": 2598
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880584564592696e-05,
+      "loss": 0.1353,
+      "step": 2599
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6878045487413634e-05,
+      "loss": 0.1369,
+      "step": 2600
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687550556842839e-05,
+      "loss": 0.1545,
+      "step": 2601
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6872964807947822e-05,
+      "loss": 0.1474,
+      "step": 2602
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68704232062829e-05,
+      "loss": 0.1485,
+      "step": 2603
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867880763744686e-05,
+      "loss": 0.1488,
+      "step": 2604
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865337480644354e-05,
+      "loss": 0.1582,
+      "step": 2605
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6862793357293176e-05,
+      "loss": 0.1388,
+      "step": 2606
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686024839400253e-05,
+      "loss": 0.1391,
+      "step": 2607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685770259108389e-05,
+      "loss": 0.1441,
+      "step": 2608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6855155948848843e-05,
+      "loss": 0.1553,
+      "step": 2609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685260846760907e-05,
+      "loss": 0.1523,
+      "step": 2610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850060147676356e-05,
+      "loss": 0.1424,
+      "step": 2611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847510989362595e-05,
+      "loss": 0.1258,
+      "step": 2612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684496099297978e-05,
+      "loss": 0.1621,
+      "step": 2613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684241015884e-05,
+      "loss": 0.1452,
+      "step": 2614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6839858487255462e-05,
+      "loss": 0.1472,
+      "step": 2615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6837305978538457e-05,
+      "loss": 0.1539,
+      "step": 2616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6834752633001388e-05,
+      "loss": 0.1567,
+      "step": 2617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832198450956766e-05,
+      "loss": 0.1476,
+      "step": 2618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6829643432717193e-05,
+      "loss": 0.147,
+      "step": 2619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6827087578595384e-05,
+      "loss": 0.1571,
+      "step": 2620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682453088890414e-05,
+      "loss": 0.1644,
+      "step": 2621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821973363956385e-05,
+      "loss": 0.1657,
+      "step": 2622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681941500406513e-05,
+      "loss": 0.1479,
+      "step": 2623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68168558095435e-05,
+      "loss": 0.1522,
+      "step": 2624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814295780704706e-05,
+      "loss": 0.1478,
+      "step": 2625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811734917862076e-05,
+      "loss": 0.1585,
+      "step": 2626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680917322132904e-05,
+      "loss": 0.1557,
+      "step": 2627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6806610691419117e-05,
+      "loss": 0.1421,
+      "step": 2628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680404732844594e-05,
+      "loss": 0.1588,
+      "step": 2629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801483132723234e-05,
+      "loss": 0.1422,
+      "step": 2630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6798918104564837e-05,
+      "loss": 0.165,
+      "step": 2631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796352244284688e-05,
+      "loss": 0.146,
+      "step": 2632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679378555219681e-05,
+      "loss": 0.1477,
+      "step": 2633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791218028615353e-05,
+      "loss": 0.1542,
+      "step": 2634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788649673854555e-05,
+      "loss": 0.1419,
+      "step": 2635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786080488228758e-05,
+      "loss": 0.1589,
+      "step": 2636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.67835104720524e-05,
+      "loss": 0.1443,
+      "step": 2637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678093962564003e-05,
+      "loss": 0.1392,
+      "step": 2638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778367949306295e-05,
+      "loss": 0.1492,
+      "step": 2639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677579544336594e-05,
+      "loss": 0.1627,
+      "step": 2640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773222108133823e-05,
+      "loss": 0.1329,
+      "step": 2641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770647943924884e-05,
+      "loss": 0.1567,
+      "step": 2642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768072951054187e-05,
+      "loss": 0.146,
+      "step": 2643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676549712983688e-05,
+      "loss": 0.1628,
+      "step": 2644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762920480588215e-05,
+      "loss": 0.1436,
+      "step": 2645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760343003623557e-05,
+      "loss": 0.1327,
+      "step": 2646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757764699258355e-05,
+      "loss": 0.1536,
+      "step": 2647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6755185567808185e-05,
+      "loss": 0.1454,
+      "step": 2648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752605609588687e-05,
+      "loss": 0.1406,
+      "step": 2649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750024824915636e-05,
+      "loss": 0.1549,
+      "step": 2650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674744321410489e-05,
+      "loss": 0.1417,
+      "step": 2651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674486077747242e-05,
+      "loss": 0.148,
+      "step": 2652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674227751533428e-05,
+      "loss": 0.1552,
+      "step": 2653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739693428006645e-05,
+      "loss": 0.1715,
+      "step": 2654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673710851580578e-05,
+      "loss": 0.1383,
+      "step": 2655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734522779048054e-05,
+      "loss": 0.1561,
+      "step": 2656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731936218049933e-05,
+      "loss": 0.1617,
+      "step": 2657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672934883312799e-05,
+      "loss": 0.1468,
+      "step": 2658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6726760624598895e-05,
+      "loss": 0.1712,
+      "step": 2659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724171592779422e-05,
+      "loss": 0.153,
+      "step": 2660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6721581737986438e-05,
+      "loss": 0.1573,
+      "step": 2661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671899106053692e-05,
+      "loss": 0.1435,
+      "step": 2662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671639956074794e-05,
+      "loss": 0.1427,
+      "step": 2663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713807238936673e-05,
+      "loss": 0.1531,
+      "step": 2664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711214095420398e-05,
+      "loss": 0.1531,
+      "step": 2665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708620130516487e-05,
+      "loss": 0.1483,
+      "step": 2666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6706025344542412e-05,
+      "loss": 0.1475,
+      "step": 2667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6703429737815756e-05,
+      "loss": 0.1396,
+      "step": 2668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6700833310654197e-05,
+      "loss": 0.1436,
+      "step": 2669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669823606337551e-05,
+      "loss": 0.1596,
+      "step": 2670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.669563799629757e-05,
+      "loss": 0.1704,
+      "step": 2671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693039109738357e-05,
+      "loss": 0.1331,
+      "step": 2672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690439404015958e-05,
+      "loss": 0.1597,
+      "step": 2673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687838879448536e-05,
+      "loss": 0.1479,
+      "step": 2674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6685237536354383e-05,
+      "loss": 0.1458,
+      "step": 2675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6682635375051872e-05,
+      "loss": 0.1532,
+      "step": 2676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680032395859483e-05,
+      "loss": 0.1385,
+      "step": 2677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6677428599095796e-05,
+      "loss": 0.1637,
+      "step": 2678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6674823985079495e-05,
+      "loss": 0.1496,
+      "step": 2679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6672218554129352e-05,
+      "loss": 0.1493,
+      "step": 2680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6669612306564253e-05,
+      "loss": 0.1667,
+      "step": 2681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666700524270317e-05,
+      "loss": 0.1442,
+      "step": 2682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664397362865192e-05,
+      "loss": 0.15,
+      "step": 2683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666178866736949e-05,
+      "loss": 0.1513,
+      "step": 2684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665917915653535e-05,
+      "loss": 0.149,
+      "step": 2685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6656568830682143e-05,
+      "loss": 0.1519,
+      "step": 2686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653957690129353e-05,
+      "loss": 0.1503,
+      "step": 2687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6651345735196555e-05,
+      "loss": 0.1485,
+      "step": 2688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648732966203432e-05,
+      "loss": 0.1487,
+      "step": 2689
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6646119383469757e-05,
+      "loss": 0.1548,
+      "step": 2690
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643504987315412e-05,
+      "loss": 0.1642,
+      "step": 2691
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640889778060367e-05,
+      "loss": 0.1414,
+      "step": 2692
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6638273756024704e-05,
+      "loss": 0.1313,
+      "step": 2693
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6635656921528594e-05,
+      "loss": 0.1455,
+      "step": 2694
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6633039274892317e-05,
+      "loss": 0.1393,
+      "step": 2695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6630420816436248e-05,
+      "loss": 0.1449,
+      "step": 2696
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627801546480856e-05,
+      "loss": 0.1628,
+      "step": 2697
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625181465346717e-05,
+      "loss": 0.1491,
+      "step": 2698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66225605733545e-05,
+      "loss": 0.1375,
+      "step": 2699
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6619938870824985e-05,
+      "loss": 0.1524,
+      "step": 2700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661731635807904e-05,
+      "loss": 0.1332,
+      "step": 2701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614693035437625e-05,
+      "loss": 0.1488,
+      "step": 2702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612068903221825e-05,
+      "loss": 0.1463,
+      "step": 2703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66094439617528e-05,
+      "loss": 0.1463,
+      "step": 2704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606818211351818e-05,
+      "loss": 0.1495,
+      "step": 2705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604191652340244e-05,
+      "loss": 0.1695,
+      "step": 2706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601564285039545e-05,
+      "loss": 0.1606,
+      "step": 2707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598936109771285e-05,
+      "loss": 0.1358,
+      "step": 2708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659630712685713e-05,
+      "loss": 0.156,
+      "step": 2709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6593677336618842e-05,
+      "loss": 0.1502,
+      "step": 2710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6591046739378275e-05,
+      "loss": 0.1574,
+      "step": 2711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588415335457397e-05,
+      "loss": 0.1593,
+      "step": 2712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585783125178262e-05,
+      "loss": 0.1519,
+      "step": 2713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6583150108863026e-05,
+      "loss": 0.1565,
+      "step": 2714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580516286833947e-05,
+      "loss": 0.1542,
+      "step": 2715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657788165941338e-05,
+      "loss": 0.1609,
+      "step": 2716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6575246226923775e-05,
+      "loss": 0.1361,
+      "step": 2717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6572609989687687e-05,
+      "loss": 0.1522,
+      "step": 2718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569972948027764e-05,
+      "loss": 0.1523,
+      "step": 2719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567335102266752e-05,
+      "loss": 0.1542,
+      "step": 2720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6564696452727506e-05,
+      "loss": 0.131,
+      "step": 2721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656205699973296e-05,
+      "loss": 0.1447,
+      "step": 2722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6559416743606162e-05,
+      "loss": 0.1425,
+      "step": 2723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655677568467026e-05,
+      "loss": 0.1615,
+      "step": 2724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6554133823248484e-05,
+      "loss": 0.1336,
+      "step": 2725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551491159664177e-05,
+      "loss": 0.1591,
+      "step": 2726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654884769424078e-05,
+      "loss": 0.1503,
+      "step": 2727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654620342730182e-05,
+      "loss": 0.163,
+      "step": 2728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6543558359170932e-05,
+      "loss": 0.1727,
+      "step": 2729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540912490171847e-05,
+      "loss": 0.1302,
+      "step": 2730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6538265820628394e-05,
+      "loss": 0.1575,
+      "step": 2731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535618350864503e-05,
+      "loss": 0.1224,
+      "step": 2732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532970081204195e-05,
+      "loss": 0.156,
+      "step": 2733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6530321011971592e-05,
+      "loss": 0.1748,
+      "step": 2734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6527671143490913e-05,
+      "loss": 0.1616,
+      "step": 2735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652502047608648e-05,
+      "loss": 0.1696,
+      "step": 2736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652236901008271e-05,
+      "loss": 0.1306,
+      "step": 2737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6519716745804112e-05,
+      "loss": 0.1576,
+      "step": 2738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517063683575304e-05,
+      "loss": 0.163,
+      "step": 2739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514409823720988e-05,
+      "loss": 0.1314,
+      "step": 2740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6511755166565973e-05,
+      "loss": 0.1429,
+      "step": 2741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6509099712435168e-05,
+      "loss": 0.1305,
+      "step": 2742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650644346165357e-05,
+      "loss": 0.1656,
+      "step": 2743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6503786414546277e-05,
+      "loss": 0.1484,
+      "step": 2744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650112857143849e-05,
+      "loss": 0.1546,
+      "step": 2745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498469932655503e-05,
+      "loss": 0.1472,
+      "step": 2746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6495810498522704e-05,
+      "loss": 0.1527,
+      "step": 2747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493150269365584e-05,
+      "loss": 0.1467,
+      "step": 2748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6490489245509732e-05,
+      "loss": 0.1457,
+      "step": 2749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648782742728083e-05,
+      "loss": 0.1625,
+      "step": 2750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485164815004654e-05,
+      "loss": 0.1537,
+      "step": 2751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648250140900709e-05,
+      "loss": 0.1553,
+      "step": 2752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6479837209614107e-05,
+      "loss": 0.1489,
+      "step": 2753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647717221715178e-05,
+      "loss": 0.1517,
+      "step": 2754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474506431946277e-05,
+      "loss": 0.154,
+      "step": 2755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6471839854323862e-05,
+      "loss": 0.1432,
+      "step": 2756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469172484610902e-05,
+      "loss": 0.1445,
+      "step": 2757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466504323133857e-05,
+      "loss": 0.1559,
+      "step": 2758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6463835370219286e-05,
+      "loss": 0.1407,
+      "step": 2759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646116562619384e-05,
+      "loss": 0.1453,
+      "step": 2760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645849509138427e-05,
+      "loss": 0.1516,
+      "step": 2761
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645582376611742e-05,
+      "loss": 0.1638,
+      "step": 2762
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6453151650720244e-05,
+      "loss": 0.1654,
+      "step": 2763
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6450478745519776e-05,
+      "loss": 0.1528,
+      "step": 2764
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644780505084315e-05,
+      "loss": 0.1528,
+      "step": 2765
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445130567017612e-05,
+      "loss": 0.1447,
+      "step": 2766
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442455294370482e-05,
+      "loss": 0.1609,
+      "step": 2767
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643977923322919e-05,
+      "loss": 0.1483,
+      "step": 2768
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437102383921266e-05,
+      "loss": 0.1323,
+      "step": 2769
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6434424746774328e-05,
+      "loss": 0.1617,
+      "step": 2770
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6431746322116086e-05,
+      "loss": 0.1613,
+      "step": 2771
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642906711027436e-05,
+      "loss": 0.1459,
+      "step": 2772
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642638711157706e-05,
+      "loss": 0.1507,
+      "step": 2773
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6423706326352186e-05,
+      "loss": 0.1463,
+      "step": 2774
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6421024754927848e-05,
+      "loss": 0.1455,
+      "step": 2775
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6418342397632238e-05,
+      "loss": 0.1489,
+      "step": 2776
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415659254793654e-05,
+      "loss": 0.1485,
+      "step": 2777
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412975326740485e-05,
+      "loss": 0.1435,
+      "step": 2778
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6410290613801216e-05,
+      "loss": 0.1508,
+      "step": 2779
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407605116304433e-05,
+      "loss": 0.142,
+      "step": 2780
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404918834578817e-05,
+      "loss": 0.1515,
+      "step": 2781
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6402231768953137e-05,
+      "loss": 0.1673,
+      "step": 2782
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639954391975627e-05,
+      "loss": 0.1555,
+      "step": 2783
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639685528731718e-05,
+      "loss": 0.1689,
+      "step": 2784
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6394165871964926e-05,
+      "loss": 0.1398,
+      "step": 2785
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391475674028672e-05,
+      "loss": 0.1373,
+      "step": 2786
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388784693837668e-05,
+      "loss": 0.1357,
+      "step": 2787
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638609293172127e-05,
+      "loss": 0.1368,
+      "step": 2788
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383400388008913e-05,
+      "loss": 0.1546,
+      "step": 2789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380707063030147e-05,
+      "loss": 0.1722,
+      "step": 2790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378012957114607e-05,
+      "loss": 0.1519,
+      "step": 2791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375318070592022e-05,
+      "loss": 0.134,
+      "step": 2792
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372622403792227e-05,
+      "loss": 0.1346,
+      "step": 2793
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636992595704514e-05,
+      "loss": 0.1467,
+      "step": 2794
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367228730680786e-05,
+      "loss": 0.1273,
+      "step": 2795
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364530725029267e-05,
+      "loss": 0.137,
+      "step": 2796
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361831940420808e-05,
+      "loss": 0.1361,
+      "step": 2797
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63591323771857e-05,
+      "loss": 0.1506,
+      "step": 2798
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356432035654357e-05,
+      "loss": 0.1453,
+      "step": 2799
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353730916157264e-05,
+      "loss": 0.1444,
+      "step": 2800
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6351029019025016e-05,
+      "loss": 0.1596,
+      "step": 2801
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63483263445883e-05,
+      "loss": 0.1607,
+      "step": 2802
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6345622893177894e-05,
+      "loss": 0.1436,
+      "step": 2803
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6342918665124678e-05,
+      "loss": 0.1535,
+      "step": 2804
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634021366075962e-05,
+      "loss": 0.1503,
+      "step": 2805
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633750788041379e-05,
+      "loss": 0.1452,
+      "step": 2806
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6334801324418346e-05,
+      "loss": 0.138,
+      "step": 2807
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332093993104545e-05,
+      "loss": 0.1448,
+      "step": 2808
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632938588680374e-05,
+      "loss": 0.162,
+      "step": 2809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6326677005847375e-05,
+      "loss": 0.1456,
+      "step": 2810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323967350566992e-05,
+      "loss": 0.1551,
+      "step": 2811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6321256921294225e-05,
+      "loss": 0.1494,
+      "step": 2812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6318545718360805e-05,
+      "loss": 0.1517,
+      "step": 2813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631583374209856e-05,
+      "loss": 0.1345,
+      "step": 2814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313120992839403e-05,
+      "loss": 0.1417,
+      "step": 2815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310407470915355e-05,
+      "loss": 0.1594,
+      "step": 2816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630769317665852e-05,
+      "loss": 0.1481,
+      "step": 2817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6304978110401106e-05,
+      "loss": 0.1545,
+      "step": 2818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630226227247541e-05,
+      "loss": 0.1484,
+      "step": 2819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6299545663213816e-05,
+      "loss": 0.1407,
+      "step": 2820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629682828294882e-05,
+      "loss": 0.1512,
+      "step": 2821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6294110132013e-05,
+      "loss": 0.1585,
+      "step": 2822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291391210739033e-05,
+      "loss": 0.1445,
+      "step": 2823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288671519459688e-05,
+      "loss": 0.1415,
+      "step": 2824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6285951058507827e-05,
+      "loss": 0.1604,
+      "step": 2825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283229828216408e-05,
+      "loss": 0.163,
+      "step": 2826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6280507828918487e-05,
+      "loss": 0.1396,
+      "step": 2827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277785060947203e-05,
+      "loss": 0.1271,
+      "step": 2828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6275061524635805e-05,
+      "loss": 0.1621,
+      "step": 2829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272337220317625e-05,
+      "loss": 0.1553,
+      "step": 2830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269612148326088e-05,
+      "loss": 0.1514,
+      "step": 2831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626688630899472e-05,
+      "loss": 0.1594,
+      "step": 2832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626415970265713e-05,
+      "loss": 0.1496,
+      "step": 2833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626143232964704e-05,
+      "loss": 0.1569,
+      "step": 2834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258704190298245e-05,
+      "loss": 0.1401,
+      "step": 2835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6255975284944648e-05,
+      "loss": 0.1404,
+      "step": 2836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253245613920243e-05,
+      "loss": 0.1327,
+      "step": 2837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250515177559106e-05,
+      "loss": 0.1436,
+      "step": 2838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624778397619542e-05,
+      "loss": 0.1578,
+      "step": 2839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624505201016346e-05,
+      "loss": 0.1506,
+      "step": 2840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6242319279797592e-05,
+      "loss": 0.1429,
+      "step": 2841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623958578543227e-05,
+      "loss": 0.1409,
+      "step": 2842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623685152740206e-05,
+      "loss": 0.1506,
+      "step": 2843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234116506041594e-05,
+      "loss": 0.1493,
+      "step": 2844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623138072168562e-05,
+      "loss": 0.1593,
+      "step": 2845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622864417466897e-05,
+      "loss": 0.1458,
+      "step": 2846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622590686532657e-05,
+      "loss": 0.1745,
+      "step": 2847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622316879399344e-05,
+      "loss": 0.1468,
+      "step": 2848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62204299610047e-05,
+      "loss": 0.148,
+      "step": 2849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6217690366695547e-05,
+      "loss": 0.1451,
+      "step": 2850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621495001140129e-05,
+      "loss": 0.1644,
+      "step": 2851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621220889545731e-05,
+      "loss": 0.1459,
+      "step": 2852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6209467019199104e-05,
+      "loss": 0.1543,
+      "step": 2853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6206724382962245e-05,
+      "loss": 0.149,
+      "step": 2854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620398098708241e-05,
+      "loss": 0.1362,
+      "step": 2855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.620123683189536e-05,
+      "loss": 0.1483,
+      "step": 2856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6198491917736955e-05,
+      "loss": 0.1441,
+      "step": 2857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195746244943142e-05,
+      "loss": 0.1339,
+      "step": 2858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192999813849974e-05,
+      "loss": 0.1517,
+      "step": 2859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190252624793577e-05,
+      "loss": 0.1506,
+      "step": 2860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187504678110188e-05,
+      "loss": 0.1472,
+      "step": 2861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618475597413612e-05,
+      "loss": 0.1289,
+      "step": 2862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61820065132078e-05,
+      "loss": 0.1598,
+      "step": 2863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179256295661728e-05,
+      "loss": 0.1678,
+      "step": 2864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6176505321834506e-05,
+      "loss": 0.1571,
+      "step": 2865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6173753592062822e-05,
+      "loss": 0.1374,
+      "step": 2866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617100110668347e-05,
+      "loss": 0.1481,
+      "step": 2867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616824786603332e-05,
+      "loss": 0.1507,
+      "step": 2868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6165493870449346e-05,
+      "loss": 0.1506,
+      "step": 2869
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6162739120268606e-05,
+      "loss": 0.1471,
+      "step": 2870
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6159983615828258e-05,
+      "loss": 0.1406,
+      "step": 2871
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6157227357465548e-05,
+      "loss": 0.138,
+      "step": 2872
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615447034551782e-05,
+      "loss": 0.139,
+      "step": 2873
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6151712580322502e-05,
+      "loss": 0.1401,
+      "step": 2874
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6148954062217113e-05,
+      "loss": 0.1355,
+      "step": 2875
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614619479153927e-05,
+      "loss": 0.1374,
+      "step": 2876
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6143434768626694e-05,
+      "loss": 0.1423,
+      "step": 2877
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614067399381717e-05,
+      "loss": 0.1447,
+      "step": 2878
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6137912467448596e-05,
+      "loss": 0.1471,
+      "step": 2879
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6135150189858953e-05,
+      "loss": 0.1408,
+      "step": 2880
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6132387161386325e-05,
+      "loss": 0.1401,
+      "step": 2881
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129623382368872e-05,
+      "loss": 0.1454,
+      "step": 2882
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6126858853144854e-05,
+      "loss": 0.1475,
+      "step": 2883
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124093574052627e-05,
+      "loss": 0.1428,
+      "step": 2884
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121327545430635e-05,
+      "loss": 0.1487,
+      "step": 2885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6118560767617407e-05,
+      "loss": 0.1458,
+      "step": 2886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115793240951573e-05,
+      "loss": 0.159,
+      "step": 2887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6113024965771854e-05,
+      "loss": 0.1423,
+      "step": 2888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110255942417057e-05,
+      "loss": 0.1484,
+      "step": 2889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107486171226084e-05,
+      "loss": 0.1456,
+      "step": 2890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610471565253793e-05,
+      "loss": 0.1538,
+      "step": 2891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610194438669168e-05,
+      "loss": 0.1596,
+      "step": 2892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6099172374026504e-05,
+      "loss": 0.1697,
+      "step": 2893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6096399614881675e-05,
+      "loss": 0.1525,
+      "step": 2894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609362610959655e-05,
+      "loss": 0.1492,
+      "step": 2895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609085185851058e-05,
+      "loss": 0.1603,
+      "step": 2896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608807686196331e-05,
+      "loss": 0.1341,
+      "step": 2897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608530112029437e-05,
+      "loss": 0.1538,
+      "step": 2898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082524633843482e-05,
+      "loss": 0.1557,
+      "step": 2899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607974740295046e-05,
+      "loss": 0.1336,
+      "step": 2900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607696942795521e-05,
+      "loss": 0.1405,
+      "step": 2901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074190709197738e-05,
+      "loss": 0.1453,
+      "step": 2902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607141124701812e-05,
+      "loss": 0.1489,
+      "step": 2903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068631041756544e-05,
+      "loss": 0.1325,
+      "step": 2904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606585009375328e-05,
+      "loss": 0.1536,
+      "step": 2905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063068403348685e-05,
+      "loss": 0.1425,
+      "step": 2906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606028597088321e-05,
+      "loss": 0.1396,
+      "step": 2907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057502796697403e-05,
+      "loss": 0.1475,
+      "step": 2908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605471888113189e-05,
+      "loss": 0.1534,
+      "step": 2909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605193422452741e-05,
+      "loss": 0.1479,
+      "step": 2910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604914882722476e-05,
+      "loss": 0.1447,
+      "step": 2911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6046362689564852e-05,
+      "loss": 0.143,
+      "step": 2912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604357581188869e-05,
+      "loss": 0.1547,
+      "step": 2913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6040788194537353e-05,
+      "loss": 0.1491,
+      "step": 2914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603799983785202e-05,
+      "loss": 0.1355,
+      "step": 2915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603521074217396e-05,
+      "loss": 0.1418,
+      "step": 2916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603242090784453e-05,
+      "loss": 0.1709,
+      "step": 2917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602963033520518e-05,
+      "loss": 0.1542,
+      "step": 2918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026839024597448e-05,
+      "loss": 0.1457,
+      "step": 2919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024046976362962e-05,
+      "loss": 0.1383,
+      "step": 2920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021254190843444e-05,
+      "loss": 0.1401,
+      "step": 2921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6018460668380705e-05,
+      "loss": 0.1461,
+      "step": 2922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6015666409316644e-05,
+      "loss": 0.1521,
+      "step": 2923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012871413993246e-05,
+      "loss": 0.1697,
+      "step": 2924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.60100756827526e-05,
+      "loss": 0.1275,
+      "step": 2925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600727921593687e-05,
+      "loss": 0.1462,
+      "step": 2926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600448201388832e-05,
+      "loss": 0.1539,
+      "step": 2927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6001684076949304e-05,
+      "loss": 0.1621,
+      "step": 2928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5998885405462254e-05,
+      "loss": 0.1653,
+      "step": 2929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59960859997697e-05,
+      "loss": 0.1388,
+      "step": 2930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599328586021427e-05,
+      "loss": 0.1317,
+      "step": 2931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599048498713867e-05,
+      "loss": 0.1646,
+      "step": 2932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59876833808857e-05,
+      "loss": 0.1495,
+      "step": 2933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598488104179825e-05,
+      "loss": 0.1575,
+      "step": 2934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5982077970219298e-05,
+      "loss": 0.138,
+      "step": 2935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5979274166491915e-05,
+      "loss": 0.1357,
+      "step": 2936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976469630959258e-05,
+      "loss": 0.137,
+      "step": 2937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973664363964573e-05,
+      "loss": 0.1621,
+      "step": 2938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970858365851197e-05,
+      "loss": 0.1541,
+      "step": 2939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5968051636962566e-05,
+      "loss": 0.1549,
+      "step": 2940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965244177642184e-05,
+      "loss": 0.1492,
+      "step": 2941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962435988233665e-05,
+      "loss": 0.1348,
+      "step": 2942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.59596270690807e-05,
+      "loss": 0.1395,
+      "step": 2943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5956817420527077e-05,
+      "loss": 0.1512,
+      "step": 2944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5954007042916672e-05,
+      "loss": 0.1582,
+      "step": 2945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595119593659344e-05,
+      "loss": 0.1372,
+      "step": 2946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.594838410190144e-05,
+      "loss": 0.1401,
+      "step": 2947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594557153918481e-05,
+      "loss": 0.1377,
+      "step": 2948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5942758248787774e-05,
+      "loss": 0.1487,
+      "step": 2949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939944231054667e-05,
+      "loss": 0.148,
+      "step": 2950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593712948632989e-05,
+      "loss": 0.1405,
+      "step": 2951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5934314014957934e-05,
+      "loss": 0.142,
+      "step": 2952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931497817283395e-05,
+      "loss": 0.1522,
+      "step": 2953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592868089365094e-05,
+      "loss": 0.1488,
+      "step": 2954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592586324440534e-05,
+      "loss": 0.136,
+      "step": 2955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923044869891446e-05,
+      "loss": 0.1467,
+      "step": 2956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59202257704542e-05,
+      "loss": 0.1401,
+      "step": 2957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917405946438635e-05,
+      "loss": 0.1516,
+      "step": 2958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5914585398189864e-05,
+      "loss": 0.1535,
+      "step": 2959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59117641260531e-05,
+      "loss": 0.1468,
+      "step": 2960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5908942130373632e-05,
+      "loss": 0.1447,
+      "step": 2961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906119411496855e-05,
+      "loss": 0.133,
+      "step": 2962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5903295969768232e-05,
+      "loss": 0.1509,
+      "step": 2963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5900471805533336e-05,
+      "loss": 0.1542,
+      "step": 2964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5897646919137814e-05,
+      "loss": 0.1551,
+      "step": 2965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5894821310927397e-05,
+      "loss": 0.1362,
+      "step": 2966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589199498124792e-05,
+      "loss": 0.1566,
+      "step": 2967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.58891679304453e-05,
+      "loss": 0.1627,
+      "step": 2968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588634015886553e-05,
+      "loss": 0.1518,
+      "step": 2969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5883511666854712e-05,
+      "loss": 0.156,
+      "step": 2970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5880682454759026e-05,
+      "loss": 0.1519,
+      "step": 2971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.1418,
+      "step": 2972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875021871698195e-05,
+      "loss": 0.1499,
+      "step": 2973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872190501425852e-05,
+      "loss": 0.16,
+      "step": 2974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586935841245424e-05,
+      "loss": 0.1768,
+      "step": 2975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586652560512998e-05,
+      "loss": 0.1504,
+      "step": 2976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863692079799777e-05,
+      "loss": 0.1594,
+      "step": 2977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5860857836810427e-05,
+      "loss": 0.1692,
+      "step": 2978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585802287650882e-05,
+      "loss": 0.1423,
+      "step": 2979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5855187199241917e-05,
+      "loss": 0.1681,
+      "step": 2980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5852350805356786e-05,
+      "loss": 0.1713,
+      "step": 2981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584951369520057e-05,
+      "loss": 0.1568,
+      "step": 2982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584667586912051e-05,
+      "loss": 0.1467,
+      "step": 2983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584383732746392e-05,
+      "loss": 0.1442,
+      "step": 2984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584099807057821e-05,
+      "loss": 0.1387,
+      "step": 2985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838158098810888e-05,
+      "loss": 0.1612,
+      "step": 2986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5835317412509532e-05,
+      "loss": 0.144,
+      "step": 2987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5832476012021812e-05,
+      "loss": 0.1349,
+      "step": 2988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829633897695494e-05,
+      "loss": 0.1463,
+      "step": 2989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826791069878423e-05,
+      "loss": 0.1632,
+      "step": 2990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5823947528918527e-05,
+      "loss": 0.1576,
+      "step": 2991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582110327516384e-05,
+      "loss": 0.1458,
+      "step": 2992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581825830896246e-05,
+      "loss": 0.1363,
+      "step": 2993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5815412630662586e-05,
+      "loss": 0.1537,
+      "step": 2994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581256624061251e-05,
+      "loss": 0.1503,
+      "step": 2995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809719139160592e-05,
+      "loss": 0.1488,
+      "step": 2996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58068713266553e-05,
+      "loss": 0.1475,
+      "step": 2997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5804022803445164e-05,
+      "loss": 0.1548,
+      "step": 2998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580117356987883e-05,
+      "loss": 0.1366,
+      "step": 2999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5798323626305008e-05,
+      "loss": 0.1598,
+      "step": 3000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579547297307251e-05,
+      "loss": 0.1458,
+      "step": 3001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5792621610530218e-05,
+      "loss": 0.1638,
+      "step": 3002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578976953902712e-05,
+      "loss": 0.1404,
+      "step": 3003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578691675891228e-05,
+      "loss": 0.1475,
+      "step": 3004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784063270534855e-05,
+      "loss": 0.1508,
+      "step": 3005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781209074244076e-05,
+      "loss": 0.1357,
+      "step": 3006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778354170389275e-05,
+      "loss": 0.1418,
+      "step": 3007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5775498559319856e-05,
+      "loss": 0.1517,
+      "step": 3008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577264224138533e-05,
+      "loss": 0.1405,
+      "step": 3009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769785216935277e-05,
+      "loss": 0.1403,
+      "step": 3010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576692748631937e-05,
+      "loss": 0.1595,
+      "step": 3011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576406904988736e-05,
+      "loss": 0.1487,
+      "step": 3012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761209907989106e-05,
+      "loss": 0.1462,
+      "step": 3013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575835006097453e-05,
+      "loss": 0.1367,
+      "step": 3014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575548950919365e-05,
+      "loss": 0.1397,
+      "step": 3015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752628252996577e-05,
+      "loss": 0.156,
+      "step": 3016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749766292733492e-05,
+      "loss": 0.1738,
+      "step": 3017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5746903628754672e-05,
+      "loss": 0.1549,
+      "step": 3018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5744040261410488e-05,
+      "loss": 0.1577,
+      "step": 3019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741176191051377e-05,
+      "loss": 0.1366,
+      "step": 3020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573831141802788e-05,
+      "loss": 0.1403,
+      "step": 3021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5735445942690623e-05,
+      "loss": 0.1396,
+      "step": 3022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732579765390296e-05,
+      "loss": 0.1542,
+      "step": 3023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572971288647771e-05,
+      "loss": 0.1527,
+      "step": 3024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726845306303732e-05,
+      "loss": 0.1276,
+      "step": 3025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723977025219326e-05,
+      "loss": 0.1467,
+      "step": 3026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572110804357555e-05,
+      "loss": 0.1582,
+      "step": 3027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718238361723525e-05,
+      "loss": 0.1498,
+      "step": 3028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571536798001449e-05,
+      "loss": 0.1399,
+      "step": 3029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571249689879974e-05,
+      "loss": 0.1536,
+      "step": 3030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570962511843067e-05,
+      "loss": 0.1374,
+      "step": 3031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570675263925876e-05,
+      "loss": 0.153,
+      "step": 3032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570387946163557e-05,
+      "loss": 0.1482,
+      "step": 3033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701005585912755e-05,
+      "loss": 0.1479,
+      "step": 3034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5698131012442046e-05,
+      "loss": 0.1484,
+      "step": 3035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695255741575264e-05,
+      "loss": 0.1381,
+      "step": 3036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5692379773664314e-05,
+      "loss": 0.15,
+      "step": 3037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5689503109061185e-05,
+      "loss": 0.1372,
+      "step": 3038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5686625748117952e-05,
+      "loss": 0.1414,
+      "step": 3039
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5683747691186778e-05,
+      "loss": 0.1434,
+      "step": 3040
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5680868938619913e-05,
+      "loss": 0.1699,
+      "step": 3041
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677989490769685e-05,
+      "loss": 0.1527,
+      "step": 3042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5675109347988507e-05,
+      "loss": 0.1483,
+      "step": 3043
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5672228510628888e-05,
+      "loss": 0.169,
+      "step": 3044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566934697904341e-05,
+      "loss": 0.1354,
+      "step": 3045
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566646475358474e-05,
+      "loss": 0.1623,
+      "step": 3046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5663581834605644e-05,
+      "loss": 0.1422,
+      "step": 3047
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660698222458962e-05,
+      "loss": 0.14,
+      "step": 3048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565781391749761e-05,
+      "loss": 0.1573,
+      "step": 3049
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5654928920074616e-05,
+      "loss": 0.1434,
+      "step": 3050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565204323054306e-05,
+      "loss": 0.1483,
+      "step": 3051
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564915684925613e-05,
+      "loss": 0.1524,
+      "step": 3052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5646269776567085e-05,
+      "loss": 0.138,
+      "step": 3053
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643382012829287e-05,
+      "loss": 0.1404,
+      "step": 3054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640493558396158e-05,
+      "loss": 0.1585,
+      "step": 3055
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5637604413621226e-05,
+      "loss": 0.1398,
+      "step": 3056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634714578858087e-05,
+      "loss": 0.1479,
+      "step": 3057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563182405446043e-05,
+      "loss": 0.1505,
+      "step": 3058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5628932840782033e-05,
+      "loss": 0.139,
+      "step": 3059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626040938176744e-05,
+      "loss": 0.1484,
+      "step": 3060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5623148346998513e-05,
+      "loss": 0.1441,
+      "step": 3061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562025506760136e-05,
+      "loss": 0.1472,
+      "step": 3062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561736110033939e-05,
+      "loss": 0.146,
+      "step": 3063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5614466445566806e-05,
+      "loss": 0.1535,
+      "step": 3064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561157110363788e-05,
+      "loss": 0.1531,
+      "step": 3065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560867507490698e-05,
+      "loss": 0.1399,
+      "step": 3066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605778359728542e-05,
+      "loss": 0.1558,
+      "step": 3067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56028809584571e-05,
+      "loss": 0.1522,
+      "step": 3068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599982871447273e-05,
+      "loss": 0.1646,
+      "step": 3069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559708409905375e-05,
+      "loss": 0.1427,
+      "step": 3070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559418464163132e-05,
+      "loss": 0.1471,
+      "step": 3071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591284499534845e-05,
+      "loss": 0.137,
+      "step": 3072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5588383673119273e-05,
+      "loss": 0.1319,
+      "step": 3073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585482162739643e-05,
+      "loss": 0.1456,
+      "step": 3074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582579968751068e-05,
+      "loss": 0.139,
+      "step": 3075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579677091508747e-05,
+      "loss": 0.1516,
+      "step": 3076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5576773531367965e-05,
+      "loss": 0.1486,
+      "step": 3077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573869288684087e-05,
+      "loss": 0.1464,
+      "step": 3078
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570964363812572e-05,
+      "loss": 0.1549,
+      "step": 3079
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556805875710895e-05,
+      "loss": 0.1414,
+      "step": 3080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565152468928837e-05,
+      "loss": 0.1227,
+      "step": 3081
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556224549962794e-05,
+      "loss": 0.1369,
+      "step": 3082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559337849562035e-05,
+      "loss": 0.1522,
+      "step": 3083
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556429519086996e-05,
+      "loss": 0.1371,
+      "step": 3084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555352050855878e-05,
+      "loss": 0.1547,
+      "step": 3085
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5550610818333415e-05,
+      "loss": 0.1348,
+      "step": 3086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547700448767017e-05,
+      "loss": 0.1171,
+      "step": 3087
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554478940021579e-05,
+      "loss": 0.1088,
+      "step": 3088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554187767303602e-05,
+      "loss": 0.1189,
+      "step": 3089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553896526758407e-05,
+      "loss": 0.1201,
+      "step": 3090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5536052184216392e-05,
+      "loss": 0.1172,
+      "step": 3091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533138423289526e-05,
+      "loss": 0.1149,
+      "step": 3092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530223985160076e-05,
+      "loss": 0.1161,
+      "step": 3093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5527308870184748e-05,
+      "loss": 0.1137,
+      "step": 3094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524393078720322e-05,
+      "loss": 0.1137,
+      "step": 3095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521476611123665e-05,
+      "loss": 0.1215,
+      "step": 3096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5518559467751724e-05,
+      "loss": 0.1082,
+      "step": 3097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5515641648961526e-05,
+      "loss": 0.1188,
+      "step": 3098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5512723155110185e-05,
+      "loss": 0.0989,
+      "step": 3099
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55098039865549e-05,
+      "loss": 0.1148,
+      "step": 3100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5506884143652944e-05,
+      "loss": 0.1084,
+      "step": 3101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550396362676168e-05,
+      "loss": 0.1255,
+      "step": 3102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550104243623855e-05,
+      "loss": 0.1107,
+      "step": 3103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549812057244108e-05,
+      "loss": 0.1174,
+      "step": 3104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549519803572688e-05,
+      "loss": 0.1167,
+      "step": 3105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5492274826453634e-05,
+      "loss": 0.1207,
+      "step": 3106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489350944979122e-05,
+      "loss": 0.1317,
+      "step": 3107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486426391661193e-05,
+      "loss": 0.1249,
+      "step": 3108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548350116685779e-05,
+      "loss": 0.1196,
+      "step": 3109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5480575270926928e-05,
+      "loss": 0.1104,
+      "step": 3110
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477648704226706e-05,
+      "loss": 0.117,
+      "step": 3111
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5474721467115313e-05,
+      "loss": 0.1124,
+      "step": 3112
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547179355995101e-05,
+      "loss": 0.1027,
+      "step": 3113
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468864983092146e-05,
+      "loss": 0.1106,
+      "step": 3114
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5465935736897152e-05,
+      "loss": 0.1111,
+      "step": 3115
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5463005821724536e-05,
+      "loss": 0.1168,
+      "step": 3116
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460075237932896e-05,
+      "loss": 0.1097,
+      "step": 3117
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5457143985880905e-05,
+      "loss": 0.1102,
+      "step": 3118
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454212065927317e-05,
+      "loss": 0.1173,
+      "step": 3119
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451279478430973e-05,
+      "loss": 0.105,
+      "step": 3120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5448346223750795e-05,
+      "loss": 0.104,
+      "step": 3121
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445412302245783e-05,
+      "loss": 0.1048,
+      "step": 3122
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5442477714275024e-05,
+      "loss": 0.112,
+      "step": 3123
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5439542460197678e-05,
+      "loss": 0.1242,
+      "step": 3124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543660654037299e-05,
+      "loss": 0.1161,
+      "step": 3125
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543366995516029e-05,
+      "loss": 0.1057,
+      "step": 3126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5430732704918998e-05,
+      "loss": 0.1088,
+      "step": 3127
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542779479000859e-05,
+      "loss": 0.1167,
+      "step": 3128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542485621078865e-05,
+      "loss": 0.12,
+      "step": 3129
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542191696761882e-05,
+      "loss": 0.1082,
+      "step": 3130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5418977060858846e-05,
+      "loss": 0.1102,
+      "step": 3131
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541603649086854e-05,
+      "loss": 0.1149,
+      "step": 3132
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5413095258007794e-05,
+      "loss": 0.1212,
+      "step": 3133
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5410153362636596e-05,
+      "loss": 0.1062,
+      "step": 3134
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407210805115e-05,
+      "loss": 0.1252,
+      "step": 3135
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540426758580315e-05,
+      "loss": 0.1121,
+      "step": 3136
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401323705061258e-05,
+      "loss": 0.1172,
+      "step": 3137
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5398379163249636e-05,
+      "loss": 0.1075,
+      "step": 3138
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5395433960728667e-05,
+      "loss": 0.1298,
+      "step": 3139
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539248809785881e-05,
+      "loss": 0.1093,
+      "step": 3140
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5389541575000616e-05,
+      "loss": 0.1187,
+      "step": 3141
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386594392514708e-05,
+      "loss": 0.1133,
+      "step": 3142
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538364655076179e-05,
+      "loss": 0.1149,
+      "step": 3143
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380698050102654e-05,
+      "loss": 0.1176,
+      "step": 3144
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5377748890898166e-05,
+      "loss": 0.1106,
+      "step": 3145
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537479907350927e-05,
+      "loss": 0.1094,
+      "step": 3146
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371848598297008e-05,
+      "loss": 0.1125,
+      "step": 3147
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5368897465622478e-05,
+      "loss": 0.1119,
+      "step": 3148
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365945675846868e-05,
+      "loss": 0.1002,
+      "step": 3149
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536299322933146e-05,
+      "loss": 0.1245,
+      "step": 3150
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5360040126437592e-05,
+      "loss": 0.109,
+      "step": 3151
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357086367526702e-05,
+      "loss": 0.1011,
+      "step": 3152
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5354131952960307e-05,
+      "loss": 0.1063,
+      "step": 3153
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535117688309999e-05,
+      "loss": 0.1167,
+      "step": 3154
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5348221158307426e-05,
+      "loss": 0.1063,
+      "step": 3155
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534526477894437e-05,
+      "loss": 0.1211,
+      "step": 3156
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534230774537265e-05,
+      "loss": 0.0986,
+      "step": 3157
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5339350057954178e-05,
+      "loss": 0.115,
+      "step": 3158
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336391717050947e-05,
+      "loss": 0.1052,
+      "step": 3159
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5333432723025032e-05,
+      "loss": 0.1089,
+      "step": 3160
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533047307623858e-05,
+      "loss": 0.1223,
+      "step": 3161
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532751277705383e-05,
+      "loss": 0.1121,
+      "step": 3162
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5324551825833086e-05,
+      "loss": 0.1237,
+      "step": 3163
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5321590222938748e-05,
+      "loss": 0.1099,
+      "step": 3164
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318627968733286e-05,
+      "loss": 0.1154,
+      "step": 3165
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315665063579246e-05,
+      "loss": 0.1052,
+      "step": 3166
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531270150783926e-05,
+      "loss": 0.1077,
+      "step": 3167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5309737301876045e-05,
+      "loss": 0.1237,
+      "step": 3168
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306772446052382e-05,
+      "loss": 0.1062,
+      "step": 3169
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303806940731146e-05,
+      "loss": 0.1318,
+      "step": 3170
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530084078627529e-05,
+      "loss": 0.1108,
+      "step": 3171
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5297873983047832e-05,
+      "loss": 0.1137,
+      "step": 3172
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5294906531411888e-05,
+      "loss": 0.1094,
+      "step": 3173
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291938431730642e-05,
+      "loss": 0.1075,
+      "step": 3174
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288969684367362e-05,
+      "loss": 0.1094,
+      "step": 3175
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286000289685396e-05,
+      "loss": 0.1062,
+      "step": 3176
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528303024804817e-05,
+      "loss": 0.1216,
+      "step": 3177
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5280059559819177e-05,
+      "loss": 0.0986,
+      "step": 3178
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277088225362017e-05,
+      "loss": 0.1129,
+      "step": 3179
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5274116245040337e-05,
+      "loss": 0.1152,
+      "step": 3180
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527114361921789e-05,
+      "loss": 0.1081,
+      "step": 3181
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268170348258493e-05,
+      "loss": 0.103,
+      "step": 3182
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526519643252604e-05,
+      "loss": 0.1315,
+      "step": 3183
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262221872384523e-05,
+      "loss": 0.1062,
+      "step": 3184
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259246668197987e-05,
+      "loss": 0.1069,
+      "step": 3185
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256270820330572e-05,
+      "loss": 0.1099,
+      "step": 3186
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253294329146494e-05,
+      "loss": 0.1103,
+      "step": 3187
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250317195010044e-05,
+      "loss": 0.1123,
+      "step": 3188
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52473394182856e-05,
+      "loss": 0.1149,
+      "step": 3189
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524436099933761e-05,
+      "loss": 0.1046,
+      "step": 3190
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241381938530597e-05,
+      "loss": 0.1175,
+      "step": 3191
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523840223622918e-05,
+      "loss": 0.1189,
+      "step": 3192
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523542189279804e-05,
+      "loss": 0.1026,
+      "step": 3193
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523244090860194e-05,
+      "loss": 0.1133,
+      "step": 3194
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5229459284005729e-05,
+      "loss": 0.1026,
+      "step": 3195
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5226477019374328e-05,
+      "loss": 0.119,
+      "step": 3196
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223494115072734e-05,
+      "loss": 0.1246,
+      "step": 3197
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.522051057146603e-05,
+      "loss": 0.1278,
+      "step": 3198
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521752638891937e-05,
+      "loss": 0.1146,
+      "step": 3199
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521454156779799e-05,
+      "loss": 0.1101,
+      "step": 3200
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52115561084672e-05,
+      "loss": 0.1149,
+      "step": 3201
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5208570011292398e-05,
+      "loss": 0.1239,
+      "step": 3202
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205583276639044e-05,
+      "loss": 0.1186,
+      "step": 3203
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202595904872698e-05,
+      "loss": 0.1175,
+      "step": 3204
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5199607896358969e-05,
+      "loss": 0.1097,
+      "step": 3205
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196619251463575e-05,
+      "loss": 0.1099,
+      "step": 3206
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5193629970552293e-05,
+      "loss": 0.1119,
+      "step": 3207
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190640053990982e-05,
+      "loss": 0.1277,
+      "step": 3208
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187649502145574e-05,
+      "loss": 0.1097,
+      "step": 3209
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.518465831538209e-05,
+      "loss": 0.1138,
+      "step": 3210
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181666494066618e-05,
+      "loss": 0.1087,
+      "step": 3211
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178674038565329e-05,
+      "loss": 0.114,
+      "step": 3212
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175680949244472e-05,
+      "loss": 0.1133,
+      "step": 3213
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5172687226470371e-05,
+      "loss": 0.1076,
+      "step": 3214
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516969287060943e-05,
+      "loss": 0.1044,
+      "step": 3215
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166697882028127e-05,
+      "loss": 0.1105,
+      "step": 3216
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516370226109302e-05,
+      "loss": 0.1157,
+      "step": 3217
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160706008170744e-05,
+      "loss": 0.1063,
+      "step": 3218
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157709123628016e-05,
+      "loss": 0.113,
+      "step": 3219
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5154711607831618e-05,
+      "loss": 0.1114,
+      "step": 3220
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151713461148424e-05,
+      "loss": 0.1099,
+      "step": 3221
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148714683945374e-05,
+      "loss": 0.1111,
+      "step": 3222
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5145715276589487e-05,
+      "loss": 0.1116,
+      "step": 3223
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142715239447867e-05,
+      "loss": 0.1172,
+      "step": 3224
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5139714572887686e-05,
+      "loss": 0.1063,
+      "step": 3225
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.51367132772762e-05,
+      "loss": 0.1165,
+      "step": 3226
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5133711352980736e-05,
+      "loss": 0.1121,
+      "step": 3227
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130708800368697e-05,
+      "loss": 0.1191,
+      "step": 3228
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127705619807572e-05,
+      "loss": 0.1231,
+      "step": 3229
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124701811664918e-05,
+      "loss": 0.1079,
+      "step": 3230
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5121697376308374e-05,
+      "loss": 0.1299,
+      "step": 3231
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118692314105656e-05,
+      "loss": 0.1201,
+      "step": 3232
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115686625424549e-05,
+      "loss": 0.1056,
+      "step": 3233
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5112680310632923e-05,
+      "loss": 0.1214,
+      "step": 3234
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510967337009872e-05,
+      "loss": 0.1092,
+      "step": 3235
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106665804189961e-05,
+      "loss": 0.1201,
+      "step": 3236
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5103657613274747e-05,
+      "loss": 0.1233,
+      "step": 3237
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510064879772125e-05,
+      "loss": 0.1235,
+      "step": 3238
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097639357897715e-05,
+      "loss": 0.1236,
+      "step": 3239
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509462929417247e-05,
+      "loss": 0.1225,
+      "step": 3240
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091618606913922e-05,
+      "loss": 0.1053,
+      "step": 3241
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508860729649055e-05,
+      "loss": 0.1013,
+      "step": 3242
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50855953632709e-05,
+      "loss": 0.1212,
+      "step": 3243
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082582807623616e-05,
+      "loss": 0.1107,
+      "step": 3244
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079569629917394e-05,
+      "loss": 0.1103,
+      "step": 3245
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5076555830521027e-05,
+      "loss": 0.1082,
+      "step": 3246
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073541409803369e-05,
+      "loss": 0.1063,
+      "step": 3247
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070526368133358e-05,
+      "loss": 0.1149,
+      "step": 3248
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5067510705880009e-05,
+      "loss": 0.1218,
+      "step": 3249
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064494423412404e-05,
+      "loss": 0.1186,
+      "step": 3250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061477521099713e-05,
+      "loss": 0.1216,
+      "step": 3251
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505845999931117e-05,
+      "loss": 0.1121,
+      "step": 3252
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5055441858416093e-05,
+      "loss": 0.111,
+      "step": 3253
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052423098783872e-05,
+      "loss": 0.0991,
+      "step": 3254
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049403720783974e-05,
+      "loss": 0.1039,
+      "step": 3255
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5046383724785946e-05,
+      "loss": 0.1272,
+      "step": 3256
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5043363111159404e-05,
+      "loss": 0.1299,
+      "step": 3257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040341880274038e-05,
+      "loss": 0.1166,
+      "step": 3258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.503732003249962e-05,
+      "loss": 0.1182,
+      "step": 3259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034297568205994e-05,
+      "loss": 0.1187,
+      "step": 3260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031274487763084e-05,
+      "loss": 0.109,
+      "step": 3261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028250791540881e-05,
+      "loss": 0.1146,
+      "step": 3262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025226479909458e-05,
+      "loss": 0.1157,
+      "step": 3263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5022201553238966e-05,
+      "loss": 0.1238,
+      "step": 3264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5019176011899621e-05,
+      "loss": 0.103,
+      "step": 3265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501614985626172e-05,
+      "loss": 0.122,
+      "step": 3266
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5013123086695633e-05,
+      "loss": 0.1164,
+      "step": 3267
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010095703571816e-05,
+      "loss": 0.1245,
+      "step": 3268
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5007067707260787e-05,
+      "loss": 0.103,
+      "step": 3269
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500403909813314e-05,
+      "loss": 0.1079,
+      "step": 3270
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500100987655955e-05,
+      "loss": 0.1139,
+      "step": 3271
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997980042910767e-05,
+      "loss": 0.1161,
+      "step": 3272
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499494959755761e-05,
+      "loss": 0.1166,
+      "step": 3273
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4991918540870978e-05,
+      "loss": 0.124,
+      "step": 3274
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988886873221842e-05,
+      "loss": 0.1157,
+      "step": 3275
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4985854594981249e-05,
+      "loss": 0.1065,
+      "step": 3276
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498282170652032e-05,
+      "loss": 0.1111,
+      "step": 3277
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979788208210249e-05,
+      "loss": 0.1112,
+      "step": 3278
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497675410042231e-05,
+      "loss": 0.1124,
+      "step": 3279
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973719383527849e-05,
+      "loss": 0.1055,
+      "step": 3280
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970684057898287e-05,
+      "loss": 0.1232,
+      "step": 3281
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4967648123905115e-05,
+      "loss": 0.112,
+      "step": 3282
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964611581919901e-05,
+      "loss": 0.1059,
+      "step": 3283
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961574432314291e-05,
+      "loss": 0.1133,
+      "step": 3284
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4958536675460004e-05,
+      "loss": 0.1184,
+      "step": 3285
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955498311728832e-05,
+      "loss": 0.1075,
+      "step": 3286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4952459341492636e-05,
+      "loss": 0.1081,
+      "step": 3287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494941976512336e-05,
+      "loss": 0.1142,
+      "step": 3288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494637958299302e-05,
+      "loss": 0.1048,
+      "step": 3289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4943338795473703e-05,
+      "loss": 0.1225,
+      "step": 3290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940297402937573e-05,
+      "loss": 0.1181,
+      "step": 3291
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937255405756868e-05,
+      "loss": 0.1149,
+      "step": 3292
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4934212804303899e-05,
+      "loss": 0.1153,
+      "step": 3293
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931169598951045e-05,
+      "loss": 0.1208,
+      "step": 3294
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928125790070773e-05,
+      "loss": 0.114,
+      "step": 3295
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4925081378035612e-05,
+      "loss": 0.1081,
+      "step": 3296
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922036363218169e-05,
+      "loss": 0.1203,
+      "step": 3297
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918990745991122e-05,
+      "loss": 0.1057,
+      "step": 3298
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915944526727231e-05,
+      "loss": 0.1123,
+      "step": 3299
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4912897705799317e-05,
+      "loss": 0.1284,
+      "step": 3300
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909850283580282e-05,
+      "loss": 0.1196,
+      "step": 3301
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906802260443105e-05,
+      "loss": 0.1061,
+      "step": 3302
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4903753636760834e-05,
+      "loss": 0.1138,
+      "step": 3303
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490070441290659e-05,
+      "loss": 0.1199,
+      "step": 3304
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489765458925357e-05,
+      "loss": 0.1214,
+      "step": 3305
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4894604166175039e-05,
+      "loss": 0.1255,
+      "step": 3306
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891553144044342e-05,
+      "loss": 0.1042,
+      "step": 3307
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888501523234894e-05,
+      "loss": 0.1027,
+      "step": 3308
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4885449304120184e-05,
+      "loss": 0.1191,
+      "step": 3309
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882396487073773e-05,
+      "loss": 0.1231,
+      "step": 3310
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4879343072469299e-05,
+      "loss": 0.1151,
+      "step": 3311
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876289060680466e-05,
+      "loss": 0.1155,
+      "step": 3312
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487323445208106e-05,
+      "loss": 0.111,
+      "step": 3313
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4870179247044933e-05,
+      "loss": 0.1074,
+      "step": 3314
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867123445946013e-05,
+      "loss": 0.1115,
+      "step": 3315
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4864067049158304e-05,
+      "loss": 0.1268,
+      "step": 3316
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4861010057055873e-05,
+      "loss": 0.1165,
+      "step": 3317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4857952470012871e-05,
+      "loss": 0.1287,
+      "step": 3318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854894288403512e-05,
+      "loss": 0.1323,
+      "step": 3319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4851835512602093e-05,
+      "loss": 0.1291,
+      "step": 3320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848776142982975e-05,
+      "loss": 0.111,
+      "step": 3321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845716179920597e-05,
+      "loss": 0.1158,
+      "step": 3322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.484265562378947e-05,
+      "loss": 0.1169,
+      "step": 3323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839594474964179e-05,
+      "loss": 0.1072,
+      "step": 3324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483653273381937e-05,
+      "loss": 0.1143,
+      "step": 3325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833470400729777e-05,
+      "loss": 0.1233,
+      "step": 3326
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48304074760702e-05,
+      "loss": 0.123,
+      "step": 3327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827343960215508e-05,
+      "loss": 0.117,
+      "step": 3328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4824279853540646e-05,
+      "loss": 0.1118,
+      "step": 3329
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821215156420638e-05,
+      "loss": 0.1202,
+      "step": 3330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4818149869230564e-05,
+      "loss": 0.1177,
+      "step": 3331
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481508399234559e-05,
+      "loss": 0.1249,
+      "step": 3332
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4812017526140947e-05,
+      "loss": 0.1208,
+      "step": 3333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808950470991945e-05,
+      "loss": 0.1256,
+      "step": 3334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805882827273958e-05,
+      "loss": 0.1111,
+      "step": 3335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480281459536244e-05,
+      "loss": 0.1257,
+      "step": 3336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479974577563291e-05,
+      "loss": 0.1176,
+      "step": 3337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4796676368460963e-05,
+      "loss": 0.115,
+      "step": 3338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479360637422226e-05,
+      "loss": 0.1259,
+      "step": 3339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4790535793292545e-05,
+      "loss": 0.1041,
+      "step": 3340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4787464626047625e-05,
+      "loss": 0.1086,
+      "step": 3341
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784392872863383e-05,
+      "loss": 0.1263,
+      "step": 3342
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781320534115766e-05,
+      "loss": 0.1088,
+      "step": 3343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4778247610180803e-05,
+      "loss": 0.1086,
+      "step": 3344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477517410143459e-05,
+      "loss": 0.1168,
+      "step": 3345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4772100008253295e-05,
+      "loss": 0.1153,
+      "step": 3346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4769025331013156e-05,
+      "loss": 0.124,
+      "step": 3347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4765950070090484e-05,
+      "loss": 0.1084,
+      "step": 3348
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762874225861663e-05,
+      "loss": 0.1109,
+      "step": 3349
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759797798703144e-05,
+      "loss": 0.1208,
+      "step": 3350
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4756720788991451e-05,
+      "loss": 0.1092,
+      "step": 3351
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753643197103184e-05,
+      "loss": 0.1223,
+      "step": 3352
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475056502341501e-05,
+      "loss": 0.1145,
+      "step": 3353
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747486268303663e-05,
+      "loss": 0.1189,
+      "step": 3354
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744406932145959e-05,
+      "loss": 0.1231,
+      "step": 3355
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741327015318772e-05,
+      "loss": 0.1177,
+      "step": 3356
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473824651819906e-05,
+      "loss": 0.1169,
+      "step": 3357
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4735165441163846e-05,
+      "loss": 0.1062,
+      "step": 3358
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732083784590221e-05,
+      "loss": 0.1145,
+      "step": 3359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.472900154885535e-05,
+      "loss": 0.1194,
+      "step": 3360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725918734336469e-05,
+      "loss": 0.1279,
+      "step": 3361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722835341410886e-05,
+      "loss": 0.1107,
+      "step": 3362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4719751370455977e-05,
+      "loss": 0.1073,
+      "step": 3363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716666821849192e-05,
+      "loss": 0.1091,
+      "step": 3364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4713581695968047e-05,
+      "loss": 0.1229,
+      "step": 3365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710495993190136e-05,
+      "loss": 0.1093,
+      "step": 3366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707409713893117e-05,
+      "loss": 0.1044,
+      "step": 3367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4704322858454718e-05,
+      "loss": 0.1197,
+      "step": 3368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701235427252744e-05,
+      "loss": 0.1092,
+      "step": 3369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4698147420665065e-05,
+      "loss": 0.1119,
+      "step": 3370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695058839069622e-05,
+      "loss": 0.1126,
+      "step": 3371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4691969682844431e-05,
+      "loss": 0.1159,
+      "step": 3372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688879952367571e-05,
+      "loss": 0.1163,
+      "step": 3373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4685789648017197e-05,
+      "loss": 0.1155,
+      "step": 3374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4682698770171533e-05,
+      "loss": 0.1048,
+      "step": 3375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4679607319208869e-05,
+      "loss": 0.1096,
+      "step": 3376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4676515295507571e-05,
+      "loss": 0.117,
+      "step": 3377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673422699446078e-05,
+      "loss": 0.123,
+      "step": 3378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4670329531402884e-05,
+      "loss": 0.1122,
+      "step": 3379
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4667235791756573e-05,
+      "loss": 0.1154,
+      "step": 3380
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466414148088578e-05,
+      "loss": 0.1153,
+      "step": 3381
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.466104659916922e-05,
+      "loss": 0.1131,
+      "step": 3382
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465795114698568e-05,
+      "loss": 0.1142,
+      "step": 3383
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4654855124714015e-05,
+      "loss": 0.1128,
+      "step": 3384
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651758532733142e-05,
+      "loss": 0.1253,
+      "step": 3385
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4648661371422058e-05,
+      "loss": 0.1112,
+      "step": 3386
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645563641159824e-05,
+      "loss": 0.1301,
+      "step": 3387
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642465342325574e-05,
+      "loss": 0.107,
+      "step": 3388
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4639366475298506e-05,
+      "loss": 0.1179,
+      "step": 3389
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636267040457899e-05,
+      "loss": 0.1189,
+      "step": 3390
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633167038183087e-05,
+      "loss": 0.1217,
+      "step": 3391
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630066468853481e-05,
+      "loss": 0.1128,
+      "step": 3392
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626965332848561e-05,
+      "loss": 0.1323,
+      "step": 3393
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4623863630547876e-05,
+      "loss": 0.1104,
+      "step": 3394
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620761362331046e-05,
+      "loss": 0.1263,
+      "step": 3395
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617658528577759e-05,
+      "loss": 0.1167,
+      "step": 3396
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4614555129667772e-05,
+      "loss": 0.1111,
+      "step": 3397
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4611451165980905e-05,
+      "loss": 0.1189,
+      "step": 3398
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460834663789706e-05,
+      "loss": 0.1254,
+      "step": 3399
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4605241545796199e-05,
+      "loss": 0.1074,
+      "step": 3400
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602135890058356e-05,
+      "loss": 0.1071,
+      "step": 3401
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.459902967106363e-05,
+      "loss": 0.1101,
+      "step": 3402
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595922889192195e-05,
+      "loss": 0.1195,
+      "step": 3403
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592815544824292e-05,
+      "loss": 0.1196,
+      "step": 3404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589707638340227e-05,
+      "loss": 0.1088,
+      "step": 3405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586599170120376e-05,
+      "loss": 0.1154,
+      "step": 3406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4583490140545192e-05,
+      "loss": 0.1008,
+      "step": 3407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4580380549995183e-05,
+      "loss": 0.1064,
+      "step": 3408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4577270398850942e-05,
+      "loss": 0.1133,
+      "step": 3409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.457415968749311e-05,
+      "loss": 0.1188,
+      "step": 3410
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571048416302416e-05,
+      "loss": 0.1099,
+      "step": 3411
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567936585659647e-05,
+      "loss": 0.1057,
+      "step": 3412
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4564824195945662e-05,
+      "loss": 0.1117,
+      "step": 3413
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4561711247541384e-05,
+      "loss": 0.1204,
+      "step": 3414
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4558597740827813e-05,
+      "loss": 0.1161,
+      "step": 3415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555483676186007e-05,
+      "loss": 0.1334,
+      "step": 3416
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552369053997097e-05,
+      "loss": 0.1058,
+      "step": 3417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4549253874642289e-05,
+      "loss": 0.1244,
+      "step": 3418
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546138138502845e-05,
+      "loss": 0.1102,
+      "step": 3419
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4543021845960104e-05,
+      "loss": 0.1109,
+      "step": 3420
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.1306,
+      "step": 3421
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4536787593190411e-05,
+      "loss": 0.1222,
+      "step": 3422
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533669633726474e-05,
+      "loss": 0.1248,
+      "step": 3423
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453055111938526e-05,
+      "loss": 0.1132,
+      "step": 3424
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452743205054845e-05,
+      "loss": 0.1156,
+      "step": 3425
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524312427597784e-05,
+      "loss": 0.1046,
+      "step": 3426
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4521192250915076e-05,
+      "loss": 0.1258,
+      "step": 3427
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4518071520882206e-05,
+      "loss": 0.1123,
+      "step": 3428
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451495023788112e-05,
+      "loss": 0.1117,
+      "step": 3429
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511828402293833e-05,
+      "loss": 0.12,
+      "step": 3430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4508706014502426e-05,
+      "loss": 0.1327,
+      "step": 3431
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4505583074889053e-05,
+      "loss": 0.1238,
+      "step": 3432
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4502459583835928e-05,
+      "loss": 0.1104,
+      "step": 3433
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499335541725337e-05,
+      "loss": 0.1149,
+      "step": 3434
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496210948939633e-05,
+      "loss": 0.1061,
+      "step": 3435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4493085805861233e-05,
+      "loss": 0.1294,
+      "step": 3436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448996011287263e-05,
+      "loss": 0.115,
+      "step": 3437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486833870356374e-05,
+      "loss": 0.1131,
+      "step": 3438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448370707869509e-05,
+      "loss": 0.1247,
+      "step": 3439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4480579738271463e-05,
+      "loss": 0.1112,
+      "step": 3440
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447745184946825e-05,
+      "loss": 0.1239,
+      "step": 3441
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474323412668278e-05,
+      "loss": 0.1164,
+      "step": 3442
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471194428254435e-05,
+      "loss": 0.1075,
+      "step": 3443
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446806489660968e-05,
+      "loss": 0.1242,
+      "step": 3444
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4464934818117035e-05,
+      "loss": 0.1093,
+      "step": 3445
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461804193159589e-05,
+      "loss": 0.1146,
+      "step": 3446
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458673022120506e-05,
+      "loss": 0.1282,
+      "step": 3447
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4455541305383007e-05,
+      "loss": 0.1194,
+      "step": 3448
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.445240904333039e-05,
+      "loss": 0.1156,
+      "step": 3449
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449276236346e-05,
+      "loss": 0.1227,
+      "step": 3450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446142884813278e-05,
+      "loss": 0.1183,
+      "step": 3451
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443008989115708e-05,
+      "loss": 0.1149,
+      "step": 3452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439874549636848e-05,
+      "loss": 0.1106,
+      "step": 3453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4436739566760326e-05,
+      "loss": 0.1227,
+      "step": 3454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443360404086983e-05,
+      "loss": 0.112,
+      "step": 3455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4430467972349122e-05,
+      "loss": 0.1119,
+      "step": 3456
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4427331361582024e-05,
+      "loss": 0.1225,
+      "step": 3457
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4424194208952427e-05,
+      "loss": 0.1209,
+      "step": 3458
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4421056514844287e-05,
+      "loss": 0.129,
+      "step": 3459
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441791827964163e-05,
+      "loss": 0.1143,
+      "step": 3460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4414779503728543e-05,
+      "loss": 0.1145,
+      "step": 3461
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411640187489181e-05,
+      "loss": 0.1219,
+      "step": 3462
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4408500331307775e-05,
+      "loss": 0.1114,
+      "step": 3463
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44053599355686e-05,
+      "loss": 0.1105,
+      "step": 3464
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440221900065602e-05,
+      "loss": 0.1169,
+      "step": 3465
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4399077526954451e-05,
+      "loss": 0.1104,
+      "step": 3466
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4395935514848377e-05,
+      "loss": 0.1204,
+      "step": 3467
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392792964722351e-05,
+      "loss": 0.1334,
+      "step": 3468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389649876960992e-05,
+      "loss": 0.1141,
+      "step": 3469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4386506251948986e-05,
+      "loss": 0.117,
+      "step": 3470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383362090071076e-05,
+      "loss": 0.1164,
+      "step": 3471
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4380217391712082e-05,
+      "loss": 0.1137,
+      "step": 3472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377072157256881e-05,
+      "loss": 0.1245,
+      "step": 3473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373926387090424e-05,
+      "loss": 0.1113,
+      "step": 3474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370780081597721e-05,
+      "loss": 0.1229,
+      "step": 3475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4367633241163846e-05,
+      "loss": 0.1115,
+      "step": 3476
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364485866173947e-05,
+      "loss": 0.1033,
+      "step": 3477
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361337957013227e-05,
+      "loss": 0.1105,
+      "step": 3478
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358189514066963e-05,
+      "loss": 0.1069,
+      "step": 3479
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4355040537720491e-05,
+      "loss": 0.105,
+      "step": 3480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.435189102835922e-05,
+      "loss": 0.1118,
+      "step": 3481
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348740986368617e-05,
+      "loss": 0.1146,
+      "step": 3482
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345590412134214e-05,
+      "loss": 0.1102,
+      "step": 3483
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4342439306041613e-05,
+      "loss": 0.1178,
+      "step": 3484
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433928766847648e-05,
+      "loss": 0.1113,
+      "step": 3485
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336135499824546e-05,
+      "loss": 0.114,
+      "step": 3486
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43329828004716e-05,
+      "loss": 0.1039,
+      "step": 3487
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329829570803511e-05,
+      "loss": 0.1172,
+      "step": 3488
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326675811206194e-05,
+      "loss": 0.1172,
+      "step": 3489
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4323521522065643e-05,
+      "loss": 0.1107,
+      "step": 3490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320366703767913e-05,
+      "loss": 0.1283,
+      "step": 3491
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317211356699124e-05,
+      "loss": 0.1162,
+      "step": 3492
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.431405548124546e-05,
+      "loss": 0.108,
+      "step": 3493
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4310899077793167e-05,
+      "loss": 0.103,
+      "step": 3494
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430774214672856e-05,
+      "loss": 0.1157,
+      "step": 3495
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304584688438015e-05,
+      "loss": 0.1195,
+      "step": 3496
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430142670330798e-05,
+      "loss": 0.1226,
+      "step": 3497
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4298268191724951e-05,
+      "loss": 0.102,
+      "step": 3498
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295109154075511e-05,
+      "loss": 0.1224,
+      "step": 3499
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429194959074629e-05,
+      "loss": 0.1155,
+      "step": 3500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288789502123985e-05,
+      "loss": 0.1233,
+      "step": 3501
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4285628888595366e-05,
+      "loss": 0.1166,
+      "step": 3502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282467750547257e-05,
+      "loss": 0.1093,
+      "step": 3503
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4279306088366555e-05,
+      "loss": 0.1138,
+      "step": 3504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276143902440213e-05,
+      "loss": 0.1159,
+      "step": 3505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272981193155257e-05,
+      "loss": 0.1178,
+      "step": 3506
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4269817960898766e-05,
+      "loss": 0.1198,
+      "step": 3507
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266654206057892e-05,
+      "loss": 0.1278,
+      "step": 3508
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426348992901985e-05,
+      "loss": 0.1116,
+      "step": 3509
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426032513017191e-05,
+      "loss": 0.1274,
+      "step": 3510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257159809901419e-05,
+      "loss": 0.1088,
+      "step": 3511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253993968595782e-05,
+      "loss": 0.1124,
+      "step": 3512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250827606642464e-05,
+      "loss": 0.1173,
+      "step": 3513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4247660724428995e-05,
+      "loss": 0.1232,
+      "step": 3514
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244493322342975e-05,
+      "loss": 0.1168,
+      "step": 3515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241325400772064e-05,
+      "loss": 0.1144,
+      "step": 3516
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238156960103985e-05,
+      "loss": 0.1177,
+      "step": 3517
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423498800072652e-05,
+      "loss": 0.1169,
+      "step": 3518
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4231818523027523e-05,
+      "loss": 0.1173,
+      "step": 3519
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228648527394905e-05,
+      "loss": 0.1168,
+      "step": 3520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4225478014216643e-05,
+      "loss": 0.1094,
+      "step": 3521
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4222306983880779e-05,
+      "loss": 0.1145,
+      "step": 3522
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219135436775412e-05,
+      "loss": 0.1024,
+      "step": 3523
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215963373288717e-05,
+      "loss": 0.1063,
+      "step": 3524
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4212790793808915e-05,
+      "loss": 0.1276,
+      "step": 3525
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4209617698724305e-05,
+      "loss": 0.1151,
+      "step": 3526
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206444088423239e-05,
+      "loss": 0.1122,
+      "step": 3527
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420326996329414e-05,
+      "loss": 0.1207,
+      "step": 3528
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200095323725487e-05,
+      "loss": 0.1019,
+      "step": 3529
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419692017010583e-05,
+      "loss": 0.1151,
+      "step": 3530
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419374450282377e-05,
+      "loss": 0.1109,
+      "step": 3531
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4190568322267981e-05,
+      "loss": 0.1179,
+      "step": 3532
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4187391628827198e-05,
+      "loss": 0.1168,
+      "step": 3533
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4184214422890217e-05,
+      "loss": 0.1337,
+      "step": 3534
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181036704845896e-05,
+      "loss": 0.1112,
+      "step": 3535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177858475083159e-05,
+      "loss": 0.1106,
+      "step": 3536
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417467973399099e-05,
+      "loss": 0.1023,
+      "step": 3537
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4171500481958433e-05,
+      "loss": 0.1166,
+      "step": 3538
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168320719374604e-05,
+      "loss": 0.1224,
+      "step": 3539
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416514044662867e-05,
+      "loss": 0.1206,
+      "step": 3540
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161959664109866e-05,
+      "loss": 0.1169,
+      "step": 3541
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158778372207489e-05,
+      "loss": 0.1172,
+      "step": 3542
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4155596571310903e-05,
+      "loss": 0.1069,
+      "step": 3543
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152414261809524e-05,
+      "loss": 0.111,
+      "step": 3544
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149231444092837e-05,
+      "loss": 0.1171,
+      "step": 3545
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4146048118550388e-05,
+      "loss": 0.1243,
+      "step": 3546
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414286428557179e-05,
+      "loss": 0.1099,
+      "step": 3547
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139679945546705e-05,
+      "loss": 0.1187,
+      "step": 3548
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4136495098864873e-05,
+      "loss": 0.124,
+      "step": 3549
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4133309745916084e-05,
+      "loss": 0.1152,
+      "step": 3550
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41301238870902e-05,
+      "loss": 0.1179,
+      "step": 3551
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126937522777129e-05,
+      "loss": 0.1198,
+      "step": 3552
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123750653366858e-05,
+      "loss": 0.1149,
+      "step": 3553
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120563279249428e-05,
+      "loss": 0.1034,
+      "step": 3554
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4117375400814945e-05,
+      "loss": 0.1116,
+      "step": 3555
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411418701845357e-05,
+      "loss": 0.1208,
+      "step": 3556
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110998132555534e-05,
+      "loss": 0.1024,
+      "step": 3557
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107808743511124e-05,
+      "loss": 0.106,
+      "step": 3558
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4104618851710688e-05,
+      "loss": 0.1205,
+      "step": 3559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101428457544643e-05,
+      "loss": 0.1204,
+      "step": 3560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409823756140346e-05,
+      "loss": 0.1166,
+      "step": 3561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095046163677673e-05,
+      "loss": 0.1171,
+      "step": 3562
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091854264757878e-05,
+      "loss": 0.1186,
+      "step": 3563
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4088661865034735e-05,
+      "loss": 0.1051,
+      "step": 3564
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085468964898956e-05,
+      "loss": 0.1221,
+      "step": 3565
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408227556474133e-05,
+      "loss": 0.1209,
+      "step": 3566
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4079081664952697e-05,
+      "loss": 0.1165,
+      "step": 3567
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4075887265923954e-05,
+      "loss": 0.1145,
+      "step": 3568
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4072692368046066e-05,
+      "loss": 0.0943,
+      "step": 3569
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4069496971710057e-05,
+      "loss": 0.1276,
+      "step": 3570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.406630107730702e-05,
+      "loss": 0.108,
+      "step": 3571
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4063104685228092e-05,
+      "loss": 0.116,
+      "step": 3572
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4059907795864486e-05,
+      "loss": 0.1048,
+      "step": 3573
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405671040960747e-05,
+      "loss": 0.12,
+      "step": 3574
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4053512526848369e-05,
+      "loss": 0.1066,
+      "step": 3575
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4050314147978577e-05,
+      "loss": 0.1128,
+      "step": 3576
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047115273389542e-05,
+      "loss": 0.1193,
+      "step": 3577
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043915903472777e-05,
+      "loss": 0.1159,
+      "step": 3578
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040716038619857e-05,
+      "loss": 0.1112,
+      "step": 3579
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4037515679222408e-05,
+      "loss": 0.1162,
+      "step": 3580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4034314825672127e-05,
+      "loss": 0.1295,
+      "step": 3581
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031113478360767e-05,
+      "loss": 0.1114,
+      "step": 3582
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402791163768014e-05,
+      "loss": 0.1238,
+      "step": 3583
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4024709304022123e-05,
+      "loss": 0.13,
+      "step": 3584
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402150647777865e-05,
+      "loss": 0.1225,
+      "step": 3585
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018303159341713e-05,
+      "loss": 0.1144,
+      "step": 3586
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015099349103372e-05,
+      "loss": 0.0969,
+      "step": 3587
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4011895047455738e-05,
+      "loss": 0.1135,
+      "step": 3588
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008690254790986e-05,
+      "loss": 0.1223,
+      "step": 3589
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005484971501357e-05,
+      "loss": 0.1207,
+      "step": 3590
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4002279197979144e-05,
+      "loss": 0.1306,
+      "step": 3591
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999072934616702e-05,
+      "loss": 0.1243,
+      "step": 3592
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995866181806448e-05,
+      "loss": 0.1227,
+      "step": 3593
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3992658939940855e-05,
+      "loss": 0.1206,
+      "step": 3594
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3989451209412458e-05,
+      "loss": 0.0974,
+      "step": 3595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986242990613856e-05,
+      "loss": 0.1153,
+      "step": 3596
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3983034283937702e-05,
+      "loss": 0.1184,
+      "step": 3597
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3979825089776708e-05,
+      "loss": 0.1265,
+      "step": 3598
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976615408523652e-05,
+      "loss": 0.108,
+      "step": 3599
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3973405240571366e-05,
+      "loss": 0.1181,
+      "step": 3600
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970194586312746e-05,
+      "loss": 0.1182,
+      "step": 3601
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966983446140741e-05,
+      "loss": 0.1203,
+      "step": 3602
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963771820448368e-05,
+      "loss": 0.1122,
+      "step": 3603
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3960559709628697e-05,
+      "loss": 0.1188,
+      "step": 3604
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3957347114074855e-05,
+      "loss": 0.1076,
+      "step": 3605
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3954134034180038e-05,
+      "loss": 0.1178,
+      "step": 3606
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3950920470337494e-05,
+      "loss": 0.1219,
+      "step": 3607
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947706422940532e-05,
+      "loss": 0.1083,
+      "step": 3608
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3944491892382524e-05,
+      "loss": 0.1185,
+      "step": 3609
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394127687905689e-05,
+      "loss": 0.1087,
+      "step": 3610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393806138335712e-05,
+      "loss": 0.1101,
+      "step": 3611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934845405676763e-05,
+      "loss": 0.1295,
+      "step": 3612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931628946409416e-05,
+      "loss": 0.1238,
+      "step": 3613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3928412005948755e-05,
+      "loss": 0.1082,
+      "step": 3614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925194584688489e-05,
+      "loss": 0.1155,
+      "step": 3615
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3921976683022407e-05,
+      "loss": 0.11,
+      "step": 3616
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918758301344344e-05,
+      "loss": 0.1076,
+      "step": 3617
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3915539440048203e-05,
+      "loss": 0.1114,
+      "step": 3618
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912320099527939e-05,
+      "loss": 0.1037,
+      "step": 3619
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3909100280177569e-05,
+      "loss": 0.1103,
+      "step": 3620
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905879982391168e-05,
+      "loss": 0.105,
+      "step": 3621
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390265920656287e-05,
+      "loss": 0.1119,
+      "step": 3622
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3899437953086865e-05,
+      "loss": 0.1063,
+      "step": 3623
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896216222357404e-05,
+      "loss": 0.1173,
+      "step": 3624
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3892994014768796e-05,
+      "loss": 0.122,
+      "step": 3625
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889771330715408e-05,
+      "loss": 0.1221,
+      "step": 3626
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886548170591664e-05,
+      "loss": 0.121,
+      "step": 3627
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883324534792051e-05,
+      "loss": 0.1178,
+      "step": 3628
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880100423711108e-05,
+      "loss": 0.1187,
+      "step": 3629
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876875837743431e-05,
+      "loss": 0.1173,
+      "step": 3630
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873650777283684e-05,
+      "loss": 0.1134,
+      "step": 3631
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870425242726583e-05,
+      "loss": 0.1102,
+      "step": 3632
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867199234466901e-05,
+      "loss": 0.1177,
+      "step": 3633
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863972752899472e-05,
+      "loss": 0.1103,
+      "step": 3634
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860745798419181e-05,
+      "loss": 0.1092,
+      "step": 3635
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3857518371420984e-05,
+      "loss": 0.1148,
+      "step": 3636
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854290472299879e-05,
+      "loss": 0.1176,
+      "step": 3637
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385106210145093e-05,
+      "loss": 0.1175,
+      "step": 3638
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3847833259269264e-05,
+      "loss": 0.1151,
+      "step": 3639
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844603946150057e-05,
+      "loss": 0.1158,
+      "step": 3640
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841374162488543e-05,
+      "loss": 0.1078,
+      "step": 3641
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3838143908680022e-05,
+      "loss": 0.1072,
+      "step": 3642
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834913185119841e-05,
+      "loss": 0.1172,
+      "step": 3643
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383168199220341e-05,
+      "loss": 0.1197,
+      "step": 3644
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38284503303262e-05,
+      "loss": 0.1084,
+      "step": 3645
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382521819988373e-05,
+      "loss": 0.1151,
+      "step": 3646
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821985601271584e-05,
+      "loss": 0.1199,
+      "step": 3647
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38187525348854e-05,
+      "loss": 0.1057,
+      "step": 3648
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815519001120872e-05,
+      "loss": 0.1057,
+      "step": 3649
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3812285000373757e-05,
+      "loss": 0.102,
+      "step": 3650
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809050533039863e-05,
+      "loss": 0.1201,
+      "step": 3651
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.380581559951506e-05,
+      "loss": 0.1169,
+      "step": 3652
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3802580200195267e-05,
+      "loss": 0.1153,
+      "step": 3653
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799344335476473e-05,
+      "loss": 0.1148,
+      "step": 3654
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796108005754712e-05,
+      "loss": 0.1149,
+      "step": 3655
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379287121142608e-05,
+      "loss": 0.1183,
+      "step": 3656
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378963395288673e-05,
+      "loss": 0.1121,
+      "step": 3657
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786396230532871e-05,
+      "loss": 0.1146,
+      "step": 3658
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3783158044760767e-05,
+      "loss": 0.1172,
+      "step": 3659
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3779919395966743e-05,
+      "loss": 0.1155,
+      "step": 3660
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776680284547176e-05,
+      "loss": 0.1122,
+      "step": 3661
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773440710898505e-05,
+      "loss": 0.1112,
+      "step": 3662
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.377020067541722e-05,
+      "loss": 0.1279,
+      "step": 3663
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766960178499874e-05,
+      "loss": 0.1174,
+      "step": 3664
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3763719220543063e-05,
+      "loss": 0.114,
+      "step": 3665
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376047780194346e-05,
+      "loss": 0.1213,
+      "step": 3666
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757235923097777e-05,
+      "loss": 0.1177,
+      "step": 3667
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753993584402787e-05,
+      "loss": 0.106,
+      "step": 3668
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750750786255327e-05,
+      "loss": 0.1025,
+      "step": 3669
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3747507529052282e-05,
+      "loss": 0.115,
+      "step": 3670
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374426381319059e-05,
+      "loss": 0.1111,
+      "step": 3671
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741019639067256e-05,
+      "loss": 0.1022,
+      "step": 3672
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737775007079334e-05,
+      "loss": 0.1195,
+      "step": 3673
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3734529917623936e-05,
+      "loss": 0.1315,
+      "step": 3674
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373128437109823e-05,
+      "loss": 0.1133,
+      "step": 3675
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3728038367899439e-05,
+      "loss": 0.1224,
+      "step": 3676
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.372479190842484e-05,
+      "loss": 0.1316,
+      "step": 3677
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721544993071774e-05,
+      "loss": 0.1217,
+      "step": 3678
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3718297622237624e-05,
+      "loss": 0.1144,
+      "step": 3679
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3715049796319842e-05,
+      "loss": 0.1053,
+      "step": 3680
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371180151571593e-05,
+      "loss": 0.131,
+      "step": 3681
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708552780823443e-05,
+      "loss": 0.1289,
+      "step": 3682
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370530359204e-05,
+      "loss": 0.127,
+      "step": 3683
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702053949763268e-05,
+      "loss": 0.1114,
+      "step": 3684
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369880385439097e-05,
+      "loss": 0.1182,
+      "step": 3685
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695553306320886e-05,
+      "loss": 0.116,
+      "step": 3686
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692302305950853e-05,
+      "loss": 0.1014,
+      "step": 3687
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3689050853678767e-05,
+      "loss": 0.1279,
+      "step": 3688
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685798949902566e-05,
+      "loss": 0.1219,
+      "step": 3689
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682546595020255e-05,
+      "loss": 0.1145,
+      "step": 3690
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3679293789429888e-05,
+      "loss": 0.1147,
+      "step": 3691
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3676040533529582e-05,
+      "loss": 0.127,
+      "step": 3692
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3672786827717502e-05,
+      "loss": 0.1056,
+      "step": 3693
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669532672391868e-05,
+      "loss": 0.1296,
+      "step": 3694
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366627806795096e-05,
+      "loss": 0.1071,
+      "step": 3695
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.366302301479311e-05,
+      "loss": 0.1067,
+      "step": 3696
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36597675133167e-05,
+      "loss": 0.1103,
+      "step": 3697
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3656511563920178e-05,
+      "loss": 0.1179,
+      "step": 3698
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365325516700204e-05,
+      "loss": 0.1121,
+      "step": 3699
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3649998322960833e-05,
+      "loss": 0.1108,
+      "step": 3700
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364674103219517e-05,
+      "loss": 0.1115,
+      "step": 3701
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643483295103703e-05,
+      "loss": 0.1139,
+      "step": 3702
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640225112085155e-05,
+      "loss": 0.1104,
+      "step": 3703
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636966483538289e-05,
+      "loss": 0.1231,
+      "step": 3704
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633707409861939e-05,
+      "loss": 0.121,
+      "step": 3705
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630447891454976e-05,
+      "loss": 0.1107,
+      "step": 3706
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362718792871634e-05,
+      "loss": 0.1033,
+      "step": 3707
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3623927522045014e-05,
+      "loss": 0.1194,
+      "step": 3708
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620666671840042e-05,
+      "loss": 0.1171,
+      "step": 3709
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361740537850052e-05,
+      "loss": 0.1222,
+      "step": 3710
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3614143642425596e-05,
+      "loss": 0.1116,
+      "step": 3711
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3610881464014481e-05,
+      "loss": 0.1174,
+      "step": 3712
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3607618843666431e-05,
+      "loss": 0.1242,
+      "step": 3713
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604355781780756e-05,
+      "loss": 0.1103,
+      "step": 3714
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360109227875683e-05,
+      "loss": 0.1021,
+      "step": 3715
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3597828334994065e-05,
+      "loss": 0.1243,
+      "step": 3716
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594563950891945e-05,
+      "loss": 0.1148,
+      "step": 3717
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3591299126849994e-05,
+      "loss": 0.1126,
+      "step": 3718
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35880338632678e-05,
+      "loss": 0.1132,
+      "step": 3719
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584768160544995e-05,
+      "loss": 0.1138,
+      "step": 3720
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.358150201908127e-05,
+      "loss": 0.1074,
+      "step": 3721
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578235439276368e-05,
+      "loss": 0.1108,
+      "step": 3722
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574968421530089e-05,
+      "loss": 0.1258,
+      "step": 3723
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571700966242283e-05,
+      "loss": 0.1144,
+      "step": 3724
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3568433073812857e-05,
+      "loss": 0.1155,
+      "step": 3725
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565164744641765e-05,
+      "loss": 0.1175,
+      "step": 3726
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561895979129022e-05,
+      "loss": 0.123,
+      "step": 3727
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3558626777674695e-05,
+      "loss": 0.1211,
+      "step": 3728
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35553571406789e-05,
+      "loss": 0.1107,
+      "step": 3729
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3552087068541811e-05,
+      "loss": 0.115,
+      "step": 3730
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354881656166365e-05,
+      "loss": 0.1159,
+      "step": 3731
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545545620444697e-05,
+      "loss": 0.1213,
+      "step": 3732
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3542274245285284e-05,
+      "loss": 0.1254,
+      "step": 3733
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539002436585795e-05,
+      "loss": 0.1107,
+      "step": 3734
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535730194746668e-05,
+      "loss": 0.1188,
+      "step": 3735
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3532457520168393e-05,
+      "loss": 0.1164,
+      "step": 3736
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529184413251517e-05,
+      "loss": 0.111,
+      "step": 3737
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352591087439663e-05,
+      "loss": 0.1222,
+      "step": 3738
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522636904004388e-05,
+      "loss": 0.1138,
+      "step": 3739
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351936250247549e-05,
+      "loss": 0.1073,
+      "step": 3740
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516087670210693e-05,
+      "loss": 0.1189,
+      "step": 3741
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512812407610802e-05,
+      "loss": 0.1234,
+      "step": 3742
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3509536715076683e-05,
+      "loss": 0.1052,
+      "step": 3743
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506260593009239e-05,
+      "loss": 0.1115,
+      "step": 3744
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502984041809443e-05,
+      "loss": 0.1324,
+      "step": 3745
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499707061878313e-05,
+      "loss": 0.1128,
+      "step": 3746
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3496429653616919e-05,
+      "loss": 0.1214,
+      "step": 3747
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3493151817426382e-05,
+      "loss": 0.1108,
+      "step": 3748
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489873553707881e-05,
+      "loss": 0.1106,
+      "step": 3749
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348659486286264e-05,
+      "loss": 0.1133,
+      "step": 3750
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348331574529194e-05,
+      "loss": 0.1239,
+      "step": 3751
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480036201397118e-05,
+      "loss": 0.1228,
+      "step": 3752
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347675623157955e-05,
+      "loss": 0.1229,
+      "step": 3753
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347347583624068e-05,
+      "loss": 0.1232,
+      "step": 3754
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470195015781989e-05,
+      "loss": 0.1063,
+      "step": 3755
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466913770605026e-05,
+      "loss": 0.1074,
+      "step": 3756
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3463632101111379e-05,
+      "loss": 0.1144,
+      "step": 3757
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460350007702691e-05,
+      "loss": 0.1159,
+      "step": 3758
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3457067490780663e-05,
+      "loss": 0.1204,
+      "step": 3759
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345378455074704e-05,
+      "loss": 0.1148,
+      "step": 3760
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3450501188003623e-05,
+      "loss": 0.113,
+      "step": 3761
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344721740295227e-05,
+      "loss": 0.1192,
+      "step": 3762
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443933195994871e-05,
+      "loss": 0.1158,
+      "step": 3763
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440648567533392e-05,
+      "loss": 0.1122,
+      "step": 3764
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437363517969834e-05,
+      "loss": 0.1059,
+      "step": 3765
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343407804770626e-05,
+      "loss": 0.1158,
+      "step": 3766
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430792157144776e-05,
+      "loss": 0.1085,
+      "step": 3767
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427505846687548e-05,
+      "loss": 0.1199,
+      "step": 3768
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3424219116736783e-05,
+      "loss": 0.1227,
+      "step": 3769
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3420931967694746e-05,
+      "loss": 0.1069,
+      "step": 3770
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417644399963754e-05,
+      "loss": 0.1096,
+      "step": 3771
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3414356413946173e-05,
+      "loss": 0.1158,
+      "step": 3772
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411068010044418e-05,
+      "loss": 0.1185,
+      "step": 3773
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407779188660963e-05,
+      "loss": 0.1179,
+      "step": 3774
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3404489950198328e-05,
+      "loss": 0.1376,
+      "step": 3775
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401200295059078e-05,
+      "loss": 0.1184,
+      "step": 3776
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3397910223645836e-05,
+      "loss": 0.1116,
+      "step": 3777
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3394619736361278e-05,
+      "loss": 0.1171,
+      "step": 3778
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391328833608129e-05,
+      "loss": 0.1065,
+      "step": 3779
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338803751578916e-05,
+      "loss": 0.1093,
+      "step": 3780
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3384745783307198e-05,
+      "loss": 0.1087,
+      "step": 3781
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381453636565118e-05,
+      "loss": 0.116,
+      "step": 3782
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378161075965848e-05,
+      "loss": 0.1158,
+      "step": 3783
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3374868101912369e-05,
+      "loss": 0.1182,
+      "step": 3784
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33715747148077e-05,
+      "loss": 0.1133,
+      "step": 3785
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336828091505493e-05,
+      "loss": 0.1023,
+      "step": 3786
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336498670305718e-05,
+      "loss": 0.1137,
+      "step": 3787
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3361692079217634e-05,
+      "loss": 0.1107,
+      "step": 3788
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358397043939522e-05,
+      "loss": 0.1102,
+      "step": 3789
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355101597626123e-05,
+      "loss": 0.1088,
+      "step": 3790
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.335180574068077e-05,
+      "loss": 0.1216,
+      "step": 3791
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3348509473506841e-05,
+      "loss": 0.1184,
+      "step": 3792
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345212796507769e-05,
+      "loss": 0.1229,
+      "step": 3793
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341915710087035e-05,
+      "loss": 0.1039,
+      "step": 3794
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333861821464817e-05,
+      "loss": 0.1145,
+      "step": 3795
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335320310594755e-05,
+      "loss": 0.1143,
+      "step": 3796
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332021998330422e-05,
+      "loss": 0.1159,
+      "step": 3797
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3328723278258857e-05,
+      "loss": 0.109,
+      "step": 3798
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3325424150783783e-05,
+      "loss": 0.1202,
+      "step": 3799
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322124616308987e-05,
+      "loss": 0.1188,
+      "step": 3800
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318824675238298e-05,
+      "loss": 0.1123,
+      "step": 3801
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3315524327975598e-05,
+      "loss": 0.1235,
+      "step": 3802
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312223574924815e-05,
+      "loss": 0.1137,
+      "step": 3803
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3308922416489936e-05,
+      "loss": 0.0961,
+      "step": 3804
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330562085307498e-05,
+      "loss": 0.1083,
+      "step": 3805
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302318885084038e-05,
+      "loss": 0.1124,
+      "step": 3806
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299016512921233e-05,
+      "loss": 0.1217,
+      "step": 3807
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3295713736990741e-05,
+      "loss": 0.1133,
+      "step": 3808
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292410557696793e-05,
+      "loss": 0.1057,
+      "step": 3809
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3289106975443666e-05,
+      "loss": 0.119,
+      "step": 3810
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285802990635687e-05,
+      "loss": 0.1176,
+      "step": 3811
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328249860367723e-05,
+      "loss": 0.1076,
+      "step": 3812
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279193814972719e-05,
+      "loss": 0.1211,
+      "step": 3813
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275888624926634e-05,
+      "loss": 0.1173,
+      "step": 3814
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272583033943494e-05,
+      "loss": 0.1092,
+      "step": 3815
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3269277042427869e-05,
+      "loss": 0.1308,
+      "step": 3816
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3265970650784388e-05,
+      "loss": 0.1254,
+      "step": 3817
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326266385941771e-05,
+      "loss": 0.1049,
+      "step": 3818
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3259356668732564e-05,
+      "loss": 0.1213,
+      "step": 3819
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3256049079133715e-05,
+      "loss": 0.12,
+      "step": 3820
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252741091025979e-05,
+      "loss": 0.1129,
+      "step": 3821
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249432704814223e-05,
+      "loss": 0.1067,
+      "step": 3822
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.324612392090336e-05,
+      "loss": 0.109,
+      "step": 3823
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242814739698353e-05,
+      "loss": 0.1114,
+      "step": 3824
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3239505161604216e-05,
+      "loss": 0.1151,
+      "step": 3825
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236195187026012e-05,
+      "loss": 0.1152,
+      "step": 3826
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3232884816368841e-05,
+      "loss": 0.1186,
+      "step": 3827
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3229574050037863e-05,
+      "loss": 0.1174,
+      "step": 3828
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3226262888438293e-05,
+      "loss": 0.1025,
+      "step": 3829
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222951331975373e-05,
+      "loss": 0.1228,
+      "step": 3830
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219639381054412e-05,
+      "loss": 0.1213,
+      "step": 3831
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216327036080759e-05,
+      "loss": 0.1101,
+      "step": 3832
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213014297459813e-05,
+      "loss": 0.1286,
+      "step": 3833
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209701165597019e-05,
+      "loss": 0.1028,
+      "step": 3834
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206387640897882e-05,
+      "loss": 0.1208,
+      "step": 3835
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203073723767932e-05,
+      "loss": 0.1136,
+      "step": 3836
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3199759414612768e-05,
+      "loss": 0.117,
+      "step": 3837
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3196444713838028e-05,
+      "loss": 0.1127,
+      "step": 3838
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.31931296218494e-05,
+      "loss": 0.1191,
+      "step": 3839
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189814139052615e-05,
+      "loss": 0.1135,
+      "step": 3840
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3186498265853462e-05,
+      "loss": 0.1086,
+      "step": 3841
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183182002657768e-05,
+      "loss": 0.1053,
+      "step": 3842
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3179865349871412e-05,
+      "loss": 0.1196,
+      "step": 3843
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3176548307900321e-05,
+      "loss": 0.1164,
+      "step": 3844
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3173230877150467e-05,
+      "loss": 0.1158,
+      "step": 3845
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169913058027874e-05,
+      "loss": 0.1163,
+      "step": 3846
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3166594850938612e-05,
+      "loss": 0.1066,
+      "step": 3847
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3163276256288793e-05,
+      "loss": 0.0971,
+      "step": 3848
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159957274484584e-05,
+      "loss": 0.1178,
+      "step": 3849
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3156637905932196e-05,
+      "loss": 0.1169,
+      "step": 3850
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153318151037886e-05,
+      "loss": 0.1078,
+      "step": 3851
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149998010207959e-05,
+      "loss": 0.1165,
+      "step": 3852
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146677483848773e-05,
+      "loss": 0.115,
+      "step": 3853
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143356572366724e-05,
+      "loss": 0.11,
+      "step": 3854
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3140035276168259e-05,
+      "loss": 0.1045,
+      "step": 3855
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136713595659875e-05,
+      "loss": 0.1133,
+      "step": 3856
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133391531248111e-05,
+      "loss": 0.1021,
+      "step": 3857
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3130069083339563e-05,
+      "loss": 0.1112,
+      "step": 3858
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3126746252340856e-05,
+      "loss": 0.1271,
+      "step": 3859
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312342303865868e-05,
+      "loss": 0.1218,
+      "step": 3860
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3120099442699758e-05,
+      "loss": 0.1216,
+      "step": 3861
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311677546487087e-05,
+      "loss": 0.1298,
+      "step": 3862
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113451105578837e-05,
+      "loss": 0.1159,
+      "step": 3863
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3110126365230528e-05,
+      "loss": 0.1215,
+      "step": 3864
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3106801244232866e-05,
+      "loss": 0.1107,
+      "step": 3865
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103475742992804e-05,
+      "loss": 0.1205,
+      "step": 3866
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3100149861917355e-05,
+      "loss": 0.112,
+      "step": 3867
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3096823601413573e-05,
+      "loss": 0.1275,
+      "step": 3868
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093496961888565e-05,
+      "loss": 0.1087,
+      "step": 3869
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.1152,
+      "step": 3870
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3086842547403502e-05,
+      "loss": 0.1152,
+      "step": 3871
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308351477325788e-05,
+      "loss": 0.1081,
+      "step": 3872
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3080186621719902e-05,
+      "loss": 0.1169,
+      "step": 3873
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076858093196897e-05,
+      "loss": 0.1114,
+      "step": 3874
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073529188096252e-05,
+      "loss": 0.121,
+      "step": 3875
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3070199906825385e-05,
+      "loss": 0.1201,
+      "step": 3876
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066870249791776e-05,
+      "loss": 0.1237,
+      "step": 3877
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3063540217402934e-05,
+      "loss": 0.1119,
+      "step": 3878
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3060209810066426e-05,
+      "loss": 0.1121,
+      "step": 3879
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3056879028189866e-05,
+      "loss": 0.1154,
+      "step": 3880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053547872180905e-05,
+      "loss": 0.1174,
+      "step": 3881
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3050216342447245e-05,
+      "loss": 0.1146,
+      "step": 3882
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3046884439396632e-05,
+      "loss": 0.1167,
+      "step": 3883
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043552163436859e-05,
+      "loss": 0.116,
+      "step": 3884
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040219514975765e-05,
+      "loss": 0.119,
+      "step": 3885
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3036886494421235e-05,
+      "loss": 0.1122,
+      "step": 3886
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033553102181198e-05,
+      "loss": 0.1121,
+      "step": 3887
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3030219338663628e-05,
+      "loss": 0.1138,
+      "step": 3888
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026885204276547e-05,
+      "loss": 0.1165,
+      "step": 3889
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023550699428015e-05,
+      "loss": 0.112,
+      "step": 3890
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3020215824526154e-05,
+      "loss": 0.1078,
+      "step": 3891
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301688057997911e-05,
+      "loss": 0.1201,
+      "step": 3892
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301354496619509e-05,
+      "loss": 0.11,
+      "step": 3893
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3010208983582337e-05,
+      "loss": 0.1037,
+      "step": 3894
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3006872632549147e-05,
+      "loss": 0.1261,
+      "step": 3895
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003535913503857e-05,
+      "loss": 0.123,
+      "step": 3896
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3000198826854846e-05,
+      "loss": 0.1136,
+      "step": 3897
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996861373010543e-05,
+      "loss": 0.1091,
+      "step": 3898
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2993523552379421e-05,
+      "loss": 0.1129,
+      "step": 3899
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2990185365369993e-05,
+      "loss": 0.1015,
+      "step": 3900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986846812390828e-05,
+      "loss": 0.0999,
+      "step": 3901
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983507893850526e-05,
+      "loss": 0.1086,
+      "step": 3902
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.298016861015774e-05,
+      "loss": 0.1062,
+      "step": 3903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2976828961721167e-05,
+      "loss": 0.113,
+      "step": 3904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973488948949545e-05,
+      "loss": 0.1188,
+      "step": 3905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2970148572251662e-05,
+      "loss": 0.135,
+      "step": 3906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2966807832036348e-05,
+      "loss": 0.118,
+      "step": 3907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963466728712474e-05,
+      "loss": 0.1138,
+      "step": 3908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2960125262688966e-05,
+      "loss": 0.0978,
+      "step": 3909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2956783434374778e-05,
+      "loss": 0.1122,
+      "step": 3910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953441244178922e-05,
+      "loss": 0.1203,
+      "step": 3911
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2950098692510449e-05,
+      "loss": 0.1076,
+      "step": 3912
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2946755779778456e-05,
+      "loss": 0.1061,
+      "step": 3913
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943412506392083e-05,
+      "loss": 0.1226,
+      "step": 3914
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940068872760513e-05,
+      "loss": 0.1081,
+      "step": 3915
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936724879292975e-05,
+      "loss": 0.1126,
+      "step": 3916
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2933380526398743e-05,
+      "loss": 0.1226,
+      "step": 3917
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930035814487128e-05,
+      "loss": 0.1091,
+      "step": 3918
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926690743967497e-05,
+      "loss": 0.1261,
+      "step": 3919
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2923345315249251e-05,
+      "loss": 0.1086,
+      "step": 3920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2919999528741837e-05,
+      "loss": 0.1248,
+      "step": 3921
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916653384854752e-05,
+      "loss": 0.1235,
+      "step": 3922
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2913306883997528e-05,
+      "loss": 0.1225,
+      "step": 3923
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2909960026579742e-05,
+      "loss": 0.1283,
+      "step": 3924
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906612813011019e-05,
+      "loss": 0.1171,
+      "step": 3925
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2903265243701028e-05,
+      "loss": 0.1166,
+      "step": 3926
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2899917319059476e-05,
+      "loss": 0.1054,
+      "step": 3927
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896569039496116e-05,
+      "loss": 0.113,
+      "step": 3928
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2893220405420748e-05,
+      "loss": 0.1215,
+      "step": 3929
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288987141724321e-05,
+      "loss": 0.11,
+      "step": 3930
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886522075373387e-05,
+      "loss": 0.1021,
+      "step": 3931
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883172380221207e-05,
+      "loss": 0.1164,
+      "step": 3932
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2879822332196639e-05,
+      "loss": 0.1115,
+      "step": 3933
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876471931709698e-05,
+      "loss": 0.1064,
+      "step": 3934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2873121179170434e-05,
+      "loss": 0.1016,
+      "step": 3935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2869770074988955e-05,
+      "loss": 0.1259,
+      "step": 3936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866418619575399e-05,
+      "loss": 0.1071,
+      "step": 3937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863066813339953e-05,
+      "loss": 0.1077,
+      "step": 3938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859714656692845e-05,
+      "loss": 0.1135,
+      "step": 3939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285636215004435e-05,
+      "loss": 0.1168,
+      "step": 3940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2853009293804777e-05,
+      "loss": 0.11,
+      "step": 3941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2849656088384486e-05,
+      "loss": 0.0998,
+      "step": 3942
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846302534193878e-05,
+      "loss": 0.1158,
+      "step": 3943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842948631643392e-05,
+      "loss": 0.1141,
+      "step": 3944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2839594381143518e-05,
+      "loss": 0.1102,
+      "step": 3945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283623978310478e-05,
+      "loss": 0.1075,
+      "step": 3946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283288483793775e-05,
+      "loss": 0.1126,
+      "step": 3947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282952954605304e-05,
+      "loss": 0.1116,
+      "step": 3948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2826173907861304e-05,
+      "loss": 0.1224,
+      "step": 3949
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822817923773243e-05,
+      "loss": 0.1218,
+      "step": 3950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819461594199594e-05,
+      "loss": 0.1203,
+      "step": 3951
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.281610491955114e-05,
+      "loss": 0.1192,
+      "step": 3952
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812747900238707e-05,
+      "loss": 0.1073,
+      "step": 3953
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2809390536673161e-05,
+      "loss": 0.1232,
+      "step": 3954
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280603282926541e-05,
+      "loss": 0.1172,
+      "step": 3955
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802674778426405e-05,
+      "loss": 0.115,
+      "step": 3956
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2799316384567137e-05,
+      "loss": 0.1325,
+      "step": 3957
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795957648098645e-05,
+      "loss": 0.1154,
+      "step": 3958
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2792598569432005e-05,
+      "loss": 0.1268,
+      "step": 3959
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2789239148978331e-05,
+      "loss": 0.118,
+      "step": 3960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278587938714879e-05,
+      "loss": 0.1197,
+      "step": 3961
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2782519284354582e-05,
+      "loss": 0.1226,
+      "step": 3962
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277915884100695e-05,
+      "loss": 0.121,
+      "step": 3963
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2775798057517178e-05,
+      "loss": 0.1184,
+      "step": 3964
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2772436934296599e-05,
+      "loss": 0.128,
+      "step": 3965
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2769075471756577e-05,
+      "loss": 0.1185,
+      "step": 3966
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765713670308521e-05,
+      "loss": 0.1259,
+      "step": 3967
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276235153036389e-05,
+      "loss": 0.1132,
+      "step": 3968
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2758989052334172e-05,
+      "loss": 0.113,
+      "step": 3969
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.27556262366309e-05,
+      "loss": 0.1058,
+      "step": 3970
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2752263083665658e-05,
+      "loss": 0.1242,
+      "step": 3971
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2748899593850054e-05,
+      "loss": 0.1166,
+      "step": 3972
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745535767595754e-05,
+      "loss": 0.116,
+      "step": 3973
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742171605314452e-05,
+      "loss": 0.1214,
+      "step": 3974
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2738807107417896e-05,
+      "loss": 0.1166,
+      "step": 3975
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735442274317866e-05,
+      "loss": 0.1226,
+      "step": 3976
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732077106426178e-05,
+      "loss": 0.1307,
+      "step": 3977
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728711604154702e-05,
+      "loss": 0.1092,
+      "step": 3978
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272534576791534e-05,
+      "loss": 0.1231,
+      "step": 3979
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272197959812004e-05,
+      "loss": 0.1135,
+      "step": 3980
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271861309518079e-05,
+      "loss": 0.1212,
+      "step": 3981
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2715246259509613e-05,
+      "loss": 0.1159,
+      "step": 3982
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271187909151858e-05,
+      "loss": 0.1143,
+      "step": 3983
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2708511591619798e-05,
+      "loss": 0.111,
+      "step": 3984
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270514376022542e-05,
+      "loss": 0.1016,
+      "step": 3985
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2701775597747635e-05,
+      "loss": 0.1164,
+      "step": 3986
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2698407104598672e-05,
+      "loss": 0.1056,
+      "step": 3987
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2695038281190798e-05,
+      "loss": 0.1123,
+      "step": 3988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269166912793633e-05,
+      "loss": 0.1181,
+      "step": 3989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688299645247618e-05,
+      "loss": 0.112,
+      "step": 3990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2684929833537055e-05,
+      "loss": 0.1101,
+      "step": 3991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268155969321707e-05,
+      "loss": 0.1247,
+      "step": 3992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678189224700141e-05,
+      "loss": 0.1094,
+      "step": 3993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2674818428398777e-05,
+      "loss": 0.1186,
+      "step": 3994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2671447304725531e-05,
+      "loss": 0.1177,
+      "step": 3995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2668075854092996e-05,
+      "loss": 0.1129,
+      "step": 3996
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664704076913807e-05,
+      "loss": 0.1042,
+      "step": 3997
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661331973600635e-05,
+      "loss": 0.1137,
+      "step": 3998
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2657959544566194e-05,
+      "loss": 0.1192,
+      "step": 3999
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654586790223236e-05,
+      "loss": 0.1311,
+      "step": 4000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265121371098455e-05,
+      "loss": 0.127,
+      "step": 4001
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647840307262972e-05,
+      "loss": 0.1194,
+      "step": 4002
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2644466579471378e-05,
+      "loss": 0.1102,
+      "step": 4003
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264109252802267e-05,
+      "loss": 0.1157,
+      "step": 4004
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2637718153329808e-05,
+      "loss": 0.1168,
+      "step": 4005
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634343455805777e-05,
+      "loss": 0.1064,
+      "step": 4006
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263096843586361e-05,
+      "loss": 0.1215,
+      "step": 4007
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2627593093916377e-05,
+      "loss": 0.1084,
+      "step": 4008
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2624217430377183e-05,
+      "loss": 0.1208,
+      "step": 4009
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2620841445659182e-05,
+      "loss": 0.1126,
+      "step": 4010
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2617465140175562e-05,
+      "loss": 0.109,
+      "step": 4011
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614088514339546e-05,
+      "loss": 0.1155,
+      "step": 4012
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2610711568564402e-05,
+      "loss": 0.1279,
+      "step": 4013
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607334303263437e-05,
+      "loss": 0.1197,
+      "step": 4014
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260395671884999e-05,
+      "loss": 0.119,
+      "step": 4015
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2600578815737452e-05,
+      "loss": 0.1117,
+      "step": 4016
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597200594339245e-05,
+      "loss": 0.1164,
+      "step": 4017
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2593822055068825e-05,
+      "loss": 0.103,
+      "step": 4018
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590443198339697e-05,
+      "loss": 0.1106,
+      "step": 4019
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587064024565399e-05,
+      "loss": 0.1133,
+      "step": 4020
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583684534159506e-05,
+      "loss": 0.1215,
+      "step": 4021
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258030472753564e-05,
+      "loss": 0.1251,
+      "step": 4022
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576924605107456e-05,
+      "loss": 0.1041,
+      "step": 4023
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573544167288647e-05,
+      "loss": 0.1124,
+      "step": 4024
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2570163414492944e-05,
+      "loss": 0.1181,
+      "step": 4025
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.256678234713412e-05,
+      "loss": 0.1166,
+      "step": 4026
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2563400965625985e-05,
+      "loss": 0.1243,
+      "step": 4027
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2560019270382389e-05,
+      "loss": 0.1154,
+      "step": 4028
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556637261817217e-05,
+      "loss": 0.1108,
+      "step": 4029
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553254940344395e-05,
+      "loss": 0.1166,
+      "step": 4030
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549872306377883e-05,
+      "loss": 0.1203,
+      "step": 4031
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2546489360331686e-05,
+      "loss": 0.1133,
+      "step": 4032
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543106102619844e-05,
+      "loss": 0.1182,
+      "step": 4033
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2539722533656433e-05,
+      "loss": 0.1086,
+      "step": 4034
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2536338653855575e-05,
+      "loss": 0.1045,
+      "step": 4035
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2532954463631415e-05,
+      "loss": 0.1197,
+      "step": 4036
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2529569963398149e-05,
+      "loss": 0.1213,
+      "step": 4037
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252618515357001e-05,
+      "loss": 0.1099,
+      "step": 4038
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522800034561265e-05,
+      "loss": 0.1199,
+      "step": 4039
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2519414606786215e-05,
+      "loss": 0.1104,
+      "step": 4040
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2516028870659208e-05,
+      "loss": 0.1148,
+      "step": 4041
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512642826594626e-05,
+      "loss": 0.113,
+      "step": 4042
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509256475006883e-05,
+      "loss": 0.1128,
+      "step": 4043
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250586981631044e-05,
+      "loss": 0.1177,
+      "step": 4044
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250248285091979e-05,
+      "loss": 0.1243,
+      "step": 4045
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499095579249468e-05,
+      "loss": 0.1055,
+      "step": 4046
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2495708001714035e-05,
+      "loss": 0.1238,
+      "step": 4047
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492320118728109e-05,
+      "loss": 0.1291,
+      "step": 4048
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2488931930706323e-05,
+      "loss": 0.1247,
+      "step": 4049
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485543438063369e-05,
+      "loss": 0.1151,
+      "step": 4050
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2482154641213954e-05,
+      "loss": 0.1237,
+      "step": 4051
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2478765540572843e-05,
+      "loss": 0.1269,
+      "step": 4052
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475376136554825e-05,
+      "loss": 0.1304,
+      "step": 4053
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471986429574735e-05,
+      "loss": 0.1043,
+      "step": 4054
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2468596420047432e-05,
+      "loss": 0.1068,
+      "step": 4055
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2465206108387827e-05,
+      "loss": 0.1121,
+      "step": 4056
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461815495010856e-05,
+      "loss": 0.1183,
+      "step": 4057
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2458424580331505e-05,
+      "loss": 0.1258,
+      "step": 4058
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455033364764781e-05,
+      "loss": 0.1162,
+      "step": 4059
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2451641848725746e-05,
+      "loss": 0.1245,
+      "step": 4060
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448250032629478e-05,
+      "loss": 0.1148,
+      "step": 4061
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244485791689111e-05,
+      "loss": 0.1246,
+      "step": 4062
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2441465501925797e-05,
+      "loss": 0.1274,
+      "step": 4063
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438072788148744e-05,
+      "loss": 0.1047,
+      "step": 4064
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434679775975183e-05,
+      "loss": 0.1227,
+      "step": 4065
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2431286465820386e-05,
+      "loss": 0.1111,
+      "step": 4066
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2427892858099666e-05,
+      "loss": 0.1205,
+      "step": 4067
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424498953228362e-05,
+      "loss": 0.1234,
+      "step": 4068
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2421104751621855e-05,
+      "loss": 0.1224,
+      "step": 4069
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2417710253695565e-05,
+      "loss": 0.1161,
+      "step": 4070
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414315459864948e-05,
+      "loss": 0.1071,
+      "step": 4071
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241092037054549e-05,
+      "loss": 0.1166,
+      "step": 4072
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407524986152718e-05,
+      "loss": 0.1209,
+      "step": 4073
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404129307102195e-05,
+      "loss": 0.1158,
+      "step": 4074
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2400733333809516e-05,
+      "loss": 0.1147,
+      "step": 4075
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2397337066690321e-05,
+      "loss": 0.1124,
+      "step": 4076
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393940506160273e-05,
+      "loss": 0.1066,
+      "step": 4077
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390543652635088e-05,
+      "loss": 0.1303,
+      "step": 4078
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2387146506530496e-05,
+      "loss": 0.108,
+      "step": 4079
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383749068262286e-05,
+      "loss": 0.1162,
+      "step": 4080
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2380351338246262e-05,
+      "loss": 0.1199,
+      "step": 4081
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376953316898282e-05,
+      "loss": 0.113,
+      "step": 4082
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373555004634225e-05,
+      "loss": 0.1108,
+      "step": 4083
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2370156401870015e-05,
+      "loss": 0.1063,
+      "step": 4084
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2366757509021603e-05,
+      "loss": 0.1134,
+      "step": 4085
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2363358326504987e-05,
+      "loss": 0.1131,
+      "step": 4086
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2359958854736187e-05,
+      "loss": 0.1091,
+      "step": 4087
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235655909413127e-05,
+      "loss": 0.1225,
+      "step": 4088
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2353159045106336e-05,
+      "loss": 0.1088,
+      "step": 4089
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2349758708077514e-05,
+      "loss": 0.1128,
+      "step": 4090
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346358083460975e-05,
+      "loss": 0.1272,
+      "step": 4091
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342957171672919e-05,
+      "loss": 0.117,
+      "step": 4092
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339555973129588e-05,
+      "loss": 0.0972,
+      "step": 4093
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336154488247255e-05,
+      "loss": 0.1135,
+      "step": 4094
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332752717442224e-05,
+      "loss": 0.1084,
+      "step": 4095
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232935066113085e-05,
+      "loss": 0.1166,
+      "step": 4096
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2325948319729499e-05,
+      "loss": 0.1072,
+      "step": 4097
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322545693654595e-05,
+      "loss": 0.1112,
+      "step": 4098
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231914278332258e-05,
+      "loss": 0.1124,
+      "step": 4099
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2315739589149938e-05,
+      "loss": 0.1097,
+      "step": 4100
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231233611155319e-05,
+      "loss": 0.1116,
+      "step": 4101
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2308932350948888e-05,
+      "loss": 0.1161,
+      "step": 4102
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2305528307753615e-05,
+      "loss": 0.1103,
+      "step": 4103
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2302123982384003e-05,
+      "loss": 0.1286,
+      "step": 4104
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2298719375256695e-05,
+      "loss": 0.1154,
+      "step": 4105
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295314486788392e-05,
+      "loss": 0.1153,
+      "step": 4106
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291909317395817e-05,
+      "loss": 0.1124,
+      "step": 4107
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2288503867495728e-05,
+      "loss": 0.1166,
+      "step": 4108
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285098137504918e-05,
+      "loss": 0.1164,
+      "step": 4109
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281692127840219e-05,
+      "loss": 0.1083,
+      "step": 4110
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2278285838918492e-05,
+      "loss": 0.1077,
+      "step": 4111
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2274879271156632e-05,
+      "loss": 0.1152,
+      "step": 4112
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271472424971576e-05,
+      "loss": 0.1002,
+      "step": 4113
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2268065300780281e-05,
+      "loss": 0.1092,
+      "step": 4114
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264657898999754e-05,
+      "loss": 0.1095,
+      "step": 4115
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261250220047019e-05,
+      "loss": 0.1186,
+      "step": 4116
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225784226433915e-05,
+      "loss": 0.1091,
+      "step": 4117
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254434032293245e-05,
+      "loss": 0.1239,
+      "step": 4118
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2251025524326439e-05,
+      "loss": 0.1212,
+      "step": 4119
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2247616740855898e-05,
+      "loss": 0.1134,
+      "step": 4120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244207682298832e-05,
+      "loss": 0.111,
+      "step": 4121
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240798349072467e-05,
+      "loss": 0.1071,
+      "step": 4122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237388741594079e-05,
+      "loss": 0.1161,
+      "step": 4123
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233978860280968e-05,
+      "loss": 0.1195,
+      "step": 4124
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.223056870555047e-05,
+      "loss": 0.1251,
+      "step": 4125
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227158277819956e-05,
+      "loss": 0.1161,
+      "step": 4126
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2223747577506833e-05,
+      "loss": 0.1121,
+      "step": 4127
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2220336605028532e-05,
+      "loss": 0.1142,
+      "step": 4128
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2216925360802523e-05,
+      "loss": 0.1153,
+      "step": 4129
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2213513845246312e-05,
+      "loss": 0.119,
+      "step": 4130
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2210102058777433e-05,
+      "loss": 0.1121,
+      "step": 4131
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2206690001813459e-05,
+      "loss": 0.1054,
+      "step": 4132
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203277674771993e-05,
+      "loss": 0.1206,
+      "step": 4133
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219986507807067e-05,
+      "loss": 0.1286,
+      "step": 4134
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2196452212127155e-05,
+      "loss": 0.1252,
+      "step": 4135
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2193039077359155e-05,
+      "loss": 0.1128,
+      "step": 4136
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.21896256741844e-05,
+      "loss": 0.1072,
+      "step": 4137
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2186212003020658e-05,
+      "loss": 0.1043,
+      "step": 4138
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2182798064285737e-05,
+      "loss": 0.1213,
+      "step": 4139
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217938385839746e-05,
+      "loss": 0.1109,
+      "step": 4140
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175969385773695e-05,
+      "loss": 0.122,
+      "step": 4141
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172554646832345e-05,
+      "loss": 0.1115,
+      "step": 4142
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2169139641991336e-05,
+      "loss": 0.1051,
+      "step": 4143
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165724371668635e-05,
+      "loss": 0.1167,
+      "step": 4144
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216230883628224e-05,
+      "loss": 0.1227,
+      "step": 4145
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2158893036250172e-05,
+      "loss": 0.1043,
+      "step": 4146
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155476971990499e-05,
+      "loss": 0.1141,
+      "step": 4147
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152060643921308e-05,
+      "loss": 0.1119,
+      "step": 4148
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.214864405246073e-05,
+      "loss": 0.1175,
+      "step": 4149
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2145227198026919e-05,
+      "loss": 0.1084,
+      "step": 4150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141810081038068e-05,
+      "loss": 0.1081,
+      "step": 4151
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2138392701912395e-05,
+      "loss": 0.1181,
+      "step": 4152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213497506106816e-05,
+      "loss": 0.128,
+      "step": 4153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131557158923642e-05,
+      "loss": 0.1037,
+      "step": 4154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2128138995897162e-05,
+      "loss": 0.1041,
+      "step": 4155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2124720572407075e-05,
+      "loss": 0.1035,
+      "step": 4156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.212130188887176e-05,
+      "loss": 0.1248,
+      "step": 4157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117882945709626e-05,
+      "loss": 0.1119,
+      "step": 4158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2114463743339124e-05,
+      "loss": 0.1196,
+      "step": 4159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111044282178728e-05,
+      "loss": 0.119,
+      "step": 4160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107624562646948e-05,
+      "loss": 0.1141,
+      "step": 4161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104204585162325e-05,
+      "loss": 0.1196,
+      "step": 4162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100784350143433e-05,
+      "loss": 0.111,
+      "step": 4163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097363858008872e-05,
+      "loss": 0.1136,
+      "step": 4164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2093943109177279e-05,
+      "loss": 0.1055,
+      "step": 4165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2090522104067319e-05,
+      "loss": 0.12,
+      "step": 4166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087100843097695e-05,
+      "loss": 0.1194,
+      "step": 4167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208367932668713e-05,
+      "loss": 0.1129,
+      "step": 4168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080257555254389e-05,
+      "loss": 0.1101,
+      "step": 4169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207683552921826e-05,
+      "loss": 0.1183,
+      "step": 4170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2073413248997567e-05,
+      "loss": 0.1097,
+      "step": 4171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069990715011164e-05,
+      "loss": 0.1061,
+      "step": 4172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.206656792767794e-05,
+      "loss": 0.1104,
+      "step": 4173
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2063144887416802e-05,
+      "loss": 0.1102,
+      "step": 4174
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059721594646709e-05,
+      "loss": 0.1083,
+      "step": 4175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2056298049786628e-05,
+      "loss": 0.1111,
+      "step": 4176
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2052874253255573e-05,
+      "loss": 0.1173,
+      "step": 4177
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049450205472585e-05,
+      "loss": 0.1112,
+      "step": 4178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204602590685673e-05,
+      "loss": 0.1231,
+      "step": 4179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204260135782711e-05,
+      "loss": 0.1131,
+      "step": 4180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2039176558802862e-05,
+      "loss": 0.1106,
+      "step": 4181
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203575151020314e-05,
+      "loss": 0.1219,
+      "step": 4182
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2032326212447143e-05,
+      "loss": 0.1192,
+      "step": 4183
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028900665954089e-05,
+      "loss": 0.1187,
+      "step": 4184
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2025474871143236e-05,
+      "loss": 0.1123,
+      "step": 4185
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2022048828433868e-05,
+      "loss": 0.1167,
+      "step": 4186
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.20186225382453e-05,
+      "loss": 0.1055,
+      "step": 4187
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2015196000996872e-05,
+      "loss": 0.099,
+      "step": 4188
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011769217107964e-05,
+      "loss": 0.1172,
+      "step": 4189
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200834218699798e-05,
+      "loss": 0.105,
+      "step": 4190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004914911086352e-05,
+      "loss": 0.1182,
+      "step": 4191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001487389792548e-05,
+      "loss": 0.1083,
+      "step": 4192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998059623536063e-05,
+      "loss": 0.1199,
+      "step": 4193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1994631612736428e-05,
+      "loss": 0.115,
+      "step": 4194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991203357813186e-05,
+      "loss": 0.1157,
+      "step": 4195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987774859185932e-05,
+      "loss": 0.1059,
+      "step": 4196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1984346117274279e-05,
+      "loss": 0.1102,
+      "step": 4197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1980917132497871e-05,
+      "loss": 0.1142,
+      "step": 4198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977487905276384e-05,
+      "loss": 0.1275,
+      "step": 4199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1974058436029523e-05,
+      "loss": 0.1143,
+      "step": 4200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1970628725177016e-05,
+      "loss": 0.1154,
+      "step": 4201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967198773138629e-05,
+      "loss": 0.132,
+      "step": 4202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196376858033416e-05,
+      "loss": 0.1159,
+      "step": 4203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960338147183423e-05,
+      "loss": 0.1077,
+      "step": 4204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956907474106277e-05,
+      "loss": 0.1073,
+      "step": 4205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.19534765615226e-05,
+      "loss": 0.1113,
+      "step": 4206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950045409852305e-05,
+      "loss": 0.1188,
+      "step": 4207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946614019515326e-05,
+      "loss": 0.1216,
+      "step": 4208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943182390931638e-05,
+      "loss": 0.1163,
+      "step": 4209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939750524521238e-05,
+      "loss": 0.1115,
+      "step": 4210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1936318420704152e-05,
+      "loss": 0.1266,
+      "step": 4211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1932886079900434e-05,
+      "loss": 0.1113,
+      "step": 4212
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1929453502530173e-05,
+      "loss": 0.1192,
+      "step": 4213
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926020689013483e-05,
+      "loss": 0.1046,
+      "step": 4214
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1922587639770502e-05,
+      "loss": 0.1218,
+      "step": 4215
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919154355221407e-05,
+      "loss": 0.1171,
+      "step": 4216
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19157208357864e-05,
+      "loss": 0.1037,
+      "step": 4217
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1912287081885708e-05,
+      "loss": 0.1237,
+      "step": 4218
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190885309393959e-05,
+      "loss": 0.1227,
+      "step": 4219
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190541887236833e-05,
+      "loss": 0.1083,
+      "step": 4220
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1901984417592249e-05,
+      "loss": 0.1077,
+      "step": 4221
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898549730031685e-05,
+      "loss": 0.1153,
+      "step": 4222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895114810107016e-05,
+      "loss": 0.1087,
+      "step": 4223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189167965823864e-05,
+      "loss": 0.105,
+      "step": 4224
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888244274846985e-05,
+      "loss": 0.1064,
+      "step": 4225
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1884808660352512e-05,
+      "loss": 0.1248,
+      "step": 4226
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881372815175702e-05,
+      "loss": 0.1127,
+      "step": 4227
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1877936739737075e-05,
+      "loss": 0.1065,
+      "step": 4228
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874500434457171e-05,
+      "loss": 0.1181,
+      "step": 4229
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871063899756562e-05,
+      "loss": 0.1102,
+      "step": 4230
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867627136055841e-05,
+      "loss": 0.1075,
+      "step": 4231
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864190143775645e-05,
+      "loss": 0.1052,
+      "step": 4232
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860752923336616e-05,
+      "loss": 0.1119,
+      "step": 4233
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857315475159446e-05,
+      "loss": 0.118,
+      "step": 4234
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185387779966484e-05,
+      "loss": 0.1155,
+      "step": 4235
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185043989727354e-05,
+      "loss": 0.126,
+      "step": 4236
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847001768406307e-05,
+      "loss": 0.1226,
+      "step": 4237
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1843563413483941e-05,
+      "loss": 0.1201,
+      "step": 4238
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840124832927258e-05,
+      "loss": 0.1071,
+      "step": 4239
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836686027157111e-05,
+      "loss": 0.1242,
+      "step": 4240
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1833246996594376e-05,
+      "loss": 0.1171,
+      "step": 4241
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1829807741659954e-05,
+      "loss": 0.113,
+      "step": 4242
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1826368262774779e-05,
+      "loss": 0.1145,
+      "step": 4243
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.182292856035981e-05,
+      "loss": 0.1132,
+      "step": 4244
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181948863483603e-05,
+      "loss": 0.1192,
+      "step": 4245
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1816048486624456e-05,
+      "loss": 0.1238,
+      "step": 4246
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812608116146126e-05,
+      "loss": 0.1101,
+      "step": 4247
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1809167523822115e-05,
+      "loss": 0.1096,
+      "step": 4248
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.180572671007351e-05,
+      "loss": 0.1114,
+      "step": 4249
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1802285675321437e-05,
+      "loss": 0.1237,
+      "step": 4250
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798844419987045e-05,
+      "loss": 0.1038,
+      "step": 4251
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1795402944491512e-05,
+      "loss": 0.1155,
+      "step": 4252
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791961249256036e-05,
+      "loss": 0.1048,
+      "step": 4253
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1788519334701853e-05,
+      "loss": 0.1249,
+      "step": 4254
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1785077201250218e-05,
+      "loss": 0.1144,
+      "step": 4255
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781634849322412e-05,
+      "loss": 0.1178,
+      "step": 4256
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778192279339748e-05,
+      "loss": 0.1111,
+      "step": 4257
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1774749491723565e-05,
+      "loss": 0.1093,
+      "step": 4258
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771306486895224e-05,
+      "loss": 0.1203,
+      "step": 4259
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1767863265276118e-05,
+      "loss": 0.106,
+      "step": 4260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1764419827287661e-05,
+      "loss": 0.1116,
+      "step": 4261
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760976173351299e-05,
+      "loss": 0.1123,
+      "step": 4262
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17575323038885e-05,
+      "loss": 0.1143,
+      "step": 4263
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754088219320762e-05,
+      "loss": 0.1184,
+      "step": 4264
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1750643920069608e-05,
+      "loss": 0.1226,
+      "step": 4265
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747199406556584e-05,
+      "loss": 0.1076,
+      "step": 4266
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1743754679203269e-05,
+      "loss": 0.1045,
+      "step": 4267
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174030973843126e-05,
+      "loss": 0.1151,
+      "step": 4268
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736864584662187e-05,
+      "loss": 0.1106,
+      "step": 4269
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1733419218317701e-05,
+      "loss": 0.1057,
+      "step": 4270
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729973639819485e-05,
+      "loss": 0.1073,
+      "step": 4271
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1726527849589244e-05,
+      "loss": 0.1073,
+      "step": 4272
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172308184804871e-05,
+      "loss": 0.1217,
+      "step": 4273
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1719635635619637e-05,
+      "loss": 0.1198,
+      "step": 4274
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1716189212723808e-05,
+      "loss": 0.1136,
+      "step": 4275
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712742579783035e-05,
+      "loss": 0.1195,
+      "step": 4276
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1709295737219149e-05,
+      "loss": 0.1195,
+      "step": 4277
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1705848685454011e-05,
+      "loss": 0.117,
+      "step": 4278
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702401424909514e-05,
+      "loss": 0.1094,
+      "step": 4279
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698953956007557e-05,
+      "loss": 0.1168,
+      "step": 4280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1695506279170084e-05,
+      "loss": 0.1033,
+      "step": 4281
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692058394819056e-05,
+      "loss": 0.1188,
+      "step": 4282
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1688610303376462e-05,
+      "loss": 0.1258,
+      "step": 4283
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685162005264314e-05,
+      "loss": 0.105,
+      "step": 4284
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681713500904652e-05,
+      "loss": 0.1128,
+      "step": 4285
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678264790719535e-05,
+      "loss": 0.1221,
+      "step": 4286
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1674815875131056e-05,
+      "loss": 0.1171,
+      "step": 4287
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671366754561327e-05,
+      "loss": 0.1157,
+      "step": 4288
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667917429432484e-05,
+      "loss": 0.1209,
+      "step": 4289
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1664467900166699e-05,
+      "loss": 0.1123,
+      "step": 4290
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661018167186155e-05,
+      "loss": 0.1136,
+      "step": 4291
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1657568230913068e-05,
+      "loss": 0.1154,
+      "step": 4292
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654118091769672e-05,
+      "loss": 0.1093,
+      "step": 4293
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650667750178238e-05,
+      "loss": 0.1212,
+      "step": 4294
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1647217206561052e-05,
+      "loss": 0.1156,
+      "step": 4295
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1643766461340424e-05,
+      "loss": 0.1162,
+      "step": 4296
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640315514938698e-05,
+      "loss": 0.1179,
+      "step": 4297
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1636864367778226e-05,
+      "loss": 0.1168,
+      "step": 4298
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633413020281406e-05,
+      "loss": 0.1038,
+      "step": 4299
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629961472870643e-05,
+      "loss": 0.113,
+      "step": 4300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1626509725968376e-05,
+      "loss": 0.1166,
+      "step": 4301
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1623057779997062e-05,
+      "loss": 0.1076,
+      "step": 4302
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619605635379192e-05,
+      "loss": 0.1113,
+      "step": 4303
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1616153292537269e-05,
+      "loss": 0.1288,
+      "step": 4304
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1612700751893829e-05,
+      "loss": 0.1128,
+      "step": 4305
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.160924801387143e-05,
+      "loss": 0.1103,
+      "step": 4306
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605795078892652e-05,
+      "loss": 0.1184,
+      "step": 4307
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602341947380103e-05,
+      "loss": 0.1092,
+      "step": 4308
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159888861975641e-05,
+      "loss": 0.1222,
+      "step": 4309
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159543509644423e-05,
+      "loss": 0.1218,
+      "step": 4310
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591981377866234e-05,
+      "loss": 0.1105,
+      "step": 4311
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588527464445134e-05,
+      "loss": 0.1204,
+      "step": 4312
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1585073356603649e-05,
+      "loss": 0.1096,
+      "step": 4313
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581619054764529e-05,
+      "loss": 0.103,
+      "step": 4314
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1578164559350552e-05,
+      "loss": 0.126,
+      "step": 4315
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574709870784506e-05,
+      "loss": 0.1102,
+      "step": 4316
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.157125498948922e-05,
+      "loss": 0.1164,
+      "step": 4317
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567799915887532e-05,
+      "loss": 0.1008,
+      "step": 4318
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.1336,
+      "step": 4319
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1560889193456448e-05,
+      "loss": 0.1166,
+      "step": 4320
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1557433545472863e-05,
+      "loss": 0.1073,
+      "step": 4321
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553977706874484e-05,
+      "loss": 0.1261,
+      "step": 4322
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550521678084279e-05,
+      "loss": 0.1191,
+      "step": 4323
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1547065459525231e-05,
+      "loss": 0.1188,
+      "step": 4324
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543609051620348e-05,
+      "loss": 0.1032,
+      "step": 4325
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154015245479266e-05,
+      "loss": 0.1141,
+      "step": 4326
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536695669465223e-05,
+      "loss": 0.1166,
+      "step": 4327
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1533238696061112e-05,
+      "loss": 0.1103,
+      "step": 4328
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1529781535003425e-05,
+      "loss": 0.1141,
+      "step": 4329
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.152632418671529e-05,
+      "loss": 0.1289,
+      "step": 4330
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1522866651619849e-05,
+      "loss": 0.1029,
+      "step": 4331
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519408930140273e-05,
+      "loss": 0.1209,
+      "step": 4332
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515951022699753e-05,
+      "loss": 0.1207,
+      "step": 4333
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1512492929721503e-05,
+      "loss": 0.124,
+      "step": 4334
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509034651628761e-05,
+      "loss": 0.1273,
+      "step": 4335
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1505576188844784e-05,
+      "loss": 0.1085,
+      "step": 4336
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.150211754179286e-05,
+      "loss": 0.1152,
+      "step": 4337
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149865871089629e-05,
+      "loss": 0.1185,
+      "step": 4338
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.14951996965784e-05,
+      "loss": 0.113,
+      "step": 4339
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1491740499262546e-05,
+      "loss": 0.12,
+      "step": 4340
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488281119372092e-05,
+      "loss": 0.1002,
+      "step": 4341
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484821557330435e-05,
+      "loss": 0.1234,
+      "step": 4342
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1481361813560997e-05,
+      "loss": 0.114,
+      "step": 4343
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.147790188848721e-05,
+      "loss": 0.1153,
+      "step": 4344
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474441782532542e-05,
+      "loss": 0.1093,
+      "step": 4345
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1470981496120474e-05,
+      "loss": 0.1133,
+      "step": 4346
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.146752102967451e-05,
+      "loss": 0.1146,
+      "step": 4347
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464060383618182e-05,
+      "loss": 0.1101,
+      "step": 4348
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1460599558375033e-05,
+      "loss": 0.1192,
+      "step": 4349
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457138554368637e-05,
+      "loss": 0.1133,
+      "step": 4350
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453677372022588e-05,
+      "loss": 0.1129,
+      "step": 4351
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1450216011760505e-05,
+      "loss": 0.1118,
+      "step": 4352
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446754474006018e-05,
+      "loss": 0.1053,
+      "step": 4353
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443292759182789e-05,
+      "loss": 0.1084,
+      "step": 4354
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.14398308677145e-05,
+      "loss": 0.1339,
+      "step": 4355
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143636880002485e-05,
+      "loss": 0.1149,
+      "step": 4356
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432906556537563e-05,
+      "loss": 0.1084,
+      "step": 4357
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1429444137676391e-05,
+      "loss": 0.1088,
+      "step": 4358
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142598154386509e-05,
+      "loss": 0.1148,
+      "step": 4359
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422518775527453e-05,
+      "loss": 0.1133,
+      "step": 4360
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1419055833087289e-05,
+      "loss": 0.1313,
+      "step": 4361
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1415592716968429e-05,
+      "loss": 0.1033,
+      "step": 4362
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1412129427594722e-05,
+      "loss": 0.1141,
+      "step": 4363
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140866596539005e-05,
+      "loss": 0.1305,
+      "step": 4364
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1405202330778293e-05,
+      "loss": 0.122,
+      "step": 4365
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1401738524183378e-05,
+      "loss": 0.1213,
+      "step": 4366
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1398274546029238e-05,
+      "loss": 0.1237,
+      "step": 4367
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1394810396739828e-05,
+      "loss": 0.119,
+      "step": 4368
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.139134607673913e-05,
+      "loss": 0.1097,
+      "step": 4369
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1387881586451141e-05,
+      "loss": 0.1051,
+      "step": 4370
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384416926299882e-05,
+      "loss": 0.1218,
+      "step": 4371
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380952096709392e-05,
+      "loss": 0.1094,
+      "step": 4372
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1377487098103734e-05,
+      "loss": 0.1219,
+      "step": 4373
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137402193090699e-05,
+      "loss": 0.1135,
+      "step": 4374
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1370556595543265e-05,
+      "loss": 0.1171,
+      "step": 4375
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367091092436682e-05,
+      "loss": 0.1163,
+      "step": 4376
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136362542201138e-05,
+      "loss": 0.1188,
+      "step": 4377
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1360159584691529e-05,
+      "loss": 0.1094,
+      "step": 4378
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1356693580901311e-05,
+      "loss": 0.1025,
+      "step": 4379
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1353227411064936e-05,
+      "loss": 0.1188,
+      "step": 4380
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349761075606623e-05,
+      "loss": 0.1211,
+      "step": 4381
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346294574950622e-05,
+      "loss": 0.1028,
+      "step": 4382
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342827909521198e-05,
+      "loss": 0.1301,
+      "step": 4383
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339361079742636e-05,
+      "loss": 0.1052,
+      "step": 4384
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1335894086039248e-05,
+      "loss": 0.1188,
+      "step": 4385
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332426928835352e-05,
+      "loss": 0.114,
+      "step": 4386
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328959608555299e-05,
+      "loss": 0.1095,
+      "step": 4387
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1325492125623458e-05,
+      "loss": 0.1109,
+      "step": 4388
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322024480464211e-05,
+      "loss": 0.1175,
+      "step": 4389
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1318556673501966e-05,
+      "loss": 0.1123,
+      "step": 4390
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315088705161151e-05,
+      "loss": 0.1062,
+      "step": 4391
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1311620575866206e-05,
+      "loss": 0.1169,
+      "step": 4392
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13081522860416e-05,
+      "loss": 0.1258,
+      "step": 4393
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1304683836111823e-05,
+      "loss": 0.1128,
+      "step": 4394
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130121522650137e-05,
+      "loss": 0.1197,
+      "step": 4395
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129774645763477e-05,
+      "loss": 0.0938,
+      "step": 4396
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294277529936568e-05,
+      "loss": 0.1197,
+      "step": 4397
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1290808443831324e-05,
+      "loss": 0.1187,
+      "step": 4398
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1287339199743626e-05,
+      "loss": 0.1159,
+      "step": 4399
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128386979809807e-05,
+      "loss": 0.1155,
+      "step": 4400
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128040023931928e-05,
+      "loss": 0.1171,
+      "step": 4401
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12769305238319e-05,
+      "loss": 0.1213,
+      "step": 4402
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127346065206058e-05,
+      "loss": 0.1069,
+      "step": 4403
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269990624430005e-05,
+      "loss": 0.1154,
+      "step": 4404
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1266520441364872e-05,
+      "loss": 0.1232,
+      "step": 4405
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263050103289898e-05,
+      "loss": 0.1077,
+      "step": 4406
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259579610629823e-05,
+      "loss": 0.1185,
+      "step": 4407
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1256108963809393e-05,
+      "loss": 0.1083,
+      "step": 4408
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252638163253387e-05,
+      "loss": 0.113,
+      "step": 4409
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1249167209386594e-05,
+      "loss": 0.1242,
+      "step": 4410
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245696102633832e-05,
+      "loss": 0.1059,
+      "step": 4411
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1242224843419923e-05,
+      "loss": 0.1149,
+      "step": 4412
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123875343216972e-05,
+      "loss": 0.1051,
+      "step": 4413
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1235281869308088e-05,
+      "loss": 0.1184,
+      "step": 4414
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231810155259915e-05,
+      "loss": 0.1074,
+      "step": 4415
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1228338290450103e-05,
+      "loss": 0.1194,
+      "step": 4416
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1224866275303574e-05,
+      "loss": 0.0996,
+      "step": 4417
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221394110245271e-05,
+      "loss": 0.1131,
+      "step": 4418
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217921795700156e-05,
+      "loss": 0.1186,
+      "step": 4419
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.12144493320932e-05,
+      "loss": 0.1073,
+      "step": 4420
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1210976719849404e-05,
+      "loss": 0.1119,
+      "step": 4421
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207503959393784e-05,
+      "loss": 0.1113,
+      "step": 4422
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204031051151366e-05,
+      "loss": 0.1073,
+      "step": 4423
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1200557995547201e-05,
+      "loss": 0.1073,
+      "step": 4424
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1197084793006363e-05,
+      "loss": 0.1165,
+      "step": 4425
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1193611443953933e-05,
+      "loss": 0.1254,
+      "step": 4426
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119013794881502e-05,
+      "loss": 0.1321,
+      "step": 4427
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186664308014738e-05,
+      "loss": 0.1101,
+      "step": 4428
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1183190521978236e-05,
+      "loss": 0.1011,
+      "step": 4429
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1179716591130666e-05,
+      "loss": 0.122,
+      "step": 4430
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117624251589721e-05,
+      "loss": 0.1085,
+      "step": 4431
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172768296703056e-05,
+      "loss": 0.1143,
+      "step": 4432
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169293933973416e-05,
+      "loss": 0.1064,
+      "step": 4433
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165819428133516e-05,
+      "loss": 0.0985,
+      "step": 4434
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162344779608605e-05,
+      "loss": 0.1189,
+      "step": 4435
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1158869988823945e-05,
+      "loss": 0.1041,
+      "step": 4436
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155395056204818e-05,
+      "loss": 0.1173,
+      "step": 4437
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.115191998217652e-05,
+      "loss": 0.1084,
+      "step": 4438
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148444767164368e-05,
+      "loss": 0.1194,
+      "step": 4439
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1144969411593693e-05,
+      "loss": 0.1187,
+      "step": 4440
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141493915889847e-05,
+      "loss": 0.1103,
+      "step": 4441
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138018280478198e-05,
+      "loss": 0.1139,
+      "step": 4442
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134542505784127e-05,
+      "loss": 0.1135,
+      "step": 4443
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1131066592233037e-05,
+      "loss": 0.1142,
+      "step": 4444
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127590540250346e-05,
+      "loss": 0.1284,
+      "step": 4445
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124114350261487e-05,
+      "loss": 0.1014,
+      "step": 4446
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1120638022691917e-05,
+      "loss": 0.1193,
+      "step": 4447
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117161557967098e-05,
+      "loss": 0.1092,
+      "step": 4448
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113684956512523e-05,
+      "loss": 0.1039,
+      "step": 4449
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1110208218753686e-05,
+      "loss": 0.1206,
+      "step": 4450
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1106731345116113e-05,
+      "loss": 0.1041,
+      "step": 4451
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1103254336025338e-05,
+      "loss": 0.1113,
+      "step": 4452
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109977719190691e-05,
+      "loss": 0.1184,
+      "step": 4453
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096299913186407e-05,
+      "loss": 0.1049,
+      "step": 4454
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092822500289402e-05,
+      "loss": 0.1193,
+      "step": 4455
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089344953641503e-05,
+      "loss": 0.1228,
+      "step": 4456
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1085867273668329e-05,
+      "loss": 0.1,
+      "step": 4457
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108238946079551e-05,
+      "loss": 0.1232,
+      "step": 4458
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.10789115154487e-05,
+      "loss": 0.112,
+      "step": 4459
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1075433438053567e-05,
+      "loss": 0.1168,
+      "step": 4460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1071955229035792e-05,
+      "loss": 0.1059,
+      "step": 4461
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1068476888821074e-05,
+      "loss": 0.1189,
+      "step": 4462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064998417835127e-05,
+      "loss": 0.1079,
+      "step": 4463
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061519816503684e-05,
+      "loss": 0.109,
+      "step": 4464
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058041085252493e-05,
+      "loss": 0.1144,
+      "step": 4465
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054562224507314e-05,
+      "loss": 0.12,
+      "step": 4466
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105108323469393e-05,
+      "loss": 0.1162,
+      "step": 4467
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1047604116238133e-05,
+      "loss": 0.1179,
+      "step": 4468
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044124869565733e-05,
+      "loss": 0.1157,
+      "step": 4469
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040645495102556e-05,
+      "loss": 0.1176,
+      "step": 4470
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1037165993274445e-05,
+      "loss": 0.0992,
+      "step": 4471
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033686364507257e-05,
+      "loss": 0.1252,
+      "step": 4472
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030206609226868e-05,
+      "loss": 0.1205,
+      "step": 4473
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1026726727859159e-05,
+      "loss": 0.1159,
+      "step": 4474
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023246720830038e-05,
+      "loss": 0.1006,
+      "step": 4475
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019766588565429e-05,
+      "loss": 0.1232,
+      "step": 4476
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1016286331491257e-05,
+      "loss": 0.1055,
+      "step": 4477
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012805950033476e-05,
+      "loss": 0.1303,
+      "step": 4478
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1009325444618053e-05,
+      "loss": 0.1189,
+      "step": 4479
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005844815670969e-05,
+      "loss": 0.1033,
+      "step": 4480
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002364063618213e-05,
+      "loss": 0.1058,
+      "step": 4481
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.09988831888858e-05,
+      "loss": 0.1157,
+      "step": 4482
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995402191899758e-05,
+      "loss": 0.1017,
+      "step": 4483
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099192107308612e-05,
+      "loss": 0.1075,
+      "step": 4484
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0988439832870947e-05,
+      "loss": 0.1173,
+      "step": 4485
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0984958471680308e-05,
+      "loss": 0.1016,
+      "step": 4486
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981476989940288e-05,
+      "loss": 0.1132,
+      "step": 4487
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977995388076983e-05,
+      "loss": 0.1056,
+      "step": 4488
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0974513666516511e-05,
+      "loss": 0.1018,
+      "step": 4489
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971031825684998e-05,
+      "loss": 0.1155,
+      "step": 4490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967549866008593e-05,
+      "loss": 0.1063,
+      "step": 4491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964067787913453e-05,
+      "loss": 0.1038,
+      "step": 4492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960585591825746e-05,
+      "loss": 0.11,
+      "step": 4493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957103278171661e-05,
+      "loss": 0.1014,
+      "step": 4494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09536208473774e-05,
+      "loss": 0.1064,
+      "step": 4495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095013829986918e-05,
+      "loss": 0.1106,
+      "step": 4496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094665563607323e-05,
+      "loss": 0.1156,
+      "step": 4497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0943172856415798e-05,
+      "loss": 0.1151,
+      "step": 4498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939689961323135e-05,
+      "loss": 0.118,
+      "step": 4499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0936206951221517e-05,
+      "loss": 0.1208,
+      "step": 4500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932723826537233e-05,
+      "loss": 0.1156,
+      "step": 4501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929240587696582e-05,
+      "loss": 0.1104,
+      "step": 4502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0925757235125877e-05,
+      "loss": 0.131,
+      "step": 4503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922273769251453e-05,
+      "loss": 0.1002,
+      "step": 4504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918790190499643e-05,
+      "loss": 0.1092,
+      "step": 4505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091530649929681e-05,
+      "loss": 0.1105,
+      "step": 4506
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0911822696069322e-05,
+      "loss": 0.1084,
+      "step": 4507
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908338781243565e-05,
+      "loss": 0.1215,
+      "step": 4508
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0904854755245933e-05,
+      "loss": 0.1059,
+      "step": 4509
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090137061850284e-05,
+      "loss": 0.122,
+      "step": 4510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089788637144071e-05,
+      "loss": 0.1148,
+      "step": 4511
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894402014485977e-05,
+      "loss": 0.1196,
+      "step": 4512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0890917548065097e-05,
+      "loss": 0.1047,
+      "step": 4513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887432972604532e-05,
+      "loss": 0.1147,
+      "step": 4514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883948288530766e-05,
+      "loss": 0.1199,
+      "step": 4515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0880463496270286e-05,
+      "loss": 0.1287,
+      "step": 4516
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876978596249596e-05,
+      "loss": 0.1161,
+      "step": 4517
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0873493588895216e-05,
+      "loss": 0.1102,
+      "step": 4518
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087000847463368e-05,
+      "loss": 0.1115,
+      "step": 4519
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866523253891525e-05,
+      "loss": 0.1202,
+      "step": 4520
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0863037927095312e-05,
+      "loss": 0.102,
+      "step": 4521
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0859552494671614e-05,
+      "loss": 0.1046,
+      "step": 4522
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.085606695704701e-05,
+      "loss": 0.1292,
+      "step": 4523
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0852581314648101e-05,
+      "loss": 0.1024,
+      "step": 4524
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849095567901488e-05,
+      "loss": 0.1063,
+      "step": 4525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845609717233801e-05,
+      "loss": 0.1079,
+      "step": 4526
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0842123763071667e-05,
+      "loss": 0.0992,
+      "step": 4527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838637705841743e-05,
+      "loss": 0.1002,
+      "step": 4528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0835151545970679e-05,
+      "loss": 0.106,
+      "step": 4529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0831665283885151e-05,
+      "loss": 0.1152,
+      "step": 4530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828178920011843e-05,
+      "loss": 0.116,
+      "step": 4531
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824692454777452e-05,
+      "loss": 0.1065,
+      "step": 4532
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0821205888608688e-05,
+      "loss": 0.1066,
+      "step": 4533
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0817719221932275e-05,
+      "loss": 0.1133,
+      "step": 4534
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0814232455174945e-05,
+      "loss": 0.1085,
+      "step": 4535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0810745588763443e-05,
+      "loss": 0.1144,
+      "step": 4536
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.080725862312453e-05,
+      "loss": 0.112,
+      "step": 4537
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803771558684977e-05,
+      "loss": 0.1272,
+      "step": 4538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0800284395871568e-05,
+      "loss": 0.1131,
+      "step": 4539
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0796797135111093e-05,
+      "loss": 0.1184,
+      "step": 4540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0793309776830365e-05,
+      "loss": 0.1057,
+      "step": 4541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.07898223214562e-05,
+      "loss": 0.1206,
+      "step": 4542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786334769415427e-05,
+      "loss": 0.1162,
+      "step": 4543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782847121134891e-05,
+      "loss": 0.1127,
+      "step": 4544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779359377041444e-05,
+      "loss": 0.1223,
+      "step": 4545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0775871537561959e-05,
+      "loss": 0.1151,
+      "step": 4546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772383603123307e-05,
+      "loss": 0.104,
+      "step": 4547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0768895574152378e-05,
+      "loss": 0.1097,
+      "step": 4548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0765407451076072e-05,
+      "loss": 0.1212,
+      "step": 4549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076191923432131e-05,
+      "loss": 0.1174,
+      "step": 4550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0758430924315004e-05,
+      "loss": 0.098,
+      "step": 4551
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0754942521484097e-05,
+      "loss": 0.11,
+      "step": 4552
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0751454026255537e-05,
+      "loss": 0.1181,
+      "step": 4553
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0747965439056274e-05,
+      "loss": 0.1068,
+      "step": 4554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744476760313285e-05,
+      "loss": 0.1178,
+      "step": 4555
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740987990453544e-05,
+      "loss": 0.1157,
+      "step": 4556
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073749912990405e-05,
+      "loss": 0.1148,
+      "step": 4557
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734010179091798e-05,
+      "loss": 0.1127,
+      "step": 4558
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730521138443812e-05,
+      "loss": 0.1042,
+      "step": 4559
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0727032008387107e-05,
+      "loss": 0.1139,
+      "step": 4560
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723542789348724e-05,
+      "loss": 0.1167,
+      "step": 4561
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720053481755705e-05,
+      "loss": 0.1027,
+      "step": 4562
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0716564086035113e-05,
+      "loss": 0.1228,
+      "step": 4563
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713074602614012e-05,
+      "loss": 0.1162,
+      "step": 4564
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0709585031919485e-05,
+      "loss": 0.1165,
+      "step": 4565
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0706095374378617e-05,
+      "loss": 0.1213,
+      "step": 4566
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702605630418511e-05,
+      "loss": 0.1083,
+      "step": 4567
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069911580046628e-05,
+      "loss": 0.1231,
+      "step": 4568
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0695625884949039e-05,
+      "loss": 0.1301,
+      "step": 4569
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692135884293926e-05,
+      "loss": 0.116,
+      "step": 4570
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688645798928084e-05,
+      "loss": 0.1071,
+      "step": 4571
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0685155629278663e-05,
+      "loss": 0.1144,
+      "step": 4572
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.068166537577282e-05,
+      "loss": 0.1221,
+      "step": 4573
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0678175038837737e-05,
+      "loss": 0.1226,
+      "step": 4574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674684618900595e-05,
+      "loss": 0.1207,
+      "step": 4575
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671194116388586e-05,
+      "loss": 0.1049,
+      "step": 4576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667703531728916e-05,
+      "loss": 0.112,
+      "step": 4577
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0664212865348798e-05,
+      "loss": 0.1183,
+      "step": 4578
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0660722117675453e-05,
+      "loss": 0.1168,
+      "step": 4579
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0657231289136118e-05,
+      "loss": 0.1185,
+      "step": 4580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653740380158038e-05,
+      "loss": 0.1212,
+      "step": 4581
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650249391168464e-05,
+      "loss": 0.1205,
+      "step": 4582
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646758322594657e-05,
+      "loss": 0.1074,
+      "step": 4583
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643267174863894e-05,
+      "loss": 0.1142,
+      "step": 4584
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0639775948403454e-05,
+      "loss": 0.1127,
+      "step": 4585
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0636284643640633e-05,
+      "loss": 0.1177,
+      "step": 4586
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063279326100273e-05,
+      "loss": 0.1213,
+      "step": 4587
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629301800917058e-05,
+      "loss": 0.1162,
+      "step": 4588
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062581026381094e-05,
+      "loss": 0.1113,
+      "step": 4589
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0622318650111703e-05,
+      "loss": 0.114,
+      "step": 4590
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618826960246687e-05,
+      "loss": 0.1181,
+      "step": 4591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0615335194643242e-05,
+      "loss": 0.1015,
+      "step": 4592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611843353728726e-05,
+      "loss": 0.1152,
+      "step": 4593
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0608351437930507e-05,
+      "loss": 0.1006,
+      "step": 4594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604859447675964e-05,
+      "loss": 0.1085,
+      "step": 4595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0601367383392475e-05,
+      "loss": 0.1113,
+      "step": 4596
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0597875245507444e-05,
+      "loss": 0.1046,
+      "step": 4597
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594383034448266e-05,
+      "loss": 0.1115,
+      "step": 4598
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0590890750642361e-05,
+      "loss": 0.1138,
+      "step": 4599
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587398394517148e-05,
+      "loss": 0.109,
+      "step": 4600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058390596650006e-05,
+      "loss": 0.0954,
+      "step": 4601
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0580413467018532e-05,
+      "loss": 0.1177,
+      "step": 4602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0576920896500019e-05,
+      "loss": 0.104,
+      "step": 4603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057342825537197e-05,
+      "loss": 0.1143,
+      "step": 4604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569935544061853e-05,
+      "loss": 0.1071,
+      "step": 4605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566442762997144e-05,
+      "loss": 0.1263,
+      "step": 4606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562949912605324e-05,
+      "loss": 0.111,
+      "step": 4607
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0559456993313886e-05,
+      "loss": 0.1097,
+      "step": 4608
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555964005550326e-05,
+      "loss": 0.1145,
+      "step": 4609
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0552470949742152e-05,
+      "loss": 0.1132,
+      "step": 4610
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0548977826316883e-05,
+      "loss": 0.1091,
+      "step": 4611
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545484635702046e-05,
+      "loss": 0.1217,
+      "step": 4612
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541991378325169e-05,
+      "loss": 0.1176,
+      "step": 4613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0538498054613793e-05,
+      "loss": 0.106,
+      "step": 4614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0535004664995471e-05,
+      "loss": 0.1141,
+      "step": 4615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0531511209897753e-05,
+      "loss": 0.1257,
+      "step": 4616
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052801768974821e-05,
+      "loss": 0.1236,
+      "step": 4617
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524524104974414e-05,
+      "loss": 0.1232,
+      "step": 4618
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0521030456003946e-05,
+      "loss": 0.1149,
+      "step": 4619
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0517536743264393e-05,
+      "loss": 0.1189,
+      "step": 4620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0514042967183354e-05,
+      "loss": 0.1085,
+      "step": 4621
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051054912818843e-05,
+      "loss": 0.1101,
+      "step": 4622
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507055226707235e-05,
+      "loss": 0.1007,
+      "step": 4623
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503561263167393e-05,
+      "loss": 0.1185,
+      "step": 4624
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500067237996523e-05,
+      "loss": 0.1103,
+      "step": 4625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0496573151622267e-05,
+      "loss": 0.1146,
+      "step": 4626
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0493079004472262e-05,
+      "loss": 0.1244,
+      "step": 4627
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0489584796974162e-05,
+      "loss": 0.1148,
+      "step": 4628
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486090529555619e-05,
+      "loss": 0.1285,
+      "step": 4629
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482596202644304e-05,
+      "loss": 0.1087,
+      "step": 4630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0479101816667885e-05,
+      "loss": 0.1039,
+      "step": 4631
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0475607372054043e-05,
+      "loss": 0.1177,
+      "step": 4632
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472112869230462e-05,
+      "loss": 0.1151,
+      "step": 4633
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468618308624834e-05,
+      "loss": 0.1213,
+      "step": 4634
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465123690664867e-05,
+      "loss": 0.1149,
+      "step": 4635
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461629015778258e-05,
+      "loss": 0.1221,
+      "step": 4636
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0458134284392727e-05,
+      "loss": 0.1181,
+      "step": 4637
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454639496935997e-05,
+      "loss": 0.1143,
+      "step": 4638
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0451144653835792e-05,
+      "loss": 0.107,
+      "step": 4639
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447649755519849e-05,
+      "loss": 0.1162,
+      "step": 4640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444154802415908e-05,
+      "loss": 0.1052,
+      "step": 4641
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440659794951721e-05,
+      "loss": 0.1154,
+      "step": 4642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437164733555041e-05,
+      "loss": 0.1103,
+      "step": 4643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043366961865363e-05,
+      "loss": 0.1049,
+      "step": 4644
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0430174450675256e-05,
+      "loss": 0.1123,
+      "step": 4645
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0426679230047695e-05,
+      "loss": 0.1173,
+      "step": 4646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423183957198724e-05,
+      "loss": 0.1129,
+      "step": 4647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0419688632556133e-05,
+      "loss": 0.1129,
+      "step": 4648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0416193256547717e-05,
+      "loss": 0.116,
+      "step": 4649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412697829601276e-05,
+      "loss": 0.1018,
+      "step": 4650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0409202352144616e-05,
+      "loss": 0.107,
+      "step": 4651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.040570682460555e-05,
+      "loss": 0.1176,
+      "step": 4652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402211247411895e-05,
+      "loss": 0.1188,
+      "step": 4653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398715620991477e-05,
+      "loss": 0.1087,
+      "step": 4654
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0395219945772125e-05,
+      "loss": 0.1076,
+      "step": 4655
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0391724222181683e-05,
+      "loss": 0.1186,
+      "step": 4656
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388228450647987e-05,
+      "loss": 0.1086,
+      "step": 4657
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0384732631598886e-05,
+      "loss": 0.1203,
+      "step": 4658
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381236765462234e-05,
+      "loss": 0.1164,
+      "step": 4659
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377740852665893e-05,
+      "loss": 0.1185,
+      "step": 4660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0374244893637728e-05,
+      "loss": 0.1196,
+      "step": 4661
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0370748888805612e-05,
+      "loss": 0.1139,
+      "step": 4662
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0367252838597421e-05,
+      "loss": 0.1157,
+      "step": 4663
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363756743441038e-05,
+      "loss": 0.1178,
+      "step": 4664
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.036026060376435e-05,
+      "loss": 0.109,
+      "step": 4665
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0356764419995253e-05,
+      "loss": 0.0964,
+      "step": 4666
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0353268192561645e-05,
+      "loss": 0.1082,
+      "step": 4667
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0349771921891432e-05,
+      "loss": 0.0919,
+      "step": 4668
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0346275608412517e-05,
+      "loss": 0.1107,
+      "step": 4669
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0342779252552822e-05,
+      "loss": 0.1079,
+      "step": 4670
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339282854740265e-05,
+      "loss": 0.1085,
+      "step": 4671
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033578641540277e-05,
+      "loss": 0.114,
+      "step": 4672
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033228993496827e-05,
+      "loss": 0.117,
+      "step": 4673
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.03287934138647e-05,
+      "loss": 0.1223,
+      "step": 4674
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0325296852519994e-05,
+      "loss": 0.1188,
+      "step": 4675
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0321800251362106e-05,
+      "loss": 0.1069,
+      "step": 4676
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318303610818983e-05,
+      "loss": 0.1119,
+      "step": 4677
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031480693131858e-05,
+      "loss": 0.1031,
+      "step": 4678
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0311310213288854e-05,
+      "loss": 0.1051,
+      "step": 4679
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0307813457157773e-05,
+      "loss": 0.1133,
+      "step": 4680
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0304316663353305e-05,
+      "loss": 0.1097,
+      "step": 4681
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0300819832303422e-05,
+      "loss": 0.12,
+      "step": 4682
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297322964436102e-05,
+      "loss": 0.1304,
+      "step": 4683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029382606017933e-05,
+      "loss": 0.1152,
+      "step": 4684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290329119961095e-05,
+      "loss": 0.1001,
+      "step": 4685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286832144209386e-05,
+      "loss": 0.1105,
+      "step": 4686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.02833351333522e-05,
+      "loss": 0.1126,
+      "step": 4687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0279838087817535e-05,
+      "loss": 0.1063,
+      "step": 4688
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276341008033398e-05,
+      "loss": 0.1068,
+      "step": 4689
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272843894427797e-05,
+      "loss": 0.1218,
+      "step": 4690
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0269346747428743e-05,
+      "loss": 0.1235,
+      "step": 4691
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265849567464256e-05,
+      "loss": 0.1164,
+      "step": 4692
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262352354962357e-05,
+      "loss": 0.0984,
+      "step": 4693
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0258855110351067e-05,
+      "loss": 0.1048,
+      "step": 4694
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255357834058418e-05,
+      "loss": 0.1095,
+      "step": 4695
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.025186052651244e-05,
+      "loss": 0.1142,
+      "step": 4696
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0248363188141175e-05,
+      "loss": 0.1111,
+      "step": 4697
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.024486581937266e-05,
+      "loss": 0.1065,
+      "step": 4698
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241368420634938e-05,
+      "loss": 0.1099,
+      "step": 4699
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237870992356061e-05,
+      "loss": 0.1298,
+      "step": 4700
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234373534964072e-05,
+      "loss": 0.1086,
+      "step": 4701
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0230876048887033e-05,
+      "loss": 0.1147,
+      "step": 4702
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227378534553e-05,
+      "loss": 0.1135,
+      "step": 4703
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223880992390038e-05,
+      "loss": 0.1101,
+      "step": 4704
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022038342282621e-05,
+      "loss": 0.1186,
+      "step": 4705
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0216885826289584e-05,
+      "loss": 0.1186,
+      "step": 4706
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0213388203208232e-05,
+      "loss": 0.1085,
+      "step": 4707
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0209890554010228e-05,
+      "loss": 0.102,
+      "step": 4708
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206392879123657e-05,
+      "loss": 0.1037,
+      "step": 4709
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202895178976591e-05,
+      "loss": 0.1008,
+      "step": 4710
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0199397453997121e-05,
+      "loss": 0.112,
+      "step": 4711
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0195899704613333e-05,
+      "loss": 0.1072,
+      "step": 4712
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192401931253317e-05,
+      "loss": 0.1293,
+      "step": 4713
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188904134345166e-05,
+      "loss": 0.1094,
+      "step": 4714
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.018540631431698e-05,
+      "loss": 0.1155,
+      "step": 4715
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181908471596852e-05,
+      "loss": 0.1179,
+      "step": 4716
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0178410606612894e-05,
+      "loss": 0.1264,
+      "step": 4717
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0174912719793202e-05,
+      "loss": 0.1154,
+      "step": 4718
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171414811565884e-05,
+      "loss": 0.1104,
+      "step": 4719
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016791688235906e-05,
+      "loss": 0.1116,
+      "step": 4720
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0164418932600829e-05,
+      "loss": 0.1022,
+      "step": 4721
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160920962719312e-05,
+      "loss": 0.1096,
+      "step": 4722
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015742297314263e-05,
+      "loss": 0.1229,
+      "step": 4723
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0153924964298897e-05,
+      "loss": 0.1164,
+      "step": 4724
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015042693661624e-05,
+      "loss": 0.1137,
+      "step": 4725
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0146928890522783e-05,
+      "loss": 0.1094,
+      "step": 4726
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143430826446649e-05,
+      "loss": 0.1026,
+      "step": 4727
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0139932744815973e-05,
+      "loss": 0.1042,
+      "step": 4728
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0136434646058883e-05,
+      "loss": 0.1058,
+      "step": 4729
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0132936530603515e-05,
+      "loss": 0.1077,
+      "step": 4730
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129438398878002e-05,
+      "loss": 0.1191,
+      "step": 4731
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.012594025131048e-05,
+      "loss": 0.1222,
+      "step": 4732
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122442088329092e-05,
+      "loss": 0.1068,
+      "step": 4733
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118943910361975e-05,
+      "loss": 0.1117,
+      "step": 4734
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011544571783728e-05,
+      "loss": 0.1164,
+      "step": 4735
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111947511183144e-05,
+      "loss": 0.1157,
+      "step": 4736
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0108449290827716e-05,
+      "loss": 0.1055,
+      "step": 4737
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104951057199144e-05,
+      "loss": 0.1186,
+      "step": 4738
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101452810725578e-05,
+      "loss": 0.1031,
+      "step": 4739
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097954551835172e-05,
+      "loss": 0.101,
+      "step": 4740
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0094456280956078e-05,
+      "loss": 0.1165,
+      "step": 4741
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090957998516448e-05,
+      "loss": 0.1255,
+      "step": 4742
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087459704944442e-05,
+      "loss": 0.1129,
+      "step": 4743
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083961400668213e-05,
+      "loss": 0.1184,
+      "step": 4744
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080463086115919e-05,
+      "loss": 0.1096,
+      "step": 4745
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076964761715725e-05,
+      "loss": 0.1246,
+      "step": 4746
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0073466427895788e-05,
+      "loss": 0.1082,
+      "step": 4747
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069968085084272e-05,
+      "loss": 0.1158,
+      "step": 4748
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0066469733709339e-05,
+      "loss": 0.114,
+      "step": 4749
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062971374199155e-05,
+      "loss": 0.1135,
+      "step": 4750
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059473006981883e-05,
+      "loss": 0.102,
+      "step": 4751
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055974632485694e-05,
+      "loss": 0.1146,
+      "step": 4752
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005247625113875e-05,
+      "loss": 0.1021,
+      "step": 4753
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004897786336922e-05,
+      "loss": 0.1089,
+      "step": 4754
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045479469605274e-05,
+      "loss": 0.0934,
+      "step": 4755
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041981070275082e-05,
+      "loss": 0.104,
+      "step": 4756
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038482665806812e-05,
+      "loss": 0.1194,
+      "step": 4757
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034984256628637e-05,
+      "loss": 0.1107,
+      "step": 4758
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003148584316873e-05,
+      "loss": 0.111,
+      "step": 4759
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0027987425855256e-05,
+      "loss": 0.0983,
+      "step": 4760
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0024489005116396e-05,
+      "loss": 0.1159,
+      "step": 4761
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020990581380315e-05,
+      "loss": 0.1093,
+      "step": 4762
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017492155075195e-05,
+      "loss": 0.1198,
+      "step": 4763
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0013993726629201e-05,
+      "loss": 0.1125,
+      "step": 4764
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0010495296470513e-05,
+      "loss": 0.1122,
+      "step": 4765
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0006996865027298e-05,
+      "loss": 0.1177,
+      "step": 4766
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003498432727738e-05,
+      "loss": 0.1118,
+      "step": 4767
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.0967,
+      "step": 4768
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996501567272267e-06,
+      "loss": 0.1075,
+      "step": 4769
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.993003134972703e-06,
+      "loss": 0.1064,
+      "step": 4770
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.989504703529493e-06,
+      "loss": 0.1019,
+      "step": 4771
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986006273370802e-06,
+      "loss": 0.1236,
+      "step": 4772
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98250784492481e-06,
+      "loss": 0.1184,
+      "step": 4773
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979009418619685e-06,
+      "loss": 0.1061,
+      "step": 4774
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975510994883609e-06,
+      "loss": 0.1035,
+      "step": 4775
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.972012574144747e-06,
+      "loss": 0.1111,
+      "step": 4776
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968514156831274e-06,
+      "loss": 0.1163,
+      "step": 4777
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965015743371368e-06,
+      "loss": 0.1172,
+      "step": 4778
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961517334193188e-06,
+      "loss": 0.1153,
+      "step": 4779
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958018929724921e-06,
+      "loss": 0.1124,
+      "step": 4780
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.95452053039473e-06,
+      "loss": 0.1002,
+      "step": 4781
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.951022136630782e-06,
+      "loss": 0.1111,
+      "step": 4782
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.947523748861255e-06,
+      "loss": 0.1027,
+      "step": 4783
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.94402536751431e-06,
+      "loss": 0.1213,
+      "step": 4784
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940526993018119e-06,
+      "loss": 0.1276,
+      "step": 4785
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937028625800847e-06,
+      "loss": 0.1007,
+      "step": 4786
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933530266290663e-06,
+      "loss": 0.1043,
+      "step": 4787
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930031914915733e-06,
+      "loss": 0.1226,
+      "step": 4788
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926533572104214e-06,
+      "loss": 0.1117,
+      "step": 4789
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.923035238284278e-06,
+      "loss": 0.108,
+      "step": 4790
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919536913884081e-06,
+      "loss": 0.1123,
+      "step": 4791
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91603859933179e-06,
+      "loss": 0.1068,
+      "step": 4792
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.912540295055563e-06,
+      "loss": 0.1101,
+      "step": 4793
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909042001483554e-06,
+      "loss": 0.1165,
+      "step": 4794
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.905543719043924e-06,
+      "loss": 0.1076,
+      "step": 4795
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902045448164828e-06,
+      "loss": 0.1017,
+      "step": 4796
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898547189274425e-06,
+      "loss": 0.1053,
+      "step": 4797
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895048942800856e-06,
+      "loss": 0.1133,
+      "step": 4798
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891550709172287e-06,
+      "loss": 0.1168,
+      "step": 4799
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.888052488816862e-06,
+      "loss": 0.1113,
+      "step": 4800
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.884554282162722e-06,
+      "loss": 0.1126,
+      "step": 4801
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881056089638026e-06,
+      "loss": 0.1088,
+      "step": 4802
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877557911670908e-06,
+      "loss": 0.1151,
+      "step": 4803
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874059748689522e-06,
+      "loss": 0.1116,
+      "step": 4804
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.870561601122003e-06,
+      "loss": 0.1184,
+      "step": 4805
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.867063469396489e-06,
+      "loss": 0.1156,
+      "step": 4806
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86356535394112e-06,
+      "loss": 0.1133,
+      "step": 4807
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860067255184027e-06,
+      "loss": 0.1148,
+      "step": 4808
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856569173553353e-06,
+      "loss": 0.1114,
+      "step": 4809
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853071109477219e-06,
+      "loss": 0.1128,
+      "step": 4810
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.849573063383763e-06,
+      "loss": 0.099,
+      "step": 4811
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.846075035701108e-06,
+      "loss": 0.1004,
+      "step": 4812
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842577026857373e-06,
+      "loss": 0.1116,
+      "step": 4813
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839079037280691e-06,
+      "loss": 0.1082,
+      "step": 4814
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.835581067399175e-06,
+      "loss": 0.1145,
+      "step": 4815
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832083117640946e-06,
+      "loss": 0.1072,
+      "step": 4816
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828585188434117e-06,
+      "loss": 0.1093,
+      "step": 4817
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.825087280206801e-06,
+      "loss": 0.1183,
+      "step": 4818
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.821589393387111e-06,
+      "loss": 0.1045,
+      "step": 4819
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818091528403148e-06,
+      "loss": 0.1049,
+      "step": 4820
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.814593685683024e-06,
+      "loss": 0.1129,
+      "step": 4821
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811095865654834e-06,
+      "loss": 0.1089,
+      "step": 4822
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807598068746687e-06,
+      "loss": 0.1182,
+      "step": 4823
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.804100295386672e-06,
+      "loss": 0.1122,
+      "step": 4824
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80060254600288e-06,
+      "loss": 0.1082,
+      "step": 4825
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797104821023412e-06,
+      "loss": 0.1106,
+      "step": 4826
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793607120876348e-06,
+      "loss": 0.1024,
+      "step": 4827
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.790109445989775e-06,
+      "loss": 0.1041,
+      "step": 4828
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786611796791775e-06,
+      "loss": 0.101,
+      "step": 4829
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.78311417371042e-06,
+      "loss": 0.1076,
+      "step": 4830
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.779616577173794e-06,
+      "loss": 0.1319,
+      "step": 4831
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776119007609962e-06,
+      "loss": 0.104,
+      "step": 4832
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.772621465447001e-06,
+      "loss": 0.1121,
+      "step": 4833
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769123951112967e-06,
+      "loss": 0.1159,
+      "step": 4834
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76562646503593e-06,
+      "loss": 0.1148,
+      "step": 4835
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762129007643946e-06,
+      "loss": 0.0989,
+      "step": 4836
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758631579365066e-06,
+      "loss": 0.1185,
+      "step": 4837
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755134180627342e-06,
+      "loss": 0.1081,
+      "step": 4838
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.751636811858827e-06,
+      "loss": 0.1183,
+      "step": 4839
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748139473487563e-06,
+      "loss": 0.1158,
+      "step": 4840
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744642165941584e-06,
+      "loss": 0.1063,
+      "step": 4841
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.741144889648938e-06,
+      "loss": 0.1143,
+      "step": 4842
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.73764764503765e-06,
+      "loss": 0.1194,
+      "step": 4843
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734150432535745e-06,
+      "loss": 0.1128,
+      "step": 4844
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.730653252571259e-06,
+      "loss": 0.1074,
+      "step": 4845
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727156105572206e-06,
+      "loss": 0.1033,
+      "step": 4846
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723658991966603e-06,
+      "loss": 0.1156,
+      "step": 4847
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.720161912182468e-06,
+      "loss": 0.1175,
+      "step": 4848
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.716664866647804e-06,
+      "loss": 0.1039,
+      "step": 4849
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.713167855790618e-06,
+      "loss": 0.1108,
+      "step": 4850
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709670880038907e-06,
+      "loss": 0.1124,
+      "step": 4851
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706173939820672e-06,
+      "loss": 0.1066,
+      "step": 4852
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.7026770355639e-06,
+      "loss": 0.1201,
+      "step": 4853
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.699180167696583e-06,
+      "loss": 0.1156,
+      "step": 4854
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.695683336646702e-06,
+      "loss": 0.125,
+      "step": 4855
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.69218654284223e-06,
+      "loss": 0.0968,
+      "step": 4856
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68868978671115e-06,
+      "loss": 0.1014,
+      "step": 4857
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685193068681423e-06,
+      "loss": 0.1173,
+      "step": 4858
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68169638918102e-06,
+      "loss": 0.1102,
+      "step": 4859
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.678199748637896e-06,
+      "loss": 0.1012,
+      "step": 4860
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.674703147480007e-06,
+      "loss": 0.1114,
+      "step": 4861
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.671206586135307e-06,
+      "loss": 0.1137,
+      "step": 4862
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667710065031732e-06,
+      "loss": 0.1038,
+      "step": 4863
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664213584597233e-06,
+      "loss": 0.1064,
+      "step": 4864
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660717145259735e-06,
+      "loss": 0.1114,
+      "step": 4865
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65722074744718e-06,
+      "loss": 0.1009,
+      "step": 4866
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.653724391587486e-06,
+      "loss": 0.1065,
+      "step": 4867
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.650228078108572e-06,
+      "loss": 0.1131,
+      "step": 4868
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646731807438358e-06,
+      "loss": 0.1193,
+      "step": 4869
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.643235580004749e-06,
+      "loss": 0.107,
+      "step": 4870
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.639739396235652e-06,
+      "loss": 0.1092,
+      "step": 4871
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636243256558967e-06,
+      "loss": 0.1169,
+      "step": 4872
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.63274716140258e-06,
+      "loss": 0.1124,
+      "step": 4873
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.629251111194393e-06,
+      "loss": 0.1094,
+      "step": 4874
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.625755106362272e-06,
+      "loss": 0.1142,
+      "step": 4875
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.622259147334109e-06,
+      "loss": 0.1147,
+      "step": 4876
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618763234537767e-06,
+      "loss": 0.1068,
+      "step": 4877
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.615267368401118e-06,
+      "loss": 0.1038,
+      "step": 4878
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611771549352018e-06,
+      "loss": 0.1171,
+      "step": 4879
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.608275777818318e-06,
+      "loss": 0.1065,
+      "step": 4880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604780054227876e-06,
+      "loss": 0.109,
+      "step": 4881
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601284379008525e-06,
+      "loss": 0.1131,
+      "step": 4882
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597788752588108e-06,
+      "loss": 0.111,
+      "step": 4883
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594293175394455e-06,
+      "loss": 0.1071,
+      "step": 4884
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.590797647855386e-06,
+      "loss": 0.1152,
+      "step": 4885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587302170398727e-06,
+      "loss": 0.1095,
+      "step": 4886
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.583806743452283e-06,
+      "loss": 0.1117,
+      "step": 4887
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58031136744387e-06,
+      "loss": 0.1187,
+      "step": 4888
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.57681604280128e-06,
+      "loss": 0.1073,
+      "step": 4889
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.573320769952307e-06,
+      "loss": 0.118,
+      "step": 4890
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.569825549324747e-06,
+      "loss": 0.1096,
+      "step": 4891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.566330381346371e-06,
+      "loss": 0.116,
+      "step": 4892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.562835266444962e-06,
+      "loss": 0.107,
+      "step": 4893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.559340205048278e-06,
+      "loss": 0.1201,
+      "step": 4894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555845197584093e-06,
+      "loss": 0.1237,
+      "step": 4895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552350244480155e-06,
+      "loss": 0.1036,
+      "step": 4896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54885534616421e-06,
+      "loss": 0.1125,
+      "step": 4897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545360503064007e-06,
+      "loss": 0.1107,
+      "step": 4898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.541865715607273e-06,
+      "loss": 0.1112,
+      "step": 4899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538370984221745e-06,
+      "loss": 0.1098,
+      "step": 4900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534876309335137e-06,
+      "loss": 0.1236,
+      "step": 4901
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531381691375168e-06,
+      "loss": 0.1152,
+      "step": 4902
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527887130769544e-06,
+      "loss": 0.1225,
+      "step": 4903
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.524392627945959e-06,
+      "loss": 0.0982,
+      "step": 4904
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52089818333212e-06,
+      "loss": 0.1014,
+      "step": 4905
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517403797355696e-06,
+      "loss": 0.1028,
+      "step": 4906
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513909470444384e-06,
+      "loss": 0.1094,
+      "step": 4907
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.510415203025844e-06,
+      "loss": 0.1189,
+      "step": 4908
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.50692099552774e-06,
+      "loss": 0.0965,
+      "step": 4909
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503426848377738e-06,
+      "loss": 0.1136,
+      "step": 4910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499932762003482e-06,
+      "loss": 0.1051,
+      "step": 4911
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496438736832612e-06,
+      "loss": 0.1074,
+      "step": 4912
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.492944773292767e-06,
+      "loss": 0.1118,
+      "step": 4913
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.489450871811574e-06,
+      "loss": 0.1119,
+      "step": 4914
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.485957032816653e-06,
+      "loss": 0.1057,
+      "step": 4915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48246325673561e-06,
+      "loss": 0.1088,
+      "step": 4916
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.478969543996059e-06,
+      "loss": 0.1086,
+      "step": 4917
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475475895025586e-06,
+      "loss": 0.1095,
+      "step": 4918
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.471982310251791e-06,
+      "loss": 0.1133,
+      "step": 4919
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.468488790102246e-06,
+      "loss": 0.1095,
+      "step": 4920
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.464995335004532e-06,
+      "loss": 0.1187,
+      "step": 4921
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.461501945386209e-06,
+      "loss": 0.122,
+      "step": 4922
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458008621674834e-06,
+      "loss": 0.1122,
+      "step": 4923
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454515364297958e-06,
+      "loss": 0.1147,
+      "step": 4924
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.451022173683115e-06,
+      "loss": 0.1124,
+      "step": 4925
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44752905025785e-06,
+      "loss": 0.1101,
+      "step": 4926
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44403599444968e-06,
+      "loss": 0.1018,
+      "step": 4927
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.440543006686119e-06,
+      "loss": 0.1203,
+      "step": 4928
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43705008739468e-06,
+      "loss": 0.1001,
+      "step": 4929
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433557237002857e-06,
+      "loss": 0.11,
+      "step": 4930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43006445593815e-06,
+      "loss": 0.1063,
+      "step": 4931
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.426571744628032e-06,
+      "loss": 0.116,
+      "step": 4932
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423079103499986e-06,
+      "loss": 0.1128,
+      "step": 4933
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.41958653298147e-06,
+      "loss": 0.0964,
+      "step": 4934
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416094033499941e-06,
+      "loss": 0.0975,
+      "step": 4935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412601605482854e-06,
+      "loss": 0.1141,
+      "step": 4936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.40910924935764e-06,
+      "loss": 0.102,
+      "step": 4937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405616965551738e-06,
+      "loss": 0.1086,
+      "step": 4938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402124754492563e-06,
+      "loss": 0.1183,
+      "step": 4939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.398632616607528e-06,
+      "loss": 0.1145,
+      "step": 4940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395140552324042e-06,
+      "loss": 0.1072,
+      "step": 4941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.391648562069495e-06,
+      "loss": 0.1091,
+      "step": 4942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388156646271276e-06,
+      "loss": 0.1115,
+      "step": 4943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38466480535676e-06,
+      "loss": 0.1137,
+      "step": 4944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381173039753317e-06,
+      "loss": 0.1194,
+      "step": 4945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.377681349888302e-06,
+      "loss": 0.1111,
+      "step": 4946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374189736189063e-06,
+      "loss": 0.1127,
+      "step": 4947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370698199082946e-06,
+      "loss": 0.1018,
+      "step": 4948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.367206738997272e-06,
+      "loss": 0.1209,
+      "step": 4949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36371535635937e-06,
+      "loss": 0.1038,
+      "step": 4950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.36022405159655e-06,
+      "loss": 0.1133,
+      "step": 4951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.35673282513611e-06,
+      "loss": 0.1225,
+      "step": 4952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353241677405348e-06,
+      "loss": 0.1143,
+      "step": 4953
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.349750608831541e-06,
+      "loss": 0.1011,
+      "step": 4954
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346259619841967e-06,
+      "loss": 0.1075,
+      "step": 4955
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.342768710863882e-06,
+      "loss": 0.1143,
+      "step": 4956
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339277882324549e-06,
+      "loss": 0.111,
+      "step": 4957
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.335787134651207e-06,
+      "loss": 0.1079,
+      "step": 4958
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332296468271085e-06,
+      "loss": 0.1129,
+      "step": 4959
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328805883611418e-06,
+      "loss": 0.1334,
+      "step": 4960
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325315381099406e-06,
+      "loss": 0.1148,
+      "step": 4961
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.321824961162265e-06,
+      "loss": 0.1081,
+      "step": 4962
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318334624227183e-06,
+      "loss": 0.1118,
+      "step": 4963
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.314844370721342e-06,
+      "loss": 0.1043,
+      "step": 4964
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31135420107192e-06,
+      "loss": 0.1123,
+      "step": 4965
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307864115706074e-06,
+      "loss": 0.1149,
+      "step": 4966
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.304374115050963e-06,
+      "loss": 0.1176,
+      "step": 4967
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.300884199533722e-06,
+      "loss": 0.1098,
+      "step": 4968
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29739436958149e-06,
+      "loss": 0.1127,
+      "step": 4969
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293904625621388e-06,
+      "loss": 0.1096,
+      "step": 4970
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290414968080518e-06,
+      "loss": 0.1178,
+      "step": 4971
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286925397385991e-06,
+      "loss": 0.1086,
+      "step": 4972
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.283435913964888e-06,
+      "loss": 0.1145,
+      "step": 4973
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279946518244296e-06,
+      "loss": 0.1109,
+      "step": 4974
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276457210651282e-06,
+      "loss": 0.1073,
+      "step": 4975
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.272967991612897e-06,
+      "loss": 0.1045,
+      "step": 4976
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269478861556194e-06,
+      "loss": 0.0986,
+      "step": 4977
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.2659898209082e-06,
+      "loss": 0.1106,
+      "step": 4978
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.262500870095953e-06,
+      "loss": 0.1173,
+      "step": 4979
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259012009546454e-06,
+      "loss": 0.122,
+      "step": 4980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255523239686718e-06,
+      "loss": 0.1204,
+      "step": 4981
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.25203456094373e-06,
+      "loss": 0.1139,
+      "step": 4982
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.248545973744468e-06,
+      "loss": 0.1069,
+      "step": 4983
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.245057478515905e-06,
+      "loss": 0.1061,
+      "step": 4984
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.241569075684998e-06,
+      "loss": 0.1045,
+      "step": 4985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238080765678693e-06,
+      "loss": 0.1095,
+      "step": 4986
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234592548923928e-06,
+      "loss": 0.1034,
+      "step": 4987
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.231104425847625e-06,
+      "loss": 0.1099,
+      "step": 4988
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227616396876698e-06,
+      "loss": 0.1158,
+      "step": 4989
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.224128462438043e-06,
+      "loss": 0.1069,
+      "step": 4990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.220640622958557e-06,
+      "loss": 0.1247,
+      "step": 4991
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21715287886511e-06,
+      "loss": 0.1068,
+      "step": 4992
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.213665230584576e-06,
+      "loss": 0.1162,
+      "step": 4993
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210177678543805e-06,
+      "loss": 0.1061,
+      "step": 4994
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.206690223169638e-06,
+      "loss": 0.1064,
+      "step": 4995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.203202864888909e-06,
+      "loss": 0.1171,
+      "step": 4996
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199715604128437e-06,
+      "loss": 0.1052,
+      "step": 4997
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196228441315028e-06,
+      "loss": 0.1074,
+      "step": 4998
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192741376875471e-06,
+      "loss": 0.1071,
+      "step": 4999
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18925441123656e-06,
+      "loss": 0.104,
+      "step": 5000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185767544825062e-06,
+      "loss": 0.1171,
+      "step": 5001
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.182280778067727e-06,
+      "loss": 0.1087,
+      "step": 5002
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.178794111391315e-06,
+      "loss": 0.1234,
+      "step": 5003
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17530754522255e-06,
+      "loss": 0.1047,
+      "step": 5004
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17182107998816e-06,
+      "loss": 0.1015,
+      "step": 5005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.168334716114854e-06,
+      "loss": 0.1095,
+      "step": 5006
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.164848454029325e-06,
+      "loss": 0.1148,
+      "step": 5007
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16136229415826e-06,
+      "loss": 0.1058,
+      "step": 5008
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.157876236928331e-06,
+      "loss": 0.1135,
+      "step": 5009
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154390282766202e-06,
+      "loss": 0.108,
+      "step": 5010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150904432098512e-06,
+      "loss": 0.1026,
+      "step": 5011
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.147418685351902e-06,
+      "loss": 0.1098,
+      "step": 5012
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.143933042952993e-06,
+      "loss": 0.1228,
+      "step": 5013
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.140447505328388e-06,
+      "loss": 0.1051,
+      "step": 5014
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.13696207290469e-06,
+      "loss": 0.0996,
+      "step": 5015
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133476746108476e-06,
+      "loss": 0.1252,
+      "step": 5016
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129991525366323e-06,
+      "loss": 0.1084,
+      "step": 5017
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.126506411104786e-06,
+      "loss": 0.1065,
+      "step": 5018
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.123021403750407e-06,
+      "loss": 0.1099,
+      "step": 5019
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11953650372972e-06,
+      "loss": 0.1113,
+      "step": 5020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116051711469236e-06,
+      "loss": 0.0936,
+      "step": 5021
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112567027395471e-06,
+      "loss": 0.1071,
+      "step": 5022
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.109082451934905e-06,
+      "loss": 0.1058,
+      "step": 5023
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105597985514027e-06,
+      "loss": 0.1092,
+      "step": 5024
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102113628559297e-06,
+      "loss": 0.1128,
+      "step": 5025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098629381497162e-06,
+      "loss": 0.1017,
+      "step": 5026
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.09514524475407e-06,
+      "loss": 0.1245,
+      "step": 5027
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091661218756438e-06,
+      "loss": 0.1148,
+      "step": 5028
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.088177303930681e-06,
+      "loss": 0.1026,
+      "step": 5029
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.084693500703193e-06,
+      "loss": 0.1073,
+      "step": 5030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08120980950036e-06,
+      "loss": 0.1136,
+      "step": 5031
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.077726230748554e-06,
+      "loss": 0.109,
+      "step": 5032
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.074242764874124e-06,
+      "loss": 0.1115,
+      "step": 5033
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070759412303422e-06,
+      "loss": 0.1098,
+      "step": 5034
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067276173462767e-06,
+      "loss": 0.1165,
+      "step": 5035
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063793048778485e-06,
+      "loss": 0.1159,
+      "step": 5036
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06031003867687e-06,
+      "loss": 0.1087,
+      "step": 5037
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056827143584206e-06,
+      "loss": 0.1082,
+      "step": 5038
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053344363926771e-06,
+      "loss": 0.106,
+      "step": 5039
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049861700130822e-06,
+      "loss": 0.1111,
+      "step": 5040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.046379152622602e-06,
+      "loss": 0.1188,
+      "step": 5041
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.042896721828345e-06,
+      "loss": 0.1238,
+      "step": 5042
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039414408174258e-06,
+      "loss": 0.1135,
+      "step": 5043
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035932212086552e-06,
+      "loss": 0.1132,
+      "step": 5044
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032450133991407e-06,
+      "loss": 0.1177,
+      "step": 5045
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028968174315003e-06,
+      "loss": 0.1024,
+      "step": 5046
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02548633348349e-06,
+      "loss": 0.1197,
+      "step": 5047
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02200461192302e-06,
+      "loss": 0.1114,
+      "step": 5048
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018523010059717e-06,
+      "loss": 0.1059,
+      "step": 5049
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015041528319696e-06,
+      "loss": 0.1149,
+      "step": 5050
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.011560167129054e-06,
+      "loss": 0.1147,
+      "step": 5051
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008078926913881e-06,
+      "loss": 0.1045,
+      "step": 5052
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.004597808100245e-06,
+      "loss": 0.1076,
+      "step": 5053
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001116811114203e-06,
+      "loss": 0.1059,
+      "step": 5054
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997635936381788e-06,
+      "loss": 0.1028,
+      "step": 5055
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994155184329036e-06,
+      "loss": 0.1117,
+      "step": 5056
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.990674555381947e-06,
+      "loss": 0.0996,
+      "step": 5057
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987194049966526e-06,
+      "loss": 0.1225,
+      "step": 5058
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.983713668508745e-06,
+      "loss": 0.1124,
+      "step": 5059
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980233411434575e-06,
+      "loss": 0.1018,
+      "step": 5060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976753279169964e-06,
+      "loss": 0.115,
+      "step": 5061
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.973273272140844e-06,
+      "loss": 0.1175,
+      "step": 5062
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969793390773139e-06,
+      "loss": 0.1194,
+      "step": 5063
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966313635492743e-06,
+      "loss": 0.1024,
+      "step": 5064
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.962834006725558e-06,
+      "loss": 0.1151,
+      "step": 5065
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959354504897444e-06,
+      "loss": 0.1194,
+      "step": 5066
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95587513043427e-06,
+      "loss": 0.1017,
+      "step": 5067
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.952395883761872e-06,
+      "loss": 0.1194,
+      "step": 5068
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94891676530607e-06,
+      "loss": 0.1164,
+      "step": 5069
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945437775492687e-06,
+      "loss": 0.1152,
+      "step": 5070
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941958914747509e-06,
+      "loss": 0.1099,
+      "step": 5071
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93848018349632e-06,
+      "loss": 0.1188,
+      "step": 5072
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935001582164877e-06,
+      "loss": 0.1131,
+      "step": 5073
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93152311117893e-06,
+      "loss": 0.1084,
+      "step": 5074
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.928044770964212e-06,
+      "loss": 0.1151,
+      "step": 5075
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.924566561946434e-06,
+      "loss": 0.1058,
+      "step": 5076
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.921088484551302e-06,
+      "loss": 0.1119,
+      "step": 5077
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91761053920449e-06,
+      "loss": 0.0976,
+      "step": 5078
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.914132726331675e-06,
+      "loss": 0.1222,
+      "step": 5079
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.9106550463585e-06,
+      "loss": 0.1048,
+      "step": 5080
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907177499710601e-06,
+      "loss": 0.1063,
+      "step": 5081
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903700086813597e-06,
+      "loss": 0.1126,
+      "step": 5082
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.90022280809309e-06,
+      "loss": 0.0973,
+      "step": 5083
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896745663974666e-06,
+      "loss": 0.1108,
+      "step": 5084
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.893268654883892e-06,
+      "loss": 0.1236,
+      "step": 5085
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.889791781246315e-06,
+      "loss": 0.102,
+      "step": 5086
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886315043487482e-06,
+      "loss": 0.121,
+      "step": 5087
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882838442032903e-06,
+      "loss": 0.1168,
+      "step": 5088
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.879361977308087e-06,
+      "loss": 0.1201,
+      "step": 5089
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875885649738513e-06,
+      "loss": 0.1122,
+      "step": 5090
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872409459749656e-06,
+      "loss": 0.113,
+      "step": 5091
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.868933407766966e-06,
+      "loss": 0.115,
+      "step": 5092
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865457494215875e-06,
+      "loss": 0.1236,
+      "step": 5093
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861981719521806e-06,
+      "loss": 0.0937,
+      "step": 5094
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858506084110153e-06,
+      "loss": 0.1176,
+      "step": 5095
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855030588406308e-06,
+      "loss": 0.1168,
+      "step": 5096
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851555232835637e-06,
+      "loss": 0.1109,
+      "step": 5097
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.848080017823482e-06,
+      "loss": 0.1146,
+      "step": 5098
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844604943795187e-06,
+      "loss": 0.1167,
+      "step": 5099
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.841130011176057e-06,
+      "loss": 0.1157,
+      "step": 5100
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.837655220391399e-06,
+      "loss": 0.116,
+      "step": 5101
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834180571866486e-06,
+      "loss": 0.1182,
+      "step": 5102
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830706066026587e-06,
+      "loss": 0.1075,
+      "step": 5103
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827231703296947e-06,
+      "loss": 0.1097,
+      "step": 5104
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.823757484102791e-06,
+      "loss": 0.113,
+      "step": 5105
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.820283408869335e-06,
+      "loss": 0.1107,
+      "step": 5106
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.816809478021764e-06,
+      "loss": 0.1132,
+      "step": 5107
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813335691985264e-06,
+      "loss": 0.1102,
+      "step": 5108
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809862051184988e-06,
+      "loss": 0.1135,
+      "step": 5109
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.806388556046069e-06,
+      "loss": 0.1051,
+      "step": 5110
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.802915206993642e-06,
+      "loss": 0.1232,
+      "step": 5111
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.799442004452799e-06,
+      "loss": 0.1134,
+      "step": 5112
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79596894884864e-06,
+      "loss": 0.1063,
+      "step": 5113
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792496040606219e-06,
+      "loss": 0.114,
+      "step": 5114
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.789023280150599e-06,
+      "loss": 0.1152,
+      "step": 5115
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.785550667906803e-06,
+      "loss": 0.1246,
+      "step": 5116
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782078204299846e-06,
+      "loss": 0.1062,
+      "step": 5117
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77860588975473e-06,
+      "loss": 0.1079,
+      "step": 5118
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775133724696426e-06,
+      "loss": 0.1126,
+      "step": 5119
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7716617095499e-06,
+      "loss": 0.0973,
+      "step": 5120
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76818984474009e-06,
+      "loss": 0.108,
+      "step": 5121
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764718130691914e-06,
+      "loss": 0.1084,
+      "step": 5122
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761246567830284e-06,
+      "loss": 0.106,
+      "step": 5123
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75777515658008e-06,
+      "loss": 0.1075,
+      "step": 5124
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754303897366173e-06,
+      "loss": 0.1094,
+      "step": 5125
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.750832790613406e-06,
+      "loss": 0.1094,
+      "step": 5126
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747361836746617e-06,
+      "loss": 0.1162,
+      "step": 5127
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.743891036190612e-06,
+      "loss": 0.1058,
+      "step": 5128
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740420389370182e-06,
+      "loss": 0.1085,
+      "step": 5129
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736949896710105e-06,
+      "loss": 0.1032,
+      "step": 5130
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.733479558635127e-06,
+      "loss": 0.0989,
+      "step": 5131
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730009375569998e-06,
+      "loss": 0.1152,
+      "step": 5132
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.726539347939422e-06,
+      "loss": 0.1113,
+      "step": 5133
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.723069476168104e-06,
+      "loss": 0.123,
+      "step": 5134
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719599760680721e-06,
+      "loss": 0.106,
+      "step": 5135
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716130201901931e-06,
+      "loss": 0.1199,
+      "step": 5136
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.712660800256379e-06,
+      "loss": 0.1118,
+      "step": 5137
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.709191556168675e-06,
+      "loss": 0.1129,
+      "step": 5138
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705722470063435e-06,
+      "loss": 0.1085,
+      "step": 5139
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702253542365235e-06,
+      "loss": 0.1084,
+      "step": 5140
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698784773498632e-06,
+      "loss": 0.1062,
+      "step": 5141
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695316163888183e-06,
+      "loss": 0.1144,
+      "step": 5142
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6918477139584e-06,
+      "loss": 0.1071,
+      "step": 5143
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688379424133797e-06,
+      "loss": 0.1007,
+      "step": 5144
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684911294838852e-06,
+      "loss": 0.1159,
+      "step": 5145
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.681443326498037e-06,
+      "loss": 0.118,
+      "step": 5146
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677975519535792e-06,
+      "loss": 0.1115,
+      "step": 5147
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.674507874376544e-06,
+      "loss": 0.1269,
+      "step": 5148
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671040391444704e-06,
+      "loss": 0.1124,
+      "step": 5149
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66757307116465e-06,
+      "loss": 0.111,
+      "step": 5150
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.664105913960757e-06,
+      "loss": 0.115,
+      "step": 5151
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660638920257367e-06,
+      "loss": 0.1018,
+      "step": 5152
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657172090478805e-06,
+      "loss": 0.1134,
+      "step": 5153
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653705425049382e-06,
+      "loss": 0.1088,
+      "step": 5154
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65023892439338e-06,
+      "loss": 0.0995,
+      "step": 5155
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646772588935068e-06,
+      "loss": 0.1088,
+      "step": 5156
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.643306419098689e-06,
+      "loss": 0.1082,
+      "step": 5157
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.639840415308475e-06,
+      "loss": 0.1026,
+      "step": 5158
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.636374577988625e-06,
+      "loss": 0.1114,
+      "step": 5159
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632908907563323e-06,
+      "loss": 0.1101,
+      "step": 5160
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.629443404456738e-06,
+      "loss": 0.1156,
+      "step": 5161
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.62597806909301e-06,
+      "loss": 0.1089,
+      "step": 5162
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622512901896268e-06,
+      "loss": 0.1124,
+      "step": 5163
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619047903290613e-06,
+      "loss": 0.1145,
+      "step": 5164
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.615583073700121e-06,
+      "loss": 0.1155,
+      "step": 5165
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.612118413548862e-06,
+      "loss": 0.1219,
+      "step": 5166
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.608653923260872e-06,
+      "loss": 0.116,
+      "step": 5167
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.605189603260175e-06,
+      "loss": 0.1109,
+      "step": 5168
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601725453970763e-06,
+      "loss": 0.1184,
+      "step": 5169
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.598261475816625e-06,
+      "loss": 0.1153,
+      "step": 5170
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59479766922171e-06,
+      "loss": 0.1047,
+      "step": 5171
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.591334034609954e-06,
+      "loss": 0.1138,
+      "step": 5172
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58787057240528e-06,
+      "loss": 0.1107,
+      "step": 5173
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584407283031571e-06,
+      "loss": 0.1142,
+      "step": 5174
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580944166912714e-06,
+      "loss": 0.1048,
+      "step": 5175
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577481224472552e-06,
+      "loss": 0.1159,
+      "step": 5176
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574018456134912e-06,
+      "loss": 0.1129,
+      "step": 5177
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.570555862323612e-06,
+      "loss": 0.1242,
+      "step": 5178
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567093443462437e-06,
+      "loss": 0.1105,
+      "step": 5179
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.563631199975153e-06,
+      "loss": 0.1064,
+      "step": 5180
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.560169132285501e-06,
+      "loss": 0.1153,
+      "step": 5181
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.556707240817213e-06,
+      "loss": 0.1004,
+      "step": 5182
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553245525993987e-06,
+      "loss": 0.1047,
+      "step": 5183
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549783988239497e-06,
+      "loss": 0.106,
+      "step": 5184
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.546322627977413e-06,
+      "loss": 0.1192,
+      "step": 5185
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542861445631363e-06,
+      "loss": 0.1185,
+      "step": 5186
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.539400441624969e-06,
+      "loss": 0.1123,
+      "step": 5187
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535939616381823e-06,
+      "loss": 0.1142,
+      "step": 5188
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532478970325493e-06,
+      "loss": 0.1226,
+      "step": 5189
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52901850387953e-06,
+      "loss": 0.1157,
+      "step": 5190
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525558217467458e-06,
+      "loss": 0.1126,
+      "step": 5191
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.522098111512791e-06,
+      "loss": 0.1109,
+      "step": 5192
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.518638186439005e-06,
+      "loss": 0.1163,
+      "step": 5193
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515178442669567e-06,
+      "loss": 0.1142,
+      "step": 5194
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511718880627915e-06,
+      "loss": 0.1154,
+      "step": 5195
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50825950073746e-06,
+      "loss": 0.114,
+      "step": 5196
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.504800303421603e-06,
+      "loss": 0.1056,
+      "step": 5197
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501341289103712e-06,
+      "loss": 0.1082,
+      "step": 5198
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.497882458207141e-06,
+      "loss": 0.1174,
+      "step": 5199
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.494423811155218e-06,
+      "loss": 0.1202,
+      "step": 5200
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.490965348371242e-06,
+      "loss": 0.1062,
+      "step": 5201
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.487507070278502e-06,
+      "loss": 0.1176,
+      "step": 5202
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48404897730025e-06,
+      "loss": 0.1121,
+      "step": 5203
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48059106985973e-06,
+      "loss": 0.1118,
+      "step": 5204
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.477133348380153e-06,
+      "loss": 0.1043,
+      "step": 5205
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.473675813284713e-06,
+      "loss": 0.128,
+      "step": 5206
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47021846499658e-06,
+      "loss": 0.1038,
+      "step": 5207
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466761303938891e-06,
+      "loss": 0.1104,
+      "step": 5208
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46330433053478e-06,
+      "loss": 0.1178,
+      "step": 5209
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.45984754520734e-06,
+      "loss": 0.1194,
+      "step": 5210
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456390948379655e-06,
+      "loss": 0.1074,
+      "step": 5211
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452934540474769e-06,
+      "loss": 0.1059,
+      "step": 5212
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.449478321915723e-06,
+      "loss": 0.117,
+      "step": 5213
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44602229312552e-06,
+      "loss": 0.1127,
+      "step": 5214
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44256645452714e-06,
+      "loss": 0.1141,
+      "step": 5215
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.439110806543555e-06,
+      "loss": 0.1176,
+      "step": 5216
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.1105,
+      "step": 5217
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432200084112473e-06,
+      "loss": 0.1031,
+      "step": 5218
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428745010510785e-06,
+      "loss": 0.1203,
+      "step": 5219
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425290129215498e-06,
+      "loss": 0.1148,
+      "step": 5220
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.421835440649453e-06,
+      "loss": 0.1159,
+      "step": 5221
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418380945235471e-06,
+      "loss": 0.1021,
+      "step": 5222
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.414926643396355e-06,
+      "loss": 0.109,
+      "step": 5223
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411472535554866e-06,
+      "loss": 0.1064,
+      "step": 5224
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408018622133767e-06,
+      "loss": 0.113,
+      "step": 5225
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404564903555777e-06,
+      "loss": 0.1041,
+      "step": 5226
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401111380243592e-06,
+      "loss": 0.1098,
+      "step": 5227
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397658052619902e-06,
+      "loss": 0.1175,
+      "step": 5228
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394204921107347e-06,
+      "loss": 0.1021,
+      "step": 5229
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.390751986128571e-06,
+      "loss": 0.1054,
+      "step": 5230
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.387299248106173e-06,
+      "loss": 0.1107,
+      "step": 5231
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.383846707462733e-06,
+      "loss": 0.1143,
+      "step": 5232
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380394364620811e-06,
+      "loss": 0.1258,
+      "step": 5233
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.376942220002936e-06,
+      "loss": 0.1045,
+      "step": 5234
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.373490274031628e-06,
+      "loss": 0.1172,
+      "step": 5235
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370038527129357e-06,
+      "loss": 0.1159,
+      "step": 5236
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366586979718596e-06,
+      "loss": 0.1114,
+      "step": 5237
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363135632221777e-06,
+      "loss": 0.1223,
+      "step": 5238
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359684485061305e-06,
+      "loss": 0.1064,
+      "step": 5239
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.356233538659578e-06,
+      "loss": 0.1132,
+      "step": 5240
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35278279343895e-06,
+      "loss": 0.1165,
+      "step": 5241
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349332249821764e-06,
+      "loss": 0.1058,
+      "step": 5242
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34588190823033e-06,
+      "loss": 0.1107,
+      "step": 5243
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.342431769086936e-06,
+      "loss": 0.1118,
+      "step": 5244
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.33898183281385e-06,
+      "loss": 0.1115,
+      "step": 5245
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.335532099833301e-06,
+      "loss": 0.1053,
+      "step": 5246
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332082570567517e-06,
+      "loss": 0.1153,
+      "step": 5247
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328633245438675e-06,
+      "loss": 0.1076,
+      "step": 5248
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.325184124868947e-06,
+      "loss": 0.1146,
+      "step": 5249
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321735209280468e-06,
+      "loss": 0.1145,
+      "step": 5250
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.31828649909535e-06,
+      "loss": 0.1073,
+      "step": 5251
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314837994735688e-06,
+      "loss": 0.1157,
+      "step": 5252
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311389696623538e-06,
+      "loss": 0.1005,
+      "step": 5253
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307941605180947e-06,
+      "loss": 0.1182,
+      "step": 5254
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.304493720829921e-06,
+      "loss": 0.1004,
+      "step": 5255
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301046043992447e-06,
+      "loss": 0.1055,
+      "step": 5256
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297598575090493e-06,
+      "loss": 0.1108,
+      "step": 5257
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.294151314545988e-06,
+      "loss": 0.1207,
+      "step": 5258
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.290704262780855e-06,
+      "loss": 0.1031,
+      "step": 5259
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287257420216967e-06,
+      "loss": 0.101,
+      "step": 5260
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.283810787276193e-06,
+      "loss": 0.1004,
+      "step": 5261
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.280364364380368e-06,
+      "loss": 0.1087,
+      "step": 5262
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276918151951295e-06,
+      "loss": 0.1118,
+      "step": 5263
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.273472150410759e-06,
+      "loss": 0.1151,
+      "step": 5264
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270026360180515e-06,
+      "loss": 0.0982,
+      "step": 5265
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.2665807816823e-06,
+      "loss": 0.11,
+      "step": 5266
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263135415337818e-06,
+      "loss": 0.116,
+      "step": 5267
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.259690261568742e-06,
+      "loss": 0.1128,
+      "step": 5268
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256245320796736e-06,
+      "loss": 0.1143,
+      "step": 5269
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252800593443417e-06,
+      "loss": 0.1079,
+      "step": 5270
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249356079930395e-06,
+      "loss": 0.1229,
+      "step": 5271
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24591178067924e-06,
+      "loss": 0.1049,
+      "step": 5272
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.242467696111501e-06,
+      "loss": 0.1153,
+      "step": 5273
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239023826648704e-06,
+      "loss": 0.1041,
+      "step": 5274
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23558017271234e-06,
+      "loss": 0.1088,
+      "step": 5275
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.232136734723887e-06,
+      "loss": 0.1132,
+      "step": 5276
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228693513104777e-06,
+      "loss": 0.1072,
+      "step": 5277
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225250508276439e-06,
+      "loss": 0.1181,
+      "step": 5278
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221807720660252e-06,
+      "loss": 0.1072,
+      "step": 5279
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218365150677592e-06,
+      "loss": 0.1092,
+      "step": 5280
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214922798749787e-06,
+      "loss": 0.1039,
+      "step": 5281
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.211480665298149e-06,
+      "loss": 0.1141,
+      "step": 5282
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208038750743967e-06,
+      "loss": 0.1101,
+      "step": 5283
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204597055508493e-06,
+      "loss": 0.1015,
+      "step": 5284
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.201155580012958e-06,
+      "loss": 0.118,
+      "step": 5285
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.197714324678567e-06,
+      "loss": 0.1208,
+      "step": 5286
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.194273289926493e-06,
+      "loss": 0.1032,
+      "step": 5287
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.19083247617789e-06,
+      "loss": 0.1311,
+      "step": 5288
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187391883853872e-06,
+      "loss": 0.122,
+      "step": 5289
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.183951513375547e-06,
+      "loss": 0.1084,
+      "step": 5290
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.180511365163972e-06,
+      "loss": 0.1062,
+      "step": 5291
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.177071439640194e-06,
+      "loss": 0.1058,
+      "step": 5292
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173631737225225e-06,
+      "loss": 0.1107,
+      "step": 5293
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.17019225834005e-06,
+      "loss": 0.1177,
+      "step": 5294
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.166753003405627e-06,
+      "loss": 0.1008,
+      "step": 5295
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16331397284289e-06,
+      "loss": 0.1133,
+      "step": 5296
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.159875167072744e-06,
+      "loss": 0.1158,
+      "step": 5297
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156436586516064e-06,
+      "loss": 0.104,
+      "step": 5298
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152998231593694e-06,
+      "loss": 0.0972,
+      "step": 5299
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149560102726466e-06,
+      "loss": 0.1081,
+      "step": 5300
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146122200335162e-06,
+      "loss": 0.1083,
+      "step": 5301
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142684524840558e-06,
+      "loss": 0.1051,
+      "step": 5302
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139247076663385e-06,
+      "loss": 0.1246,
+      "step": 5303
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13580985622436e-06,
+      "loss": 0.112,
+      "step": 5304
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13237286394416e-06,
+      "loss": 0.107,
+      "step": 5305
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12893610024344e-06,
+      "loss": 0.1116,
+      "step": 5306
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125499565542832e-06,
+      "loss": 0.114,
+      "step": 5307
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.122063260262926e-06,
+      "loss": 0.1031,
+      "step": 5308
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118627184824301e-06,
+      "loss": 0.1128,
+      "step": 5309
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.115191339647495e-06,
+      "loss": 0.1143,
+      "step": 5310
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111755725153018e-06,
+      "loss": 0.116,
+      "step": 5311
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.108320341761365e-06,
+      "loss": 0.1161,
+      "step": 5312
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.104885189892986e-06,
+      "loss": 0.1118,
+      "step": 5313
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.101450269968318e-06,
+      "loss": 0.1155,
+      "step": 5314
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.098015582407754e-06,
+      "loss": 0.1004,
+      "step": 5315
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094581127631673e-06,
+      "loss": 0.1083,
+      "step": 5316
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.091146906060414e-06,
+      "loss": 0.102,
+      "step": 5317
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.087712918114294e-06,
+      "loss": 0.1089,
+      "step": 5318
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084279164213604e-06,
+      "loss": 0.1037,
+      "step": 5319
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080845644778593e-06,
+      "loss": 0.119,
+      "step": 5320
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0774123602295e-06,
+      "loss": 0.1084,
+      "step": 5321
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073979310986524e-06,
+      "loss": 0.1057,
+      "step": 5322
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.07054649746983e-06,
+      "loss": 0.1013,
+      "step": 5323
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06711392009957e-06,
+      "loss": 0.1159,
+      "step": 5324
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06368157929585e-06,
+      "loss": 0.1159,
+      "step": 5325
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060249475478764e-06,
+      "loss": 0.1068,
+      "step": 5326
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056817609068362e-06,
+      "loss": 0.1082,
+      "step": 5327
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.053385980484676e-06,
+      "loss": 0.0995,
+      "step": 5328
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.0499545901477e-06,
+      "loss": 0.1087,
+      "step": 5329
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046523438477401e-06,
+      "loss": 0.1139,
+      "step": 5330
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.043092525893726e-06,
+      "loss": 0.1122,
+      "step": 5331
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039661852816578e-06,
+      "loss": 0.1052,
+      "step": 5332
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036231419665846e-06,
+      "loss": 0.1132,
+      "step": 5333
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.032801226861374e-06,
+      "loss": 0.1066,
+      "step": 5334
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.029371274822988e-06,
+      "loss": 0.1038,
+      "step": 5335
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.025941563970484e-06,
+      "loss": 0.1048,
+      "step": 5336
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022512094723618e-06,
+      "loss": 0.1096,
+      "step": 5337
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.019082867502132e-06,
+      "loss": 0.0975,
+      "step": 5338
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.015653882725721e-06,
+      "loss": 0.1103,
+      "step": 5339
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01222514081407e-06,
+      "loss": 0.1082,
+      "step": 5340
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008796642186817e-06,
+      "loss": 0.1108,
+      "step": 5341
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.005368387263577e-06,
+      "loss": 0.0996,
+      "step": 5342
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001940376463939e-06,
+      "loss": 0.1082,
+      "step": 5343
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.998512610207452e-06,
+      "loss": 0.1093,
+      "step": 5344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995085088913653e-06,
+      "loss": 0.1077,
+      "step": 5345
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991657813002026e-06,
+      "loss": 0.1058,
+      "step": 5346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988230782892038e-06,
+      "loss": 0.1082,
+      "step": 5347
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.984803999003131e-06,
+      "loss": 0.1163,
+      "step": 5348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981377461754704e-06,
+      "loss": 0.1087,
+      "step": 5349
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977951171566135e-06,
+      "loss": 0.1224,
+      "step": 5350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.974525128856764e-06,
+      "loss": 0.1038,
+      "step": 5351
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971099334045913e-06,
+      "loss": 0.1088,
+      "step": 5352
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967673787552863e-06,
+      "loss": 0.1074,
+      "step": 5353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964248489796861e-06,
+      "loss": 0.101,
+      "step": 5354
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.960823441197143e-06,
+      "loss": 0.1035,
+      "step": 5355
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.957398642172888e-06,
+      "loss": 0.1043,
+      "step": 5356
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.953974093143273e-06,
+      "loss": 0.1068,
+      "step": 5357
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950549794527418e-06,
+      "loss": 0.1113,
+      "step": 5358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94712574674443e-06,
+      "loss": 0.1115,
+      "step": 5359
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.943701950213374e-06,
+      "loss": 0.1056,
+      "step": 5360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940278405353294e-06,
+      "loss": 0.1136,
+      "step": 5361
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.9368551125832e-06,
+      "loss": 0.1011,
+      "step": 5362
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933432072322062e-06,
+      "loss": 0.1026,
+      "step": 5363
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930009284988838e-06,
+      "loss": 0.1068,
+      "step": 5364
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.926586751002438e-06,
+      "loss": 0.1089,
+      "step": 5365
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923164470781743e-06,
+      "loss": 0.1051,
+      "step": 5366
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919742444745614e-06,
+      "loss": 0.1147,
+      "step": 5367
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.916320673312872e-06,
+      "loss": 0.1078,
+      "step": 5368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912899156902307e-06,
+      "loss": 0.1108,
+      "step": 5369
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909477895932681e-06,
+      "loss": 0.1047,
+      "step": 5370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.906056890822724e-06,
+      "loss": 0.1075,
+      "step": 5371
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.902636141991133e-06,
+      "loss": 0.1088,
+      "step": 5372
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89921564985657e-06,
+      "loss": 0.0983,
+      "step": 5373
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895795414837679e-06,
+      "loss": 0.1132,
+      "step": 5374
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892375437353052e-06,
+      "loss": 0.1076,
+      "step": 5375
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888955717821275e-06,
+      "loss": 0.1011,
+      "step": 5376
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.885536256660883e-06,
+      "loss": 0.1065,
+      "step": 5377
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882117054290375e-06,
+      "loss": 0.11,
+      "step": 5378
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878698111128245e-06,
+      "loss": 0.109,
+      "step": 5379
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.875279427592926e-06,
+      "loss": 0.1003,
+      "step": 5380
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87186100410284e-06,
+      "loss": 0.1217,
+      "step": 5381
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86844284107636e-06,
+      "loss": 0.1047,
+      "step": 5382
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.865024938931845e-06,
+      "loss": 0.1061,
+      "step": 5383
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86160729808761e-06,
+      "loss": 0.1044,
+      "step": 5384
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858189918961935e-06,
+      "loss": 0.1036,
+      "step": 5385
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.854772801973085e-06,
+      "loss": 0.1091,
+      "step": 5386
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.851355947539272e-06,
+      "loss": 0.1204,
+      "step": 5387
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.847939356078693e-06,
+      "loss": 0.1064,
+      "step": 5388
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844523028009506e-06,
+      "loss": 0.1137,
+      "step": 5389
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841106963749832e-06,
+      "loss": 0.1074,
+      "step": 5390
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.837691163717764e-06,
+      "loss": 0.1156,
+      "step": 5391
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.834275628331365e-06,
+      "loss": 0.1224,
+      "step": 5392
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.830860358008665e-06,
+      "loss": 0.1086,
+      "step": 5393
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827445353167657e-06,
+      "loss": 0.1246,
+      "step": 5394
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824030614226306e-06,
+      "loss": 0.116,
+      "step": 5395
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820616141602546e-06,
+      "loss": 0.0984,
+      "step": 5396
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.817201935714268e-06,
+      "loss": 0.1089,
+      "step": 5397
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.813787996979343e-06,
+      "loss": 0.1127,
+      "step": 5398
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8103743258156e-06,
+      "loss": 0.1126,
+      "step": 5399
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806960922640848e-06,
+      "loss": 0.1097,
+      "step": 5400
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.803547787872849e-06,
+      "loss": 0.1014,
+      "step": 5401
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800134921929333e-06,
+      "loss": 0.1205,
+      "step": 5402
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79672232522801e-06,
+      "loss": 0.1098,
+      "step": 5403
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79330999818654e-06,
+      "loss": 0.1042,
+      "step": 5404
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.789897941222569e-06,
+      "loss": 0.1046,
+      "step": 5405
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78648615475369e-06,
+      "loss": 0.1188,
+      "step": 5406
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.783074639197482e-06,
+      "loss": 0.1066,
+      "step": 5407
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.779663394971475e-06,
+      "loss": 0.1131,
+      "step": 5408
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77625242249317e-06,
+      "loss": 0.1122,
+      "step": 5409
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772841722180047e-06,
+      "loss": 0.1176,
+      "step": 5410
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.769431294449532e-06,
+      "loss": 0.1155,
+      "step": 5411
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.766021139719034e-06,
+      "loss": 0.1083,
+      "step": 5412
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762611258405926e-06,
+      "loss": 0.1049,
+      "step": 5413
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759201650927537e-06,
+      "loss": 0.1149,
+      "step": 5414
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755792317701173e-06,
+      "loss": 0.1155,
+      "step": 5415
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.752383259144102e-06,
+      "loss": 0.1039,
+      "step": 5416
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748974475673565e-06,
+      "loss": 0.1238,
+      "step": 5417
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745565967706757e-06,
+      "loss": 0.1167,
+      "step": 5418
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.742157735660853e-06,
+      "loss": 0.1102,
+      "step": 5419
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738749779952986e-06,
+      "loss": 0.1054,
+      "step": 5420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73534210100025e-06,
+      "loss": 0.1055,
+      "step": 5421
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73193469921972e-06,
+      "loss": 0.0999,
+      "step": 5422
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728527575028428e-06,
+      "loss": 0.1007,
+      "step": 5423
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72512072884337e-06,
+      "loss": 0.1135,
+      "step": 5424
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72171416108151e-06,
+      "loss": 0.1083,
+      "step": 5425
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718307872159784e-06,
+      "loss": 0.1098,
+      "step": 5426
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.714901862495087e-06,
+      "loss": 0.1075,
+      "step": 5427
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.711496132504276e-06,
+      "loss": 0.1093,
+      "step": 5428
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708090682604188e-06,
+      "loss": 0.1068,
+      "step": 5429
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70468551321161e-06,
+      "loss": 0.1159,
+      "step": 5430
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.701280624743306e-06,
+      "loss": 0.1122,
+      "step": 5431
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.697876017616004e-06,
+      "loss": 0.1126,
+      "step": 5432
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.694471692246387e-06,
+      "loss": 0.1001,
+      "step": 5433
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.691067649051117e-06,
+      "loss": 0.1097,
+      "step": 5434
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687663888446812e-06,
+      "loss": 0.1171,
+      "step": 5435
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.684260410850066e-06,
+      "loss": 0.1068,
+      "step": 5436
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.680857216677422e-06,
+      "loss": 0.1063,
+      "step": 5437
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677454306345408e-06,
+      "loss": 0.1041,
+      "step": 5438
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.674051680270504e-06,
+      "loss": 0.111,
+      "step": 5439
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.670649338869154e-06,
+      "loss": 0.1228,
+      "step": 5440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667247282557777e-06,
+      "loss": 0.1064,
+      "step": 5441
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.663845511752749e-06,
+      "loss": 0.105,
+      "step": 5442
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660444026870415e-06,
+      "loss": 0.1153,
+      "step": 5443
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657042828327085e-06,
+      "loss": 0.1009,
+      "step": 5444
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.653641916539029e-06,
+      "loss": 0.1157,
+      "step": 5445
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65024129192249e-06,
+      "loss": 0.1061,
+      "step": 5446
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646840954893666e-06,
+      "loss": 0.0991,
+      "step": 5447
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.643440905868731e-06,
+      "loss": 0.1098,
+      "step": 5448
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.640041145263813e-06,
+      "loss": 0.1083,
+      "step": 5449
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636641673495018e-06,
+      "loss": 0.1153,
+      "step": 5450
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6332424909784e-06,
+      "loss": 0.1116,
+      "step": 5451
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.629843598129988e-06,
+      "loss": 0.1075,
+      "step": 5452
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626444995365778e-06,
+      "loss": 0.1143,
+      "step": 5453
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.62304668310172e-06,
+      "loss": 0.1016,
+      "step": 5454
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61964866175374e-06,
+      "loss": 0.112,
+      "step": 5455
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6162509317377185e-06,
+      "loss": 0.1134,
+      "step": 5456
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6128534934695054e-06,
+      "loss": 0.1082,
+      "step": 5457
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609456347364919e-06,
+      "loss": 0.1061,
+      "step": 5458
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.606059493839727e-06,
+      "loss": 0.108,
+      "step": 5459
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602662933309684e-06,
+      "loss": 0.1069,
+      "step": 5460
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599266666190484e-06,
+      "loss": 0.1,
+      "step": 5461
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.595870692897809e-06,
+      "loss": 0.1209,
+      "step": 5462
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.592475013847287e-06,
+      "loss": 0.1074,
+      "step": 5463
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589079629454513e-06,
+      "loss": 0.1064,
+      "step": 5464
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585684540135055e-06,
+      "loss": 0.1094,
+      "step": 5465
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.582289746304436e-06,
+      "loss": 0.119,
+      "step": 5466
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.578895248378149e-06,
+      "loss": 0.0999,
+      "step": 5467
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575501046771645e-06,
+      "loss": 0.1094,
+      "step": 5468
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.572107141900338e-06,
+      "loss": 0.1095,
+      "step": 5469
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.568713534179617e-06,
+      "loss": 0.1205,
+      "step": 5470
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.565320224024819e-06,
+      "loss": 0.1195,
+      "step": 5471
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56192721185126e-06,
+      "loss": 0.1034,
+      "step": 5472
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.558534498074204e-06,
+      "loss": 0.1075,
+      "step": 5473
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555142083108894e-06,
+      "loss": 0.1073,
+      "step": 5474
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.551749967370525e-06,
+      "loss": 0.1071,
+      "step": 5475
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.548358151274257e-06,
+      "loss": 0.1113,
+      "step": 5476
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54496663523522e-06,
+      "loss": 0.1055,
+      "step": 5477
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.541575419668497e-06,
+      "loss": 0.1186,
+      "step": 5478
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538184504989145e-06,
+      "loss": 0.107,
+      "step": 5479
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.53479389161218e-06,
+      "loss": 0.1098,
+      "step": 5480
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.531403579952572e-06,
+      "loss": 0.1059,
+      "step": 5481
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528013570425272e-06,
+      "loss": 0.1039,
+      "step": 5482
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5246238634451766e-06,
+      "loss": 0.1033,
+      "step": 5483
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52123445942716e-06,
+      "loss": 0.12,
+      "step": 5484
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.517845358786048e-06,
+      "loss": 0.1096,
+      "step": 5485
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514456561936635e-06,
+      "loss": 0.099,
+      "step": 5486
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.511068069293679e-06,
+      "loss": 0.1109,
+      "step": 5487
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5076798812718945e-06,
+      "loss": 0.1129,
+      "step": 5488
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504291998285966e-06,
+      "loss": 0.1029,
+      "step": 5489
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500904420750535e-06,
+      "loss": 0.1028,
+      "step": 5490
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4975171490802115e-06,
+      "loss": 0.1103,
+      "step": 5491
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.494130183689564e-06,
+      "loss": 0.1097,
+      "step": 5492
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490743524993119e-06,
+      "loss": 0.1074,
+      "step": 5493
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.48735717340538e-06,
+      "loss": 0.1209,
+      "step": 5494
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.483971129340792e-06,
+      "loss": 0.1142,
+      "step": 5495
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4805853932137884e-06,
+      "loss": 0.1146,
+      "step": 5496
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4771999654387384e-06,
+      "loss": 0.1107,
+      "step": 5497
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.473814846429993e-06,
+      "loss": 0.1038,
+      "step": 5498
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.470430036601855e-06,
+      "loss": 0.1096,
+      "step": 5499
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467045536368588e-06,
+      "loss": 0.1069,
+      "step": 5500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4636613461444305e-06,
+      "loss": 0.1025,
+      "step": 5501
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4602774663435665e-06,
+      "loss": 0.1111,
+      "step": 5502
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456893897380158e-06,
+      "loss": 0.1134,
+      "step": 5503
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453510639668313e-06,
+      "loss": 0.1013,
+      "step": 5504
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45012769362212e-06,
+      "loss": 0.1056,
+      "step": 5505
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.44674505965561e-06,
+      "loss": 0.106,
+      "step": 5506
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.443362738182787e-06,
+      "loss": 0.1011,
+      "step": 5507
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.439980729617613e-06,
+      "loss": 0.1055,
+      "step": 5508
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4365990343740165e-06,
+      "loss": 0.1131,
+      "step": 5509
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433217652865884e-06,
+      "loss": 0.1075,
+      "step": 5510
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.429836585507062e-06,
+      "loss": 0.1009,
+      "step": 5511
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426455832711357e-06,
+      "loss": 0.0987,
+      "step": 5512
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423075394892548e-06,
+      "loss": 0.1203,
+      "step": 5513
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41969527246436e-06,
+      "loss": 0.1086,
+      "step": 5514
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416315465840496e-06,
+      "loss": 0.1071,
+      "step": 5515
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.412935975434605e-06,
+      "loss": 0.1011,
+      "step": 5516
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.409556801660306e-06,
+      "loss": 0.116,
+      "step": 5517
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.406177944931179e-06,
+      "loss": 0.1056,
+      "step": 5518
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4027994056607585e-06,
+      "loss": 0.1024,
+      "step": 5519
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.399421184262551e-06,
+      "loss": 0.1154,
+      "step": 5520
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.396043281150009e-06,
+      "loss": 0.1175,
+      "step": 5521
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.392665696736568e-06,
+      "loss": 0.099,
+      "step": 5522
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3892884314356026e-06,
+      "loss": 0.1012,
+      "step": 5523
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385911485660456e-06,
+      "loss": 0.1189,
+      "step": 5524
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.382534859824442e-06,
+      "loss": 0.1244,
+      "step": 5525
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.379158554340817e-06,
+      "loss": 0.1152,
+      "step": 5526
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.375782569622818e-06,
+      "loss": 0.1056,
+      "step": 5527
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.372406906083627e-06,
+      "loss": 0.1083,
+      "step": 5528
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369031564136394e-06,
+      "loss": 0.1084,
+      "step": 5529
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365656544194226e-06,
+      "loss": 0.1074,
+      "step": 5530
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.362281846670196e-06,
+      "loss": 0.1094,
+      "step": 5531
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.358907471977333e-06,
+      "loss": 0.1131,
+      "step": 5532
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355533420528625e-06,
+      "loss": 0.1183,
+      "step": 5533
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.352159692737028e-06,
+      "loss": 0.1001,
+      "step": 5534
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3487862890154545e-06,
+      "loss": 0.1205,
+      "step": 5535
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345413209776768e-06,
+      "loss": 0.1149,
+      "step": 5536
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34204045543381e-06,
+      "loss": 0.1036,
+      "step": 5537
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338668026399365e-06,
+      "loss": 0.1062,
+      "step": 5538
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3352959230861945e-06,
+      "loss": 0.1076,
+      "step": 5539
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3319241459070044e-06,
+      "loss": 0.11,
+      "step": 5540
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328552695274471e-06,
+      "loss": 0.1107,
+      "step": 5541
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.325181571601227e-06,
+      "loss": 0.1116,
+      "step": 5542
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321810775299861e-06,
+      "loss": 0.112,
+      "step": 5543
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3184403067829324e-06,
+      "loss": 0.1014,
+      "step": 5544
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315070166462948e-06,
+      "loss": 0.1006,
+      "step": 5545
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.311700354752385e-06,
+      "loss": 0.1153,
+      "step": 5546
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.308330872063676e-06,
+      "loss": 0.1019,
+      "step": 5547
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.304961718809205e-06,
+      "loss": 0.0997,
+      "step": 5548
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301592895401335e-06,
+      "loss": 0.1115,
+      "step": 5549
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2982244022523685e-06,
+      "loss": 0.1006,
+      "step": 5550
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.294856239774584e-06,
+      "loss": 0.0978,
+      "step": 5551
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.291488408380202e-06,
+      "loss": 0.1086,
+      "step": 5552
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288120908481423e-06,
+      "loss": 0.0955,
+      "step": 5553
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284753740490391e-06,
+      "loss": 0.1138,
+      "step": 5554
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.281386904819213e-06,
+      "loss": 0.1015,
+      "step": 5555
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278020401879964e-06,
+      "loss": 0.1054,
+      "step": 5556
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274654232084662e-06,
+      "loss": 0.1134,
+      "step": 5557
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.271288395845302e-06,
+      "loss": 0.1033,
+      "step": 5558
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267922893573827e-06,
+      "loss": 0.0975,
+      "step": 5559
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.264557725682137e-06,
+      "loss": 0.1019,
+      "step": 5560
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.261192892582105e-06,
+      "loss": 0.1196,
+      "step": 5561
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.257828394685547e-06,
+      "loss": 0.1037,
+      "step": 5562
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2544642324042495e-06,
+      "loss": 0.1142,
+      "step": 5563
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.251100406149946e-06,
+      "loss": 0.105,
+      "step": 5564
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.247736916334346e-06,
+      "loss": 0.1169,
+      "step": 5565
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244373763369103e-06,
+      "loss": 0.109,
+      "step": 5566
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.241010947665832e-06,
+      "loss": 0.1109,
+      "step": 5567
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.237648469636115e-06,
+      "loss": 0.1081,
+      "step": 5568
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23428632969148e-06,
+      "loss": 0.1036,
+      "step": 5569
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2309245282434285e-06,
+      "loss": 0.0999,
+      "step": 5570
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.227563065703404e-06,
+      "loss": 0.1021,
+      "step": 5571
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224201942482825e-06,
+      "loss": 0.1164,
+      "step": 5572
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220841158993057e-06,
+      "loss": 0.1167,
+      "step": 5573
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.217480715645421e-06,
+      "loss": 0.0997,
+      "step": 5574
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214120612851213e-06,
+      "loss": 0.1083,
+      "step": 5575
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.210760851021669e-06,
+      "loss": 0.1085,
+      "step": 5576
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.207401430567998e-06,
+      "loss": 0.1022,
+      "step": 5577
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204042351901359e-06,
+      "loss": 0.112,
+      "step": 5578
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.200683615432864e-06,
+      "loss": 0.1145,
+      "step": 5579
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197325221573599e-06,
+      "loss": 0.1125,
+      "step": 5580
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193967170734594e-06,
+      "loss": 0.1142,
+      "step": 5581
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.190609463326842e-06,
+      "loss": 0.1155,
+      "step": 5582
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.187252099761295e-06,
+      "loss": 0.1263,
+      "step": 5583
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183895080448863e-06,
+      "loss": 0.1074,
+      "step": 5584
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.180538405800411e-06,
+      "loss": 0.1146,
+      "step": 5585
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1771820762267595e-06,
+      "loss": 0.1141,
+      "step": 5586
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173826092138701e-06,
+      "loss": 0.1059,
+      "step": 5587
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170470453946963e-06,
+      "loss": 0.1178,
+      "step": 5588
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167115162062254e-06,
+      "loss": 0.1119,
+      "step": 5589
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163760216895224e-06,
+      "loss": 0.1181,
+      "step": 5590
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.160405618856483e-06,
+      "loss": 0.1063,
+      "step": 5591
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1570513683566104e-06,
+      "loss": 0.1123,
+      "step": 5592
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153697465806124e-06,
+      "loss": 0.1078,
+      "step": 5593
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.150343911615518e-06,
+      "loss": 0.1077,
+      "step": 5594
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1469907061952245e-06,
+      "loss": 0.1094,
+      "step": 5595
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.143637849955654e-06,
+      "loss": 0.1142,
+      "step": 5596
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.140285343307159e-06,
+      "loss": 0.0994,
+      "step": 5597
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136933186660049e-06,
+      "loss": 0.1104,
+      "step": 5598
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133581380424605e-06,
+      "loss": 0.1114,
+      "step": 5599
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.130229925011047e-06,
+      "loss": 0.1095,
+      "step": 5600
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.126878820829568e-06,
+      "loss": 0.102,
+      "step": 5601
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123528068290309e-06,
+      "loss": 0.1134,
+      "step": 5602
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1201776678033655e-06,
+      "loss": 0.1068,
+      "step": 5603
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116827619778795e-06,
+      "loss": 0.0948,
+      "step": 5604
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1134779246266135e-06,
+      "loss": 0.1058,
+      "step": 5605
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.110128582756792e-06,
+      "loss": 0.1164,
+      "step": 5606
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1067795945792526e-06,
+      "loss": 0.113,
+      "step": 5607
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1034309605038865e-06,
+      "loss": 0.1023,
+      "step": 5608
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10008268094053e-06,
+      "loss": 0.1104,
+      "step": 5609
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.096734756298975e-06,
+      "loss": 0.1038,
+      "step": 5610
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.093387186988983e-06,
+      "loss": 0.0994,
+      "step": 5611
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0900399734202596e-06,
+      "loss": 0.1012,
+      "step": 5612
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.086693116002475e-06,
+      "loss": 0.1112,
+      "step": 5613
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0833466151452516e-06,
+      "loss": 0.1047,
+      "step": 5614
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0800004712581635e-06,
+      "loss": 0.1063,
+      "step": 5615
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.076654684750753e-06,
+      "loss": 0.117,
+      "step": 5616
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.073309256032504e-06,
+      "loss": 0.1153,
+      "step": 5617
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069964185512874e-06,
+      "loss": 0.1051,
+      "step": 5618
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.06661947360126e-06,
+      "loss": 0.1083,
+      "step": 5619
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063275120707027e-06,
+      "loss": 0.111,
+      "step": 5620
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.059931127239491e-06,
+      "loss": 0.1095,
+      "step": 5621
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056587493607918e-06,
+      "loss": 0.1036,
+      "step": 5622
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.053244220221547e-06,
+      "loss": 0.1187,
+      "step": 5623
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049901307489553e-06,
+      "loss": 0.1082,
+      "step": 5624
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046558755821082e-06,
+      "loss": 0.1188,
+      "step": 5625
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.043216565625226e-06,
+      "loss": 0.0936,
+      "step": 5626
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.039874737311038e-06,
+      "loss": 0.1132,
+      "step": 5627
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036533271287529e-06,
+      "loss": 0.1111,
+      "step": 5628
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.033192167963654e-06,
+      "loss": 0.0982,
+      "step": 5629
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.029851427748342e-06,
+      "loss": 0.1141,
+      "step": 5630
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026511051050456e-06,
+      "loss": 0.1034,
+      "step": 5631
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0231710382788374e-06,
+      "loss": 0.1076,
+      "step": 5632
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0198313898422645e-06,
+      "loss": 0.1103,
+      "step": 5633
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016492106149476e-06,
+      "loss": 0.1122,
+      "step": 5634
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0131531876091765e-06,
+      "loss": 0.1056,
+      "step": 5635
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.009814634630008e-06,
+      "loss": 0.1147,
+      "step": 5636
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0064764476205825e-06,
+      "loss": 0.1119,
+      "step": 5637
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003138626989457e-06,
+      "loss": 0.1137,
+      "step": 5638
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.999801173145156e-06,
+      "loss": 0.1132,
+      "step": 5639
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.996464086496147e-06,
+      "loss": 0.1017,
+      "step": 5640
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9931273674508535e-06,
+      "loss": 0.1169,
+      "step": 5641
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989791016417666e-06,
+      "loss": 0.0999,
+      "step": 5642
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986455033804912e-06,
+      "loss": 0.1065,
+      "step": 5643
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.983119420020893e-06,
+      "loss": 0.1079,
+      "step": 5644
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.979784175473852e-06,
+      "loss": 0.1135,
+      "step": 5645
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976449300571987e-06,
+      "loss": 0.1138,
+      "step": 5646
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973114795723459e-06,
+      "loss": 0.1,
+      "step": 5647
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9697806613363735e-06,
+      "loss": 0.1124,
+      "step": 5648
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966446897818806e-06,
+      "loss": 0.0967,
+      "step": 5649
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9631135055787645e-06,
+      "loss": 0.1087,
+      "step": 5650
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.959780485024237e-06,
+      "loss": 0.11,
+      "step": 5651
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956447836563145e-06,
+      "loss": 0.1056,
+      "step": 5652
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9531155606033705e-06,
+      "loss": 0.1116,
+      "step": 5653
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9497836575527596e-06,
+      "loss": 0.0991,
+      "step": 5654
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.946452127819099e-06,
+      "loss": 0.1042,
+      "step": 5655
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.943120971810136e-06,
+      "loss": 0.1088,
+      "step": 5656
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.939790189933575e-06,
+      "loss": 0.1144,
+      "step": 5657
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936459782597069e-06,
+      "loss": 0.1017,
+      "step": 5658
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.93312975020823e-06,
+      "loss": 0.1055,
+      "step": 5659
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9298000931746145e-06,
+      "loss": 0.1002,
+      "step": 5660
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926470811903752e-06,
+      "loss": 0.1077,
+      "step": 5661
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9231419068031024e-06,
+      "loss": 0.1126,
+      "step": 5662
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919813378280101e-06,
+      "loss": 0.1051,
+      "step": 5663
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916485226742125e-06,
+      "loss": 0.0996,
+      "step": 5664
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.913157452596501e-06,
+      "loss": 0.1033,
+      "step": 5665
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.103,
+      "step": 5666
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.906503038111437e-06,
+      "loss": 0.1158,
+      "step": 5667
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.903176398586428e-06,
+      "loss": 0.1109,
+      "step": 5668
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.899850138082649e-06,
+      "loss": 0.1022,
+      "step": 5669
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896524257007199e-06,
+      "loss": 0.1086,
+      "step": 5670
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8931987557671384e-06,
+      "loss": 0.1068,
+      "step": 5671
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.889873634769471e-06,
+      "loss": 0.1102,
+      "step": 5672
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.886548894421166e-06,
+      "loss": 0.1076,
+      "step": 5673
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883224535129131e-06,
+      "loss": 0.1006,
+      "step": 5674
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8799005573002454e-06,
+      "loss": 0.1091,
+      "step": 5675
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.876576961341326e-06,
+      "loss": 0.1121,
+      "step": 5676
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.873253747659147e-06,
+      "loss": 0.1007,
+      "step": 5677
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.86993091666044e-06,
+      "loss": 0.1211,
+      "step": 5678
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.866608468751888e-06,
+      "loss": 0.094,
+      "step": 5679
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863286404340127e-06,
+      "loss": 0.1107,
+      "step": 5680
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8599647238317444e-06,
+      "loss": 0.1001,
+      "step": 5681
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.856643427633278e-06,
+      "loss": 0.1019,
+      "step": 5682
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853322516151231e-06,
+      "loss": 0.0981,
+      "step": 5683
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850001989792041e-06,
+      "loss": 0.1066,
+      "step": 5684
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.846681848962118e-06,
+      "loss": 0.1056,
+      "step": 5685
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8433620940678045e-06,
+      "loss": 0.1186,
+      "step": 5686
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8400427255154175e-06,
+      "loss": 0.1024,
+      "step": 5687
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.83672374371121e-06,
+      "loss": 0.1044,
+      "step": 5688
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.833405149061391e-06,
+      "loss": 0.1088,
+      "step": 5689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8300869419721296e-06,
+      "loss": 0.1046,
+      "step": 5690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.826769122849533e-06,
+      "loss": 0.1086,
+      "step": 5691
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.823451692099683e-06,
+      "loss": 0.1186,
+      "step": 5692
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.820134650128593e-06,
+      "loss": 0.1124,
+      "step": 5693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.816817997342236e-06,
+      "loss": 0.1139,
+      "step": 5694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8135017341465424e-06,
+      "loss": 0.1114,
+      "step": 5695
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.810185860947386e-06,
+      "loss": 0.102,
+      "step": 5696
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806870378150604e-06,
+      "loss": 0.1056,
+      "step": 5697
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.803555286161973e-06,
+      "loss": 0.1157,
+      "step": 5698
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800240585387235e-06,
+      "loss": 0.1015,
+      "step": 5699
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.796926276232071e-06,
+      "loss": 0.1117,
+      "step": 5700
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793612359102122e-06,
+      "loss": 0.1026,
+      "step": 5701
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.790298834402982e-06,
+      "loss": 0.1123,
+      "step": 5702
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.786985702540187e-06,
+      "loss": 0.122,
+      "step": 5703
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7836729639192444e-06,
+      "loss": 0.1052,
+      "step": 5704
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.780360618945592e-06,
+      "loss": 0.1023,
+      "step": 5705
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.77704866802463e-06,
+      "loss": 0.1047,
+      "step": 5706
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.773737111561712e-06,
+      "loss": 0.1151,
+      "step": 5707
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770425949962136e-06,
+      "loss": 0.1081,
+      "step": 5708
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.767115183631163e-06,
+      "loss": 0.1147,
+      "step": 5709
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7638048129739935e-06,
+      "loss": 0.1017,
+      "step": 5710
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7604948383957855e-06,
+      "loss": 0.1057,
+      "step": 5711
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.757185260301651e-06,
+      "loss": 0.1096,
+      "step": 5712
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.753876079096643e-06,
+      "loss": 0.1195,
+      "step": 5713
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750567295185781e-06,
+      "loss": 0.1103,
+      "step": 5714
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.747258908974023e-06,
+      "loss": 0.1106,
+      "step": 5715
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.743950920866289e-06,
+      "loss": 0.1093,
+      "step": 5716
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7406433312674355e-06,
+      "loss": 0.1081,
+      "step": 5717
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737336140582291e-06,
+      "loss": 0.1115,
+      "step": 5718
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.734029349215618e-06,
+      "loss": 0.0962,
+      "step": 5719
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7307229575721334e-06,
+      "loss": 0.1019,
+      "step": 5720
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.727416966056509e-06,
+      "loss": 0.102,
+      "step": 5721
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7241113750733674e-06,
+      "loss": 0.1007,
+      "step": 5722
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720806185027282e-06,
+      "loss": 0.1045,
+      "step": 5723
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7175013963227755e-06,
+      "loss": 0.1056,
+      "step": 5724
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.714197009364316e-06,
+      "loss": 0.1132,
+      "step": 5725
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.710893024556338e-06,
+      "loss": 0.1117,
+      "step": 5726
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7075894423032086e-06,
+      "loss": 0.0966,
+      "step": 5727
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.704286263009263e-06,
+      "loss": 0.125,
+      "step": 5728
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700983487078771e-06,
+      "loss": 0.1147,
+      "step": 5729
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697681114915964e-06,
+      "loss": 0.0979,
+      "step": 5730
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694379146925021e-06,
+      "loss": 0.0965,
+      "step": 5731
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.691077583510069e-06,
+      "loss": 0.1035,
+      "step": 5732
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687776425075188e-06,
+      "loss": 0.112,
+      "step": 5733
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.684475672024404e-06,
+      "loss": 0.1116,
+      "step": 5734
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.681175324761705e-06,
+      "loss": 0.1054,
+      "step": 5735
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677875383691018e-06,
+      "loss": 0.1193,
+      "step": 5736
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.674575849216219e-06,
+      "loss": 0.1052,
+      "step": 5737
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671276721741149e-06,
+      "loss": 0.1131,
+      "step": 5738
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6679780016695775e-06,
+      "loss": 0.1019,
+      "step": 5739
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664679689405247e-06,
+      "loss": 0.1138,
+      "step": 5740
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661381785351833e-06,
+      "loss": 0.1133,
+      "step": 5741
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.658084289912969e-06,
+      "loss": 0.0979,
+      "step": 5742
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.654787203492234e-06,
+      "loss": 0.1067,
+      "step": 5743
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651490526493161e-06,
+      "loss": 0.0952,
+      "step": 5744
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.648194259319234e-06,
+      "loss": 0.1042,
+      "step": 5745
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.644898402373877e-06,
+      "loss": 0.1049,
+      "step": 5746
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.64160295606048e-06,
+      "loss": 0.1058,
+      "step": 5747
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.63830792078237e-06,
+      "loss": 0.1075,
+      "step": 5748
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.635013296942821e-06,
+      "loss": 0.1088,
+      "step": 5749
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631719084945074e-06,
+      "loss": 0.1022,
+      "step": 5750
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628425285192301e-06,
+      "loss": 0.0985,
+      "step": 5751
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6251318980876355e-06,
+      "loss": 0.1053,
+      "step": 5752
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621838924034153e-06,
+      "loss": 0.1011,
+      "step": 5753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.618546363434885e-06,
+      "loss": 0.1053,
+      "step": 5754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.615254216692808e-06,
+      "loss": 0.1149,
+      "step": 5755
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.611962484210843e-06,
+      "loss": 0.0951,
+      "step": 5756
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.608671166391875e-06,
+      "loss": 0.1135,
+      "step": 5757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.605380263638722e-06,
+      "loss": 0.1143,
+      "step": 5758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602089776354166e-06,
+      "loss": 0.1011,
+      "step": 5759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598799704940928e-06,
+      "loss": 0.0996,
+      "step": 5760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.595510049801675e-06,
+      "loss": 0.1119,
+      "step": 5761
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592220811339039e-06,
+      "loss": 0.1051,
+      "step": 5762
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588931989955582e-06,
+      "loss": 0.1185,
+      "step": 5763
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.585643586053831e-06,
+      "loss": 0.1064,
+      "step": 5764
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582355600036248e-06,
+      "loss": 0.1072,
+      "step": 5765
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579068032305258e-06,
+      "loss": 0.1024,
+      "step": 5766
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575780883263223e-06,
+      "loss": 0.1087,
+      "step": 5767
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572494153312454e-06,
+      "loss": 0.1094,
+      "step": 5768
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5692078428552254e-06,
+      "loss": 0.101,
+      "step": 5769
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56592195229374e-06,
+      "loss": 0.1041,
+      "step": 5770
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.562636482030167e-06,
+      "loss": 0.1018,
+      "step": 5771
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559351432466612e-06,
+      "loss": 0.1182,
+      "step": 5772
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55606680400513e-06,
+      "loss": 0.1022,
+      "step": 5773
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552782597047736e-06,
+      "loss": 0.0999,
+      "step": 5774
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.549498811996377e-06,
+      "loss": 0.11,
+      "step": 5775
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546215449252962e-06,
+      "loss": 0.1099,
+      "step": 5776
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.542932509219337e-06,
+      "loss": 0.1135,
+      "step": 5777
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539649992297311e-06,
+      "loss": 0.1103,
+      "step": 5778
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536367898888627e-06,
+      "loss": 0.0979,
+      "step": 5779
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533086229394976e-06,
+      "loss": 0.0913,
+      "step": 5780
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529804984218014e-06,
+      "loss": 0.1142,
+      "step": 5781
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5265241637593225e-06,
+      "loss": 0.1119,
+      "step": 5782
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523243768420453e-06,
+      "loss": 0.1127,
+      "step": 5783
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5199637986028855e-06,
+      "loss": 0.1157,
+      "step": 5784
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.516684254708061e-06,
+      "loss": 0.1032,
+      "step": 5785
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513405137137364e-06,
+      "loss": 0.1101,
+      "step": 5786
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510126446292121e-06,
+      "loss": 0.1132,
+      "step": 5787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50684818257362e-06,
+      "loss": 0.1182,
+      "step": 5788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503570346383081e-06,
+      "loss": 0.1013,
+      "step": 5789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.500292938121688e-06,
+      "loss": 0.1016,
+      "step": 5790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49701595819056e-06,
+      "loss": 0.1067,
+      "step": 5791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493739406990762e-06,
+      "loss": 0.0944,
+      "step": 5792
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.490463284923323e-06,
+      "loss": 0.1006,
+      "step": 5793
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487187592389199e-06,
+      "loss": 0.1071,
+      "step": 5794
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483912329789308e-06,
+      "loss": 0.1024,
+      "step": 5795
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48063749752451e-06,
+      "loss": 0.1084,
+      "step": 5796
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477363095995615e-06,
+      "loss": 0.1092,
+      "step": 5797
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4740891256033736e-06,
+      "loss": 0.1024,
+      "step": 5798
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.470815586748485e-06,
+      "loss": 0.1086,
+      "step": 5799
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.46754247983161e-06,
+      "loss": 0.1087,
+      "step": 5800
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464269805253333e-06,
+      "loss": 0.104,
+      "step": 5801
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.460997563414208e-06,
+      "loss": 0.1038,
+      "step": 5802
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45772575471472e-06,
+      "loss": 0.1139,
+      "step": 5803
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454454379555304e-06,
+      "loss": 0.0984,
+      "step": 5804
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.451183438336352e-06,
+      "loss": 0.1088,
+      "step": 5805
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.447912931458192e-06,
+      "loss": 0.0988,
+      "step": 5806
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444642859321103e-06,
+      "loss": 0.0972,
+      "step": 5807
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.441373222325306e-06,
+      "loss": 0.0978,
+      "step": 5808
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4381040208709786e-06,
+      "loss": 0.1118,
+      "step": 5809
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434835255358239e-06,
+      "loss": 0.1148,
+      "step": 5810
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.431566926187147e-06,
+      "loss": 0.0936,
+      "step": 5811
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428299033757721e-06,
+      "loss": 0.1034,
+      "step": 5812
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.425031578469914e-06,
+      "loss": 0.1027,
+      "step": 5813
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.421764560723636e-06,
+      "loss": 0.113,
+      "step": 5814
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4184979809187355e-06,
+      "loss": 0.1123,
+      "step": 5815
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.415231839455009e-06,
+      "loss": 0.1103,
+      "step": 5816
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.411966136732201e-06,
+      "loss": 0.1161,
+      "step": 5817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.408700873150005e-06,
+      "loss": 0.1065,
+      "step": 5818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.405436049108057e-06,
+      "loss": 0.1023,
+      "step": 5819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.402171665005934e-06,
+      "loss": 0.0995,
+      "step": 5820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.398907721243175e-06,
+      "loss": 0.1075,
+      "step": 5821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395644218219248e-06,
+      "loss": 0.0943,
+      "step": 5822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3923811563335716e-06,
+      "loss": 0.1014,
+      "step": 5823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389118535985522e-06,
+      "loss": 0.1124,
+      "step": 5824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.385856357574406e-06,
+      "loss": 0.1183,
+      "step": 5825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.382594621499484e-06,
+      "loss": 0.1168,
+      "step": 5826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379333328159962e-06,
+      "loss": 0.1098,
+      "step": 5827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.376072477954989e-06,
+      "loss": 0.0919,
+      "step": 5828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372812071283665e-06,
+      "loss": 0.1096,
+      "step": 5829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369552108545024e-06,
+      "loss": 0.1129,
+      "step": 5830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366292590138064e-06,
+      "loss": 0.122,
+      "step": 5831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.36303351646171e-06,
+      "loss": 0.1042,
+      "step": 5832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3597748879148495e-06,
+      "loss": 0.1048,
+      "step": 5833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356516704896303e-06,
+      "loss": 0.1035,
+      "step": 5834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.353258967804834e-06,
+      "loss": 0.1065,
+      "step": 5835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3500016770391704e-06,
+      "loss": 0.1167,
+      "step": 5836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.346744832997963e-06,
+      "loss": 0.1089,
+      "step": 5837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3434884360798255e-06,
+      "loss": 0.1027,
+      "step": 5838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340232486683302e-06,
+      "loss": 0.1115,
+      "step": 5839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336976985206894e-06,
+      "loss": 0.0983,
+      "step": 5840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.333721932049044e-06,
+      "loss": 0.102,
+      "step": 5841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.330467327608133e-06,
+      "loss": 0.1189,
+      "step": 5842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.327213172282502e-06,
+      "loss": 0.1097,
+      "step": 5843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.323959466470418e-06,
+      "loss": 0.1051,
+      "step": 5844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.320706210570113e-06,
+      "loss": 0.1119,
+      "step": 5845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3174534049797495e-06,
+      "loss": 0.1145,
+      "step": 5846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.314201050097436e-06,
+      "loss": 0.1233,
+      "step": 5847
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.310949146321237e-06,
+      "loss": 0.1023,
+      "step": 5848
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307697694049147e-06,
+      "loss": 0.107,
+      "step": 5849
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304446693679116e-06,
+      "loss": 0.1026,
+      "step": 5850
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.301196145609035e-06,
+      "loss": 0.1127,
+      "step": 5851
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.297946050236735e-06,
+      "loss": 0.0993,
+      "step": 5852
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.294696407960004e-06,
+      "loss": 0.0987,
+      "step": 5853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291447219176557e-06,
+      "loss": 0.1167,
+      "step": 5854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.288198484284074e-06,
+      "loss": 0.1124,
+      "step": 5855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.284950203680159e-06,
+      "loss": 0.1001,
+      "step": 5856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281702377762379e-06,
+      "loss": 0.1046,
+      "step": 5857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278455006928233e-06,
+      "loss": 0.1134,
+      "step": 5858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.275208091575162e-06,
+      "loss": 0.1111,
+      "step": 5859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.271961632100566e-06,
+      "loss": 0.1037,
+      "step": 5860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268715628901772e-06,
+      "loss": 0.1067,
+      "step": 5861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.265470082376066e-06,
+      "loss": 0.1208,
+      "step": 5862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2622249929206655e-06,
+      "loss": 0.0945,
+      "step": 5863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258980360932746e-06,
+      "loss": 0.0996,
+      "step": 5864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2557361868094135e-06,
+      "loss": 0.1263,
+      "step": 5865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.252492470947722e-06,
+      "loss": 0.0954,
+      "step": 5866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249249213744676e-06,
+      "loss": 0.1176,
+      "step": 5867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.246006415597214e-06,
+      "loss": 0.1082,
+      "step": 5868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242764076902226e-06,
+      "loss": 0.1158,
+      "step": 5869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239522198056545e-06,
+      "loss": 0.1008,
+      "step": 5870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.236280779456939e-06,
+      "loss": 0.1126,
+      "step": 5871
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233039821500133e-06,
+      "loss": 0.1071,
+      "step": 5872
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229799324582781e-06,
+      "loss": 0.1246,
+      "step": 5873
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226559289101498e-06,
+      "loss": 0.1002,
+      "step": 5874
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2233197154528245e-06,
+      "loss": 0.113,
+      "step": 5875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22008060403326e-06,
+      "loss": 0.1142,
+      "step": 5876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.216841955239237e-06,
+      "loss": 0.0986,
+      "step": 5877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213603769467132e-06,
+      "loss": 0.1136,
+      "step": 5878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.210366047113273e-06,
+      "loss": 0.1025,
+      "step": 5879
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.207128788573922e-06,
+      "loss": 0.0992,
+      "step": 5880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.203891994245293e-06,
+      "loss": 0.1061,
+      "step": 5881
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2006556645235315e-06,
+      "loss": 0.097,
+      "step": 5882
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1974197998047335e-06,
+      "loss": 0.1075,
+      "step": 5883
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194184400484946e-06,
+      "loss": 0.1074,
+      "step": 5884
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190949466960139e-06,
+      "loss": 0.0999,
+      "step": 5885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.187714999626246e-06,
+      "loss": 0.0971,
+      "step": 5886
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1844809988791284e-06,
+      "loss": 0.1093,
+      "step": 5887
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1812474651146035e-06,
+      "loss": 0.0949,
+      "step": 5888
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.17801439872842e-06,
+      "loss": 0.0954,
+      "step": 5889
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174781800116274e-06,
+      "loss": 0.0958,
+      "step": 5890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1715496696738035e-06,
+      "loss": 0.1137,
+      "step": 5891
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16831800779659e-06,
+      "loss": 0.1001,
+      "step": 5892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.165086814880162e-06,
+      "loss": 0.1084,
+      "step": 5893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161856091319984e-06,
+      "loss": 0.1056,
+      "step": 5894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158625837511458e-06,
+      "loss": 0.1219,
+      "step": 5895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155396053849948e-06,
+      "loss": 0.1122,
+      "step": 5896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152166740730738e-06,
+      "loss": 0.1047,
+      "step": 5897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.148937898549072e-06,
+      "loss": 0.1022,
+      "step": 5898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145709527700123e-06,
+      "loss": 0.1111,
+      "step": 5899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.14248162857902e-06,
+      "loss": 0.1067,
+      "step": 5900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13925420158082e-06,
+      "loss": 0.0995,
+      "step": 5901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.136027247100531e-06,
+      "loss": 0.1125,
+      "step": 5902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1328007655331015e-06,
+      "loss": 0.119,
+      "step": 5903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129574757273417e-06,
+      "loss": 0.114,
+      "step": 5904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1263492227163164e-06,
+      "loss": 0.1064,
+      "step": 5905
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123124162256573e-06,
+      "loss": 0.1068,
+      "step": 5906
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1198995762888965e-06,
+      "loss": 0.1131,
+      "step": 5907
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116675465207954e-06,
+      "loss": 0.116,
+      "step": 5908
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1134518294083364e-06,
+      "loss": 0.0941,
+      "step": 5909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110228669284595e-06,
+      "loss": 0.1077,
+      "step": 5910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.107005985231205e-06,
+      "loss": 0.1152,
+      "step": 5911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103783777642599e-06,
+      "loss": 0.1067,
+      "step": 5912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.100562046913137e-06,
+      "loss": 0.1048,
+      "step": 5913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.097340793437132e-06,
+      "loss": 0.1023,
+      "step": 5914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094120017608835e-06,
+      "loss": 0.1171,
+      "step": 5915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090899719822432e-06,
+      "loss": 0.1041,
+      "step": 5916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087679900472065e-06,
+      "loss": 0.1016,
+      "step": 5917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.084460559951802e-06,
+      "loss": 0.1083,
+      "step": 5918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0812416986556575e-06,
+      "loss": 0.1083,
+      "step": 5919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.078023316977598e-06,
+      "loss": 0.0927,
+      "step": 5920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.074805415311512e-06,
+      "loss": 0.1054,
+      "step": 5921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.071587994051249e-06,
+      "loss": 0.101,
+      "step": 5922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068371053590583e-06,
+      "loss": 0.0973,
+      "step": 5923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065154594323241e-06,
+      "loss": 0.1015,
+      "step": 5924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.061938616642884e-06,
+      "loss": 0.1062,
+      "step": 5925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058723120943113e-06,
+      "loss": 0.0968,
+      "step": 5926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.055508107617482e-06,
+      "loss": 0.1012,
+      "step": 5927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052293577059468e-06,
+      "loss": 0.1101,
+      "step": 5928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0490795296625085e-06,
+      "loss": 0.105,
+      "step": 5929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045865965819961e-06,
+      "loss": 0.1093,
+      "step": 5930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.042652885925146e-06,
+      "loss": 0.1053,
+      "step": 5931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0394402903713075e-06,
+      "loss": 0.1142,
+      "step": 5932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036228179551635e-06,
+      "loss": 0.1104,
+      "step": 5933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0330165538592615e-06,
+      "loss": 0.1101,
+      "step": 5934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029805413687255e-06,
+      "loss": 0.0963,
+      "step": 5935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.026594759428635e-06,
+      "loss": 0.1061,
+      "step": 5936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023384591476351e-06,
+      "loss": 0.1056,
+      "step": 5937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.020174910223293e-06,
+      "loss": 0.1067,
+      "step": 5938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.016965716062302e-06,
+      "loss": 0.1,
+      "step": 5939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0137570093861455e-06,
+      "loss": 0.1062,
+      "step": 5940
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.010548790587544e-06,
+      "loss": 0.1038,
+      "step": 5941
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.007341060059149e-06,
+      "loss": 0.1074,
+      "step": 5942
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.004133818193554e-06,
+      "loss": 0.1063,
+      "step": 5943
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0009270653833e-06,
+      "loss": 0.0994,
+      "step": 5944
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9977208020208565e-06,
+      "loss": 0.1102,
+      "step": 5945
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994515028498646e-06,
+      "loss": 0.1073,
+      "step": 5946
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.991309745209013e-06,
+      "loss": 0.111,
+      "step": 5947
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.988104952544266e-06,
+      "loss": 0.0968,
+      "step": 5948
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.984900650896633e-06,
+      "loss": 0.0997,
+      "step": 5949
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981696840658289e-06,
+      "loss": 0.0976,
+      "step": 5950
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.978493522221355e-06,
+      "loss": 0.098,
+      "step": 5951
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9752906959778775e-06,
+      "loss": 0.1155,
+      "step": 5952
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.972088362319861e-06,
+      "loss": 0.1095,
+      "step": 5953
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.968886521639234e-06,
+      "loss": 0.1074,
+      "step": 5954
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9656851743278756e-06,
+      "loss": 0.1001,
+      "step": 5955
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.962484320777594e-06,
+      "loss": 0.1018,
+      "step": 5956
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959283961380145e-06,
+      "loss": 0.1057,
+      "step": 5957
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956084096527224e-06,
+      "loss": 0.1066,
+      "step": 5958
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9528847266104576e-06,
+      "loss": 0.1074,
+      "step": 5959
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.949685852021425e-06,
+      "loss": 0.1172,
+      "step": 5960
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.946487473151635e-06,
+      "loss": 0.111,
+      "step": 5961
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.943289590392534e-06,
+      "loss": 0.0976,
+      "step": 5962
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.940092204135517e-06,
+      "loss": 0.114,
+      "step": 5963
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.936895314771911e-06,
+      "loss": 0.116,
+      "step": 5964
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933698922692983e-06,
+      "loss": 0.101,
+      "step": 5965
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9305030282899425e-06,
+      "loss": 0.1186,
+      "step": 5966
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.927307631953938e-06,
+      "loss": 0.1035,
+      "step": 5967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.924112734076053e-06,
+      "loss": 0.1201,
+      "step": 5968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.920918335047307e-06,
+      "loss": 0.1039,
+      "step": 5969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917724435258672e-06,
+      "loss": 0.1123,
+      "step": 5970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.914531035101043e-06,
+      "loss": 0.1026,
+      "step": 5971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.9113381349652695e-06,
+      "loss": 0.1054,
+      "step": 5972
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908145735242126e-06,
+      "loss": 0.0954,
+      "step": 5973
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.904953836322329e-06,
+      "loss": 0.0958,
+      "step": 5974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901762438596544e-06,
+      "loss": 0.0963,
+      "step": 5975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898571542455358e-06,
+      "loss": 0.103,
+      "step": 5976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.895381148289314e-06,
+      "loss": 0.1001,
+      "step": 5977
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8921912564888775e-06,
+      "loss": 0.1093,
+      "step": 5978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889001867444469e-06,
+      "loss": 0.1039,
+      "step": 5979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885812981546434e-06,
+      "loss": 0.1061,
+      "step": 5980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.882624599185057e-06,
+      "loss": 0.1035,
+      "step": 5981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879436720750574e-06,
+      "loss": 0.1023,
+      "step": 5982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8762493466331425e-06,
+      "loss": 0.1081,
+      "step": 5983
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873062477222875e-06,
+      "loss": 0.1104,
+      "step": 5984
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869876112909807e-06,
+      "loss": 0.1095,
+      "step": 5985
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866690254083918e-06,
+      "loss": 0.1026,
+      "step": 5986
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.863504901135129e-06,
+      "loss": 0.1097,
+      "step": 5987
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.860320054453295e-06,
+      "loss": 0.0988,
+      "step": 5988
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.857135714428215e-06,
+      "loss": 0.1034,
+      "step": 5989
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.853951881449611e-06,
+      "loss": 0.1076,
+      "step": 5990
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850768555907166e-06,
+      "loss": 0.1182,
+      "step": 5991
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847585738190481e-06,
+      "loss": 0.1126,
+      "step": 5992
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.844403428689099e-06,
+      "loss": 0.1045,
+      "step": 5993
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841221627792513e-06,
+      "loss": 0.1086,
+      "step": 5994
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.838040335890135e-06,
+      "loss": 0.1103,
+      "step": 5995
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834859553371335e-06,
+      "loss": 0.1175,
+      "step": 5996
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8316792806254e-06,
+      "loss": 0.1172,
+      "step": 5997
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8284995180415685e-06,
+      "loss": 0.1156,
+      "step": 5998
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825320266009014e-06,
+      "loss": 0.1074,
+      "step": 5999
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822141524916842e-06,
+      "loss": 0.1028,
+      "step": 6000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818963295154105e-06,
+      "loss": 0.1019,
+      "step": 6001
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815785577109785e-06,
+      "loss": 0.1067,
+      "step": 6002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.812608371172807e-06,
+      "loss": 0.0944,
+      "step": 6003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.809431677732024e-06,
+      "loss": 0.1116,
+      "step": 6004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806255497176234e-06,
+      "loss": 0.1128,
+      "step": 6005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8030798298941745e-06,
+      "loss": 0.1089,
+      "step": 6006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799904676274515e-06,
+      "loss": 0.1098,
+      "step": 6007
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796730036705862e-06,
+      "loss": 0.1103,
+      "step": 6008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793555911576761e-06,
+      "loss": 0.106,
+      "step": 6009
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.790382301275696e-06,
+      "loss": 0.1098,
+      "step": 6010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787209206191089e-06,
+      "loss": 0.1063,
+      "step": 6011
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784036626711288e-06,
+      "loss": 0.1095,
+      "step": 6012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.780864563224591e-06,
+      "loss": 0.0989,
+      "step": 6013
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.777693016119226e-06,
+      "loss": 0.1095,
+      "step": 6014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77452198578336e-06,
+      "loss": 0.1002,
+      "step": 6015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.771351472605102e-06,
+      "loss": 0.1094,
+      "step": 6016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.768181476972479e-06,
+      "loss": 0.098,
+      "step": 6017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765011999273484e-06,
+      "loss": 0.0964,
+      "step": 6018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761843039896016e-06,
+      "loss": 0.114,
+      "step": 6019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.758674599227938e-06,
+      "loss": 0.1021,
+      "step": 6020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.755506677657021e-06,
+      "loss": 0.1146,
+      "step": 6021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752339275571006e-06,
+      "loss": 0.1073,
+      "step": 6022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7491723933575395e-06,
+      "loss": 0.1048,
+      "step": 6023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746006031404221e-06,
+      "loss": 0.1144,
+      "step": 6024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742840190098585e-06,
+      "loss": 0.0976,
+      "step": 6025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73967486982809e-06,
+      "loss": 0.1166,
+      "step": 6026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7365100709801556e-06,
+      "loss": 0.0995,
+      "step": 6027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.733345793942111e-06,
+      "loss": 0.1135,
+      "step": 6028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730182039101236e-06,
+      "loss": 0.1285,
+      "step": 6029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.727018806844746e-06,
+      "loss": 0.1064,
+      "step": 6030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.723856097559787e-06,
+      "loss": 0.1037,
+      "step": 6031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.720693911633446e-06,
+      "loss": 0.0995,
+      "step": 6032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.717532249452743e-06,
+      "loss": 0.1009,
+      "step": 6033
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.714371111404639e-06,
+      "loss": 0.1029,
+      "step": 6034
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.711210497876018e-06,
+      "loss": 0.0934,
+      "step": 6035
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.708050409253716e-06,
+      "loss": 0.1065,
+      "step": 6036
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704890845924493e-06,
+      "loss": 0.1051,
+      "step": 6037
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.70173180827505e-06,
+      "loss": 0.1116,
+      "step": 6038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698573296692025e-06,
+      "loss": 0.1078,
+      "step": 6039
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695415311561989e-06,
+      "loss": 0.113,
+      "step": 6040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692257853271441e-06,
+      "loss": 0.1149,
+      "step": 6041
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.689100922206837e-06,
+      "loss": 0.1035,
+      "step": 6042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68594451875454e-06,
+      "loss": 0.102,
+      "step": 6043
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682788643300878e-06,
+      "loss": 0.1048,
+      "step": 6044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679633296232087e-06,
+      "loss": 0.1213,
+      "step": 6045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67647847793436e-06,
+      "loss": 0.1087,
+      "step": 6046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673324188793813e-06,
+      "loss": 0.1025,
+      "step": 6047
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6701704291964936e-06,
+      "loss": 0.1028,
+      "step": 6048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.667017199528404e-06,
+      "loss": 0.1037,
+      "step": 6049
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.663864500175456e-06,
+      "loss": 0.117,
+      "step": 6050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660712331523524e-06,
+      "loss": 0.1083,
+      "step": 6051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.657560693958389e-06,
+      "loss": 0.1041,
+      "step": 6052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6544095878657886e-06,
+      "loss": 0.1061,
+      "step": 6053
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6512590136313875e-06,
+      "loss": 0.1177,
+      "step": 6054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.648108971640782e-06,
+      "loss": 0.1138,
+      "step": 6055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.644959462279509e-06,
+      "loss": 0.1077,
+      "step": 6056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.641810485933038e-06,
+      "loss": 0.102,
+      "step": 6057
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638662042986777e-06,
+      "loss": 0.1012,
+      "step": 6058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635514133826058e-06,
+      "loss": 0.1028,
+      "step": 6059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632366758836157e-06,
+      "loss": 0.1147,
+      "step": 6060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629219918402282e-06,
+      "loss": 0.1084,
+      "step": 6061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626073612909577e-06,
+      "loss": 0.1054,
+      "step": 6062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622927842743118e-06,
+      "loss": 0.1176,
+      "step": 6063
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.619782608287922e-06,
+      "loss": 0.1031,
+      "step": 6064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.616637909928924e-06,
+      "loss": 0.1073,
+      "step": 6065
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.61349374805102e-06,
+      "loss": 0.1012,
+      "step": 6066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610350123039011e-06,
+      "loss": 0.1127,
+      "step": 6067
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607207035277653e-06,
+      "loss": 0.1013,
+      "step": 6068
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.604064485151628e-06,
+      "loss": 0.1222,
+      "step": 6069
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.600922473045554e-06,
+      "loss": 0.1036,
+      "step": 6070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.597780999343985e-06,
+      "loss": 0.1061,
+      "step": 6071
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5946400644314e-06,
+      "loss": 0.1047,
+      "step": 6072
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.591499668692231e-06,
+      "loss": 0.113,
+      "step": 6073
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588359812510816e-06,
+      "loss": 0.1011,
+      "step": 6074
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585220496271461e-06,
+      "loss": 0.1067,
+      "step": 6075
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.582081720358375e-06,
+      "loss": 0.0938,
+      "step": 6076
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.578943485155717e-06,
+      "loss": 0.1063,
+      "step": 6077
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.575805791047577e-06,
+      "loss": 0.1124,
+      "step": 6078
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5726686384179785e-06,
+      "loss": 0.1136,
+      "step": 6079
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.569532027650884e-06,
+      "loss": 0.1029,
+      "step": 6080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5663959591301695e-06,
+      "loss": 0.1032,
+      "step": 6081
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.563260433239678e-06,
+      "loss": 0.1056,
+      "step": 6082
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5601254503631545e-06,
+      "loss": 0.1006,
+      "step": 6083
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.556991010884294e-06,
+      "loss": 0.1091,
+      "step": 6084
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553857115186722e-06,
+      "loss": 0.0999,
+      "step": 6085
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5507237636539975e-06,
+      "loss": 0.1012,
+      "step": 6086
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547590956669613e-06,
+      "loss": 0.1158,
+      "step": 6087
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.544458694616992e-06,
+      "loss": 0.1081,
+      "step": 6088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541326977879498e-06,
+      "loss": 0.1097,
+      "step": 6089
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538195806840414e-06,
+      "loss": 0.1059,
+      "step": 6090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53506518188297e-06,
+      "loss": 0.1106,
+      "step": 6091
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531935103390323e-06,
+      "loss": 0.1025,
+      "step": 6092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5288055717455655e-06,
+      "loss": 0.0948,
+      "step": 6093
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525676587331723e-06,
+      "loss": 0.1085,
+      "step": 6094
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522548150531755e-06,
+      "loss": 0.1094,
+      "step": 6095
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.519420261728539e-06,
+      "loss": 0.1216,
+      "step": 6096
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.516292921304916e-06,
+      "loss": 0.1044,
+      "step": 6097
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.51316612964363e-06,
+      "loss": 0.1039,
+      "step": 6098
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510039887127373e-06,
+      "loss": 0.1087,
+      "step": 6099
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.506914194138768e-06,
+      "loss": 0.0976,
+      "step": 6100
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5037890510603705e-06,
+      "loss": 0.108,
+      "step": 6101
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500664458274669e-06,
+      "loss": 0.0994,
+      "step": 6102
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.497540416164073e-06,
+      "loss": 0.1138,
+      "step": 6103
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494416925110953e-06,
+      "loss": 0.1106,
+      "step": 6104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4912939854975725e-06,
+      "loss": 0.1043,
+      "step": 6105
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488171597706171e-06,
+      "loss": 0.1043,
+      "step": 6106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4850497621188835e-06,
+      "loss": 0.106,
+      "step": 6107
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4819284791177965e-06,
+      "loss": 0.1073,
+      "step": 6108
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.478807749084924e-06,
+      "loss": 0.109,
+      "step": 6109
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475687572402217e-06,
+      "loss": 0.1093,
+      "step": 6110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.472567949451556e-06,
+      "loss": 0.0967,
+      "step": 6111
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.469448880614741e-06,
+      "loss": 0.0981,
+      "step": 6112
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466330366273531e-06,
+      "loss": 0.1074,
+      "step": 6113
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.463212406809591e-06,
+      "loss": 0.1035,
+      "step": 6114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.0994,
+      "step": 6115
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.456978154039898e-06,
+      "loss": 0.1031,
+      "step": 6116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453861861497156e-06,
+      "loss": 0.103,
+      "step": 6117
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.450746125357712e-06,
+      "loss": 0.1008,
+      "step": 6118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4476309460029066e-06,
+      "loss": 0.1083,
+      "step": 6119
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.444516323813999e-06,
+      "loss": 0.121,
+      "step": 6120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4414022591721924e-06,
+      "loss": 0.1068,
+      "step": 6121
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438288752458619e-06,
+      "loss": 0.0988,
+      "step": 6122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435175804054342e-06,
+      "loss": 0.1027,
+      "step": 6123
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432063414340354e-06,
+      "loss": 0.1012,
+      "step": 6124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428951583697584e-06,
+      "loss": 0.0991,
+      "step": 6125
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425840312506894e-06,
+      "loss": 0.1084,
+      "step": 6126
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.422729601149059e-06,
+      "loss": 0.1118,
+      "step": 6127
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4196194500048186e-06,
+      "loss": 0.0998,
+      "step": 6128
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.416509859454812e-06,
+      "loss": 0.1097,
+      "step": 6129
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413400829879626e-06,
+      "loss": 0.1095,
+      "step": 6130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410292361659777e-06,
+      "loss": 0.1031,
+      "step": 6131
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407184455175712e-06,
+      "loss": 0.0991,
+      "step": 6132
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404077110807811e-06,
+      "loss": 0.1085,
+      "step": 6133
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40097032893637e-06,
+      "loss": 0.101,
+      "step": 6134
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397864109941649e-06,
+      "loss": 0.117,
+      "step": 6135
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.394758454203801e-06,
+      "loss": 0.102,
+      "step": 6136
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391653362102942e-06,
+      "loss": 0.1115,
+      "step": 6137
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.388548834019097e-06,
+      "loss": 0.1047,
+      "step": 6138
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.385444870332231e-06,
+      "loss": 0.1068,
+      "step": 6139
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382341471422242e-06,
+      "loss": 0.1103,
+      "step": 6140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.379238637668953e-06,
+      "loss": 0.1119,
+      "step": 6141
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.376136369452128e-06,
+      "loss": 0.1018,
+      "step": 6142
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.373034667151444e-06,
+      "loss": 0.1069,
+      "step": 6143
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369933531146525e-06,
+      "loss": 0.1009,
+      "step": 6144
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366832961816918e-06,
+      "loss": 0.1043,
+      "step": 6145
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363732959542105e-06,
+      "loss": 0.1068,
+      "step": 6146
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.360633524701494e-06,
+      "loss": 0.0998,
+      "step": 6147
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357534657674428e-06,
+      "loss": 0.1097,
+      "step": 6148
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.354436358840177e-06,
+      "loss": 0.1155,
+      "step": 6149
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3513386285779465e-06,
+      "loss": 0.1067,
+      "step": 6150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3482414672668625e-06,
+      "loss": 0.105,
+      "step": 6151
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.345144875285991e-06,
+      "loss": 0.1074,
+      "step": 6152
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.342048853014322e-06,
+      "loss": 0.116,
+      "step": 6153
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.338953400830782e-06,
+      "loss": 0.1099,
+      "step": 6154
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335858519114224e-06,
+      "loss": 0.1138,
+      "step": 6155
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33276420824343e-06,
+      "loss": 0.0997,
+      "step": 6156
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.329670468597119e-06,
+      "loss": 0.1001,
+      "step": 6157
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.326577300553923e-06,
+      "loss": 0.0954,
+      "step": 6158
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32348470449243e-06,
+      "loss": 0.1133,
+      "step": 6159
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3203926807911335e-06,
+      "loss": 0.1098,
+      "step": 6160
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317301229828471e-06,
+      "loss": 0.1006,
+      "step": 6161
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3142103519828045e-06,
+      "loss": 0.1051,
+      "step": 6162
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311120047632431e-06,
+      "loss": 0.1092,
+      "step": 6163
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.308030317155575e-06,
+      "loss": 0.1045,
+      "step": 6164
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304941160930378e-06,
+      "loss": 0.1151,
+      "step": 6165
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30185257933494e-06,
+      "loss": 0.1058,
+      "step": 6166
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298764572747256e-06,
+      "loss": 0.1014,
+      "step": 6167
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.295677141545284e-06,
+      "loss": 0.11,
+      "step": 6168
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.292590286106887e-06,
+      "loss": 0.1135,
+      "step": 6169
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2895040068098646e-06,
+      "loss": 0.1109,
+      "step": 6170
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.286418304031953e-06,
+      "loss": 0.1151,
+      "step": 6171
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283333178150809e-06,
+      "loss": 0.1048,
+      "step": 6172
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.280248629544027e-06,
+      "loss": 0.0799,
+      "step": 6173
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.277164658589119e-06,
+      "loss": 0.07,
+      "step": 6174
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274081265663535e-06,
+      "loss": 0.0881,
+      "step": 6175
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.270998451144654e-06,
+      "loss": 0.075,
+      "step": 6176
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267916215409783e-06,
+      "loss": 0.0766,
+      "step": 6177
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.264834558836156e-06,
+      "loss": 0.0754,
+      "step": 6178
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261753481800939e-06,
+      "loss": 0.0712,
+      "step": 6179
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2586729846812275e-06,
+      "loss": 0.0797,
+      "step": 6180
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255593067854047e-06,
+      "loss": 0.0723,
+      "step": 6181
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252513731696336e-06,
+      "loss": 0.0763,
+      "step": 6182
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.249434976584995e-06,
+      "loss": 0.0725,
+      "step": 6183
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246356802896818e-06,
+      "loss": 0.0726,
+      "step": 6184
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2432792110085494e-06,
+      "loss": 0.0726,
+      "step": 6185
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240202201296862e-06,
+      "loss": 0.0771,
+      "step": 6186
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237125774138338e-06,
+      "loss": 0.0709,
+      "step": 6187
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.234049929909519e-06,
+      "loss": 0.0703,
+      "step": 6188
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.230974668986844e-06,
+      "loss": 0.0733,
+      "step": 6189
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227899991746708e-06,
+      "loss": 0.0728,
+      "step": 6190
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.224825898565409e-06,
+      "loss": 0.0671,
+      "step": 6191
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.221752389819199e-06,
+      "loss": 0.0709,
+      "step": 6192
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218679465884238e-06,
+      "loss": 0.0699,
+      "step": 6193
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.215607127136621e-06,
+      "loss": 0.0716,
+      "step": 6194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.212535373952381e-06,
+      "loss": 0.0689,
+      "step": 6195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.209464206707455e-06,
+      "loss": 0.0759,
+      "step": 6196
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206393625777744e-06,
+      "loss": 0.0724,
+      "step": 6197
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.203323631539042e-06,
+      "loss": 0.0755,
+      "step": 6198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200254224367094e-06,
+      "loss": 0.0724,
+      "step": 6199
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197185404637562e-06,
+      "loss": 0.0674,
+      "step": 6200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194117172726042e-06,
+      "loss": 0.0726,
+      "step": 6201
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1910495290080565e-06,
+      "loss": 0.0681,
+      "step": 6202
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.187982473859052e-06,
+      "loss": 0.0708,
+      "step": 6203
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.184916007654411e-06,
+      "loss": 0.0677,
+      "step": 6204
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181850130769441e-06,
+      "loss": 0.075,
+      "step": 6205
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178784843579368e-06,
+      "loss": 0.0809,
+      "step": 6206
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1757201464593556e-06,
+      "loss": 0.0704,
+      "step": 6207
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172656039784495e-06,
+      "loss": 0.065,
+      "step": 6208
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169592523929803e-06,
+      "loss": 0.0745,
+      "step": 6209
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166529599270228e-06,
+      "loss": 0.0784,
+      "step": 6210
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163467266180631e-06,
+      "loss": 0.0659,
+      "step": 6211
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.160405525035828e-06,
+      "loss": 0.0709,
+      "step": 6212
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.157344376210528e-06,
+      "loss": 0.0676,
+      "step": 6213
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.154283820079404e-06,
+      "loss": 0.0661,
+      "step": 6214
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.151223857017027e-06,
+      "loss": 0.0728,
+      "step": 6215
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.148164487397911e-06,
+      "loss": 0.0759,
+      "step": 6216
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.145105711596494e-06,
+      "loss": 0.0736,
+      "step": 6217
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.142047529987133e-06,
+      "loss": 0.0693,
+      "step": 6218
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.138989942944133e-06,
+      "loss": 0.0742,
+      "step": 6219
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.135932950841698e-06,
+      "loss": 0.0724,
+      "step": 6220
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132876554053989e-06,
+      "loss": 0.0694,
+      "step": 6221
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1298207529550655e-06,
+      "loss": 0.0729,
+      "step": 6222
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.126765547918941e-06,
+      "loss": 0.0691,
+      "step": 6223
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123710939319535e-06,
+      "loss": 0.0679,
+      "step": 6224
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1206569275307035e-06,
+      "loss": 0.0769,
+      "step": 6225
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117603512926228e-06,
+      "loss": 0.0685,
+      "step": 6226
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.114550695879819e-06,
+      "loss": 0.0728,
+      "step": 6227
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111498476765111e-06,
+      "loss": 0.0834,
+      "step": 6228
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108446855955662e-06,
+      "loss": 0.0729,
+      "step": 6229
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.105395833824964e-06,
+      "loss": 0.0707,
+      "step": 6230
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102345410746435e-06,
+      "loss": 0.0771,
+      "step": 6231
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0992955870934114e-06,
+      "loss": 0.0679,
+      "step": 6232
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.096246363239166e-06,
+      "loss": 0.0796,
+      "step": 6233
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.093197739556895e-06,
+      "loss": 0.0677,
+      "step": 6234
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0901497164197185e-06,
+      "loss": 0.0682,
+      "step": 6235
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08710229420069e-06,
+      "loss": 0.0714,
+      "step": 6236
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.084055473272775e-06,
+      "loss": 0.0722,
+      "step": 6237
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081009254008882e-06,
+      "loss": 0.0695,
+      "step": 6238
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077963636781835e-06,
+      "loss": 0.0675,
+      "step": 6239
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.074918621964391e-06,
+      "loss": 0.0708,
+      "step": 6240
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071874209929233e-06,
+      "loss": 0.0724,
+      "step": 6241
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068830401048954e-06,
+      "loss": 0.0674,
+      "step": 6242
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.065787195696106e-06,
+      "loss": 0.0696,
+      "step": 6243
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.062744594243131e-06,
+      "loss": 0.0722,
+      "step": 6244
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.059702597062428e-06,
+      "loss": 0.0743,
+      "step": 6245
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0566612045262985e-06,
+      "loss": 0.0699,
+      "step": 6246
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053620417006981e-06,
+      "loss": 0.067,
+      "step": 6247
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.050580234876641e-06,
+      "loss": 0.075,
+      "step": 6248
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.047540658507366e-06,
+      "loss": 0.0703,
+      "step": 6249
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044501688271175e-06,
+      "loss": 0.0752,
+      "step": 6250
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.041463324539996e-06,
+      "loss": 0.0746,
+      "step": 6251
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.038425567685712e-06,
+      "loss": 0.0719,
+      "step": 6252
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0353884180801024e-06,
+      "loss": 0.0669,
+      "step": 6253
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0323518760948895e-06,
+      "loss": 0.0781,
+      "step": 6254
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029315942101716e-06,
+      "loss": 0.0705,
+      "step": 6255
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.026280616472152e-06,
+      "loss": 0.0738,
+      "step": 6256
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02324589957769e-06,
+      "loss": 0.0687,
+      "step": 6257
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020211791789753e-06,
+      "loss": 0.0741,
+      "step": 6258
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0171782934796865e-06,
+      "loss": 0.0689,
+      "step": 6259
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.014145405018757e-06,
+      "loss": 0.0663,
+      "step": 6260
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011113126778162e-06,
+      "loss": 0.0689,
+      "step": 6261
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008081459129025e-06,
+      "loss": 0.0632,
+      "step": 6262
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.005050402442391e-06,
+      "loss": 0.0787,
+      "step": 6263
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.002019957089233e-06,
+      "loss": 0.0692,
+      "step": 6264
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9989901234404535e-06,
+      "loss": 0.0742,
+      "step": 6265
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.99596090186686e-06,
+      "loss": 0.0709,
+      "step": 6266
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9929322927392174e-06,
+      "loss": 0.0782,
+      "step": 6267
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.989904296428186e-06,
+      "loss": 0.0698,
+      "step": 6268
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.986876913304368e-06,
+      "loss": 0.0701,
+      "step": 6269
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.983850143738285e-06,
+      "loss": 0.0756,
+      "step": 6270
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.980823988100383e-06,
+      "loss": 0.0771,
+      "step": 6271
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.977798446761041e-06,
+      "loss": 0.0643,
+      "step": 6272
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974773520090541e-06,
+      "loss": 0.0775,
+      "step": 6273
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971749208459123e-06,
+      "loss": 0.0693,
+      "step": 6274
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968725512236917e-06,
+      "loss": 0.0764,
+      "step": 6275
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9657024317940085e-06,
+      "loss": 0.0726,
+      "step": 6276
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.962679967500384e-06,
+      "loss": 0.0713,
+      "step": 6277
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.959658119725965e-06,
+      "loss": 0.0733,
+      "step": 6278
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956636888840599e-06,
+      "loss": 0.068,
+      "step": 6279
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.953616275214054e-06,
+      "loss": 0.0801,
+      "step": 6280
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9505962792160275e-06,
+      "loss": 0.071,
+      "step": 6281
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947576901216128e-06,
+      "loss": 0.073,
+      "step": 6282
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.94455814158391e-06,
+      "loss": 0.071,
+      "step": 6283
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.941540000688834e-06,
+      "loss": 0.0698,
+      "step": 6284
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938522478900291e-06,
+      "loss": 0.0772,
+      "step": 6285
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935505576587597e-06,
+      "loss": 0.0683,
+      "step": 6286
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932489294119993e-06,
+      "loss": 0.0689,
+      "step": 6287
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9294736318666415e-06,
+      "loss": 0.0739,
+      "step": 6288
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926458590196631e-06,
+      "loss": 0.0758,
+      "step": 6289
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923444169478978e-06,
+      "loss": 0.0748,
+      "step": 6290
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92043037008261e-06,
+      "loss": 0.0672,
+      "step": 6291
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9174171923763895e-06,
+      "loss": 0.0734,
+      "step": 6292
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9144046367291025e-06,
+      "loss": 0.074,
+      "step": 6293
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.911392703509454e-06,
+      "loss": 0.0709,
+      "step": 6294
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.908381393086079e-06,
+      "loss": 0.0692,
+      "step": 6295
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.905370705827534e-06,
+      "loss": 0.0744,
+      "step": 6296
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902360642102286e-06,
+      "loss": 0.0696,
+      "step": 6297
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.899351202278756e-06,
+      "loss": 0.0729,
+      "step": 6298
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.896342386725256e-06,
+      "loss": 0.0692,
+      "step": 6299
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89333419581004e-06,
+      "loss": 0.0753,
+      "step": 6300
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890326629901283e-06,
+      "loss": 0.0785,
+      "step": 6301
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.887319689367081e-06,
+      "loss": 0.0685,
+      "step": 6302
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884313374575458e-06,
+      "loss": 0.0734,
+      "step": 6303
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8813076858943465e-06,
+      "loss": 0.068,
+      "step": 6304
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.878302623691629e-06,
+      "loss": 0.0715,
+      "step": 6305
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875298188335081e-06,
+      "loss": 0.0687,
+      "step": 6306
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872294380192432e-06,
+      "loss": 0.0725,
+      "step": 6307
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.869291199631306e-06,
+      "loss": 0.0724,
+      "step": 6308
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866288647019269e-06,
+      "loss": 0.0698,
+      "step": 6309
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.863286722723802e-06,
+      "loss": 0.0715,
+      "step": 6310
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8602854271123145e-06,
+      "loss": 0.0777,
+      "step": 6311
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8572847605521366e-06,
+      "loss": 0.0715,
+      "step": 6312
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.854284723410513e-06,
+      "loss": 0.0764,
+      "step": 6313
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851285316054631e-06,
+      "loss": 0.0818,
+      "step": 6314
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.84828653885158e-06,
+      "loss": 0.0711,
+      "step": 6315
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.845288392168384e-06,
+      "loss": 0.0696,
+      "step": 6316
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.842290876371987e-06,
+      "loss": 0.0731,
+      "step": 6317
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839293991829256e-06,
+      "loss": 0.0649,
+      "step": 6318
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.836297738906981e-06,
+      "loss": 0.0749,
+      "step": 6319
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833302117971879e-06,
+      "loss": 0.0741,
+      "step": 6320
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8303071293905754e-06,
+      "loss": 0.0797,
+      "step": 6321
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.827312773529634e-06,
+      "loss": 0.071,
+      "step": 6322
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824319050755531e-06,
+      "loss": 0.0773,
+      "step": 6323
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821325961434674e-06,
+      "loss": 0.0773,
+      "step": 6324
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8183335059333846e-06,
+      "loss": 0.0693,
+      "step": 6325
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8153416846179125e-06,
+      "loss": 0.0652,
+      "step": 6326
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812350497854431e-06,
+      "loss": 0.0684,
+      "step": 6327
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80935994600902e-06,
+      "loss": 0.0794,
+      "step": 6328
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.80637002944771e-06,
+      "loss": 0.0706,
+      "step": 6329
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.803380748536422e-06,
+      "loss": 0.0683,
+      "step": 6330
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.800392103641032e-06,
+      "loss": 0.0643,
+      "step": 6331
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7974040951273075e-06,
+      "loss": 0.0696,
+      "step": 6332
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794416723360957e-06,
+      "loss": 0.0623,
+      "step": 6333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.79142998870761e-06,
+      "loss": 0.0661,
+      "step": 6334
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788443891532801e-06,
+      "loss": 0.0684,
+      "step": 6335
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785458432202016e-06,
+      "loss": 0.0727,
+      "step": 6336
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.78247361108063e-06,
+      "loss": 0.0767,
+      "step": 6337
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779489428533973e-06,
+      "loss": 0.073,
+      "step": 6338
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.776505884927267e-06,
+      "loss": 0.0693,
+      "step": 6339
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7735229806256745e-06,
+      "loss": 0.0708,
+      "step": 6340
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.770540715994272e-06,
+      "loss": 0.0701,
+      "step": 6341
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.76755909139806e-06,
+      "loss": 0.0656,
+      "step": 6342
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.764578107201967e-06,
+      "loss": 0.0709,
+      "step": 6343
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.761597763770825e-06,
+      "loss": 0.072,
+      "step": 6344
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.758618061469407e-06,
+      "loss": 0.068,
+      "step": 6345
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7556390006623966e-06,
+      "loss": 0.0694,
+      "step": 6346
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.752660581714403e-06,
+      "loss": 0.0729,
+      "step": 6347
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7496828049899565e-06,
+      "loss": 0.0666,
+      "step": 6348
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.746705670853508e-06,
+      "loss": 0.07,
+      "step": 6349
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743729179669429e-06,
+      "loss": 0.0711,
+      "step": 6350
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.740753331802017e-06,
+      "loss": 0.0732,
+      "step": 6351
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737778127615478e-06,
+      "loss": 0.0728,
+      "step": 6352
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.73480356747396e-06,
+      "loss": 0.068,
+      "step": 6353
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731829651741511e-06,
+      "loss": 0.0748,
+      "step": 6354
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728856380782112e-06,
+      "loss": 0.0848,
+      "step": 6355
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.725883754959668e-06,
+      "loss": 0.0708,
+      "step": 6356
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722911774637987e-06,
+      "loss": 0.0666,
+      "step": 6357
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.719940440180827e-06,
+      "loss": 0.0736,
+      "step": 6358
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.716969751951833e-06,
+      "loss": 0.0722,
+      "step": 6359
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713999710314606e-06,
+      "loss": 0.0656,
+      "step": 6360
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.711030315632636e-06,
+      "loss": 0.0703,
+      "step": 6361
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70806156826936e-06,
+      "loss": 0.0705,
+      "step": 6362
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.705093468588114e-06,
+      "loss": 0.0668,
+      "step": 6363
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.70212601695217e-06,
+      "loss": 0.0777,
+      "step": 6364
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.699159213724713e-06,
+      "loss": 0.067,
+      "step": 6365
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696193059268853e-06,
+      "loss": 0.0702,
+      "step": 6366
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.693227553947621e-06,
+      "loss": 0.073,
+      "step": 6367
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.690262698123956e-06,
+      "loss": 0.0778,
+      "step": 6368
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687298492160742e-06,
+      "loss": 0.0681,
+      "step": 6369
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.684334936420757e-06,
+      "loss": 0.0664,
+      "step": 6370
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.681372031266717e-06,
+      "loss": 0.0749,
+      "step": 6371
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.678409777061252e-06,
+      "loss": 0.0672,
+      "step": 6372
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.675448174166912e-06,
+      "loss": 0.0741,
+      "step": 6373
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6724872229461716e-06,
+      "loss": 0.0708,
+      "step": 6374
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669526923761424e-06,
+      "loss": 0.0773,
+      "step": 6375
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.666567276974974e-06,
+      "loss": 0.0688,
+      "step": 6376
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663608282949057e-06,
+      "loss": 0.07,
+      "step": 6377
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.660649942045826e-06,
+      "loss": 0.0726,
+      "step": 6378
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.657692254627354e-06,
+      "loss": 0.0711,
+      "step": 6379
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6547352210556325e-06,
+      "loss": 0.0705,
+      "step": 6380
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651778841692574e-06,
+      "loss": 0.0715,
+      "step": 6381
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.648823116900013e-06,
+      "loss": 0.068,
+      "step": 6382
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6458680470396925e-06,
+      "loss": 0.0737,
+      "step": 6383
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642913632473298e-06,
+      "loss": 0.0691,
+      "step": 6384
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6399598735624105e-06,
+      "loss": 0.0705,
+      "step": 6385
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637006770668547e-06,
+      "loss": 0.0724,
+      "step": 6386
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634054324153134e-06,
+      "loss": 0.0777,
+      "step": 6387
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.631102534377527e-06,
+      "loss": 0.0761,
+      "step": 6388
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6281514017029994e-06,
+      "loss": 0.0718,
+      "step": 6389
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625200926490728e-06,
+      "loss": 0.0781,
+      "step": 6390
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.622251109101838e-06,
+      "loss": 0.071,
+      "step": 6391
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619301949897346e-06,
+      "loss": 0.0713,
+      "step": 6392
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.616353449238212e-06,
+      "loss": 0.0672,
+      "step": 6393
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613405607485295e-06,
+      "loss": 0.0726,
+      "step": 6394
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6104584249993855e-06,
+      "loss": 0.0761,
+      "step": 6395
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607511902141191e-06,
+      "loss": 0.064,
+      "step": 6396
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.604566039271334e-06,
+      "loss": 0.0709,
+      "step": 6397
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.601620836750367e-06,
+      "loss": 0.0727,
+      "step": 6398
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598676294938746e-06,
+      "loss": 0.071,
+      "step": 6399
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.595732414196856e-06,
+      "loss": 0.0764,
+      "step": 6400
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.592789194885002e-06,
+      "loss": 0.0679,
+      "step": 6401
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.589846637363405e-06,
+      "loss": 0.0756,
+      "step": 6402
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.586904741992205e-06,
+      "loss": 0.0688,
+      "step": 6403
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.583963509131463e-06,
+      "loss": 0.0792,
+      "step": 6404
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.581022939141154e-06,
+      "loss": 0.0721,
+      "step": 6405
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.578083032381182e-06,
+      "loss": 0.0733,
+      "step": 6406
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.575143789211356e-06,
+      "loss": 0.0648,
+      "step": 6407
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572205209991412e-06,
+      "loss": 0.076,
+      "step": 6408
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.569267295081006e-06,
+      "loss": 0.0744,
+      "step": 6409
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5663300448397094e-06,
+      "loss": 0.0774,
+      "step": 6410
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5633934596270165e-06,
+      "loss": 0.0731,
+      "step": 6411
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5604575398023266e-06,
+      "loss": 0.0757,
+      "step": 6412
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5575222857249825e-06,
+      "loss": 0.0735,
+      "step": 6413
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.554587697754216e-06,
+      "loss": 0.0695,
+      "step": 6414
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.551653776249207e-06,
+      "loss": 0.0661,
+      "step": 6415
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.548720521569028e-06,
+      "loss": 0.0699,
+      "step": 6416
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545787934072685e-06,
+      "loss": 0.0729,
+      "step": 6417
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.542856014119098e-06,
+      "loss": 0.0684,
+      "step": 6418
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.539924762067105e-06,
+      "loss": 0.0747,
+      "step": 6419
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.536994178275468e-06,
+      "loss": 0.0691,
+      "step": 6420
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5340642631028484e-06,
+      "loss": 0.0765,
+      "step": 6421
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531135016907858e-06,
+      "loss": 0.0757,
+      "step": 6422
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.528206440048995e-06,
+      "loss": 0.0695,
+      "step": 6423
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.525278532884691e-06,
+      "loss": 0.069,
+      "step": 6424
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522351295773296e-06,
+      "loss": 0.076,
+      "step": 6425
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.519424729073075e-06,
+      "loss": 0.0751,
+      "step": 6426
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.516498833142211e-06,
+      "loss": 0.066,
+      "step": 6427
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513573608338806e-06,
+      "loss": 0.073,
+      "step": 6428
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510649055020881e-06,
+      "loss": 0.0718,
+      "step": 6429
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.507725173546368e-06,
+      "loss": 0.0693,
+      "step": 6430
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5048019642731245e-06,
+      "loss": 0.0721,
+      "step": 6431
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.501879427558923e-06,
+      "loss": 0.0704,
+      "step": 6432
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498957563761452e-06,
+      "loss": 0.0729,
+      "step": 6433
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496036373238322e-06,
+      "loss": 0.0738,
+      "step": 6434
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4931158563470575e-06,
+      "loss": 0.0637,
+      "step": 6435
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.490196013445102e-06,
+      "loss": 0.0749,
+      "step": 6436
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.487276844889819e-06,
+      "loss": 0.0692,
+      "step": 6437
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.484358351038478e-06,
+      "loss": 0.0639,
+      "step": 6438
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48144053224828e-06,
+      "loss": 0.0677,
+      "step": 6439
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.478523388876338e-06,
+      "loss": 0.0681,
+      "step": 6440
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475606921279679e-06,
+      "loss": 0.0682,
+      "step": 6441
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.472691129815259e-06,
+      "loss": 0.0742,
+      "step": 6442
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469776014839926e-06,
+      "loss": 0.0725,
+      "step": 6443
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466861576710482e-06,
+      "loss": 0.0692,
+      "step": 6444
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.463947815783607e-06,
+      "loss": 0.0724,
+      "step": 6445
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461034732415934e-06,
+      "loss": 0.0679,
+      "step": 6446
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.458122326963985e-06,
+      "loss": 0.0628,
+      "step": 6447
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455210599784213e-06,
+      "loss": 0.0756,
+      "step": 6448
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452299551232987e-06,
+      "loss": 0.074,
+      "step": 6449
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.449389181666589e-06,
+      "loss": 0.0702,
+      "step": 6450
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4464794914412245e-06,
+      "loss": 0.069,
+      "step": 6451
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.443570480913002e-06,
+      "loss": 0.0649,
+      "step": 6452
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440662150437968e-06,
+      "loss": 0.0723,
+      "step": 6453
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.437754500372066e-06,
+      "loss": 0.071,
+      "step": 6454
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434847531071166e-06,
+      "loss": 0.0672,
+      "step": 6455
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4319412428910534e-06,
+      "loss": 0.0689,
+      "step": 6456
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4290356361874284e-06,
+      "loss": 0.0736,
+      "step": 6457
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426130711315913e-06,
+      "loss": 0.0727,
+      "step": 6458
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.423226468632037e-06,
+      "loss": 0.0656,
+      "step": 6459
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.420322908491258e-06,
+      "loss": 0.0763,
+      "step": 6460
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417420031248937e-06,
+      "loss": 0.0755,
+      "step": 6461
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414517837260359e-06,
+      "loss": 0.0636,
+      "step": 6462
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4116163268807275e-06,
+      "loss": 0.0703,
+      "step": 6463
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408715500465157e-06,
+      "loss": 0.0742,
+      "step": 6464
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4058153583686815e-06,
+      "loss": 0.0642,
+      "step": 6465
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.402915900946254e-06,
+      "loss": 0.0773,
+      "step": 6466
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.400017128552729e-06,
+      "loss": 0.0694,
+      "step": 6467
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397119041542903e-06,
+      "loss": 0.0738,
+      "step": 6468
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394221640271459e-06,
+      "loss": 0.0711,
+      "step": 6469
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391324925093025e-06,
+      "loss": 0.0697,
+      "step": 6470
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.38842889636212e-06,
+      "loss": 0.0794,
+      "step": 6471
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.385533554433195e-06,
+      "loss": 0.0725,
+      "step": 6472
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.382638899660613e-06,
+      "loss": 0.0769,
+      "step": 6473
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.379744932398642e-06,
+      "loss": 0.0692,
+      "step": 6474
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.376851653001492e-06,
+      "loss": 0.0732,
+      "step": 6475
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373959061823254e-06,
+      "loss": 0.0734,
+      "step": 6476
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.371067159217971e-06,
+      "loss": 0.0666,
+      "step": 6477
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368175945539572e-06,
+      "loss": 0.0696,
+      "step": 6478
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365285421141916e-06,
+      "loss": 0.0727,
+      "step": 6479
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.362395586378777e-06,
+      "loss": 0.0675,
+      "step": 6480
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3595064416038425e-06,
+      "loss": 0.0659,
+      "step": 6481
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356617987170718e-06,
+      "loss": 0.0712,
+      "step": 6482
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.353730223432913e-06,
+      "loss": 0.0778,
+      "step": 6483
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.350843150743875e-06,
+      "loss": 0.0723,
+      "step": 6484
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347956769456944e-06,
+      "loss": 0.0697,
+      "step": 6485
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3450710799253885e-06,
+      "loss": 0.0702,
+      "step": 6486
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342186082502389e-06,
+      "loss": 0.0697,
+      "step": 6487
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339301777541041e-06,
+      "loss": 0.0706,
+      "step": 6488
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.336418165394356e-06,
+      "loss": 0.0726,
+      "step": 6489
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.333535246415262e-06,
+      "loss": 0.0673,
+      "step": 6490
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330653020956593e-06,
+      "loss": 0.0756,
+      "step": 6491
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.327771489371116e-06,
+      "loss": 0.0713,
+      "step": 6492
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.324890652011495e-06,
+      "loss": 0.0685,
+      "step": 6493
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322010509230319e-06,
+      "loss": 0.0726,
+      "step": 6494
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.319131061380089e-06,
+      "loss": 0.067,
+      "step": 6495
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.316252308813223e-06,
+      "loss": 0.0796,
+      "step": 6496
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3133742518820535e-06,
+      "loss": 0.0694,
+      "step": 6497
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3104968909388174e-06,
+      "loss": 0.0745,
+      "step": 6498
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.307620226335691e-06,
+      "loss": 0.0747,
+      "step": 6499
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304744258424735e-06,
+      "loss": 0.0701,
+      "step": 6500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3018689875579555e-06,
+      "loss": 0.0694,
+      "step": 6501
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298994414087246e-06,
+      "loss": 0.0675,
+      "step": 6502
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29612053836443e-06,
+      "loss": 0.0678,
+      "step": 6503
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.293247360741245e-06,
+      "loss": 0.0681,
+      "step": 6504
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290374881569331e-06,
+      "loss": 0.0694,
+      "step": 6505
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.287503101200266e-06,
+      "loss": 0.0684,
+      "step": 6506
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.284632019985512e-06,
+      "loss": 0.0699,
+      "step": 6507
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281761638276476e-06,
+      "loss": 0.0697,
+      "step": 6508
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2788919564244565e-06,
+      "loss": 0.0704,
+      "step": 6509
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2760229747806755e-06,
+      "loss": 0.0679,
+      "step": 6510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273154693696272e-06,
+      "loss": 0.0728,
+      "step": 6511
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.270287113522292e-06,
+      "loss": 0.0696,
+      "step": 6512
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.267420234609702e-06,
+      "loss": 0.0697,
+      "step": 6513
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.26455405730938e-06,
+      "loss": 0.0726,
+      "step": 6514
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.261688581972122e-06,
+      "loss": 0.0751,
+      "step": 6515
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258823808948626e-06,
+      "loss": 0.0614,
+      "step": 6516
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.255959738589517e-06,
+      "loss": 0.0743,
+      "step": 6517
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.253096371245329e-06,
+      "loss": 0.0677,
+      "step": 6518
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250233707266511e-06,
+      "loss": 0.0762,
+      "step": 6519
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247371747003426e-06,
+      "loss": 0.0682,
+      "step": 6520
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244510490806352e-06,
+      "loss": 0.0633,
+      "step": 6521
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.24164993902547e-06,
+      "loss": 0.0687,
+      "step": 6522
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238790092010897e-06,
+      "loss": 0.076,
+      "step": 6523
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23593095011264e-06,
+      "loss": 0.0703,
+      "step": 6524
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233072513680635e-06,
+      "loss": 0.0724,
+      "step": 6525
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230214783064726e-06,
+      "loss": 0.0707,
+      "step": 6526
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.227357758614672e-06,
+      "loss": 0.0776,
+      "step": 6527
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.224501440680148e-06,
+      "loss": 0.0725,
+      "step": 6528
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221645829610729e-06,
+      "loss": 0.0652,
+      "step": 6529
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.218790925755929e-06,
+      "loss": 0.0758,
+      "step": 6530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.215936729465147e-06,
+      "loss": 0.0715,
+      "step": 6531
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213083241087721e-06,
+      "loss": 0.0657,
+      "step": 6532
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210230460972882e-06,
+      "loss": 0.0665,
+      "step": 6533
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.207378389469784e-06,
+      "loss": 0.0689,
+      "step": 6534
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204527026927495e-06,
+      "loss": 0.0652,
+      "step": 6535
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.201676373694994e-06,
+      "loss": 0.0799,
+      "step": 6536
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1988264301211756e-06,
+      "loss": 0.0805,
+      "step": 6537
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.195977196554835e-06,
+      "loss": 0.0729,
+      "step": 6538
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193128673344707e-06,
+      "loss": 0.0678,
+      "step": 6539
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190280860839409e-06,
+      "loss": 0.0756,
+      "step": 6540
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187433759387491e-06,
+      "loss": 0.0783,
+      "step": 6541
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.184587369337413e-06,
+      "loss": 0.0609,
+      "step": 6542
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181741691037542e-06,
+      "loss": 0.0663,
+      "step": 6543
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178896724836163e-06,
+      "loss": 0.0694,
+      "step": 6544
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.176052471081477e-06,
+      "loss": 0.0781,
+      "step": 6545
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173208930121582e-06,
+      "loss": 0.0705,
+      "step": 6546
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170366102304508e-06,
+      "loss": 0.0708,
+      "step": 6547
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.167523987978188e-06,
+      "loss": 0.076,
+      "step": 6548
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.164682587490469e-06,
+      "loss": 0.067,
+      "step": 6549
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161841901189112e-06,
+      "loss": 0.0778,
+      "step": 6550
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.159001929421788e-06,
+      "loss": 0.0731,
+      "step": 6551
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156162672536085e-06,
+      "loss": 0.0672,
+      "step": 6552
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153324130879491e-06,
+      "loss": 0.0784,
+      "step": 6553
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.150486304799431e-06,
+      "loss": 0.0798,
+      "step": 6554
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.147649194643218e-06,
+      "loss": 0.0661,
+      "step": 6555
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1448128007580855e-06,
+      "loss": 0.076,
+      "step": 6556
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.141977123491185e-06,
+      "loss": 0.071,
+      "step": 6557
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.139142163189573e-06,
+      "loss": 0.065,
+      "step": 6558
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136307920200228e-06,
+      "loss": 0.0738,
+      "step": 6559
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.133474394870021e-06,
+      "loss": 0.0739,
+      "step": 6560
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130641587545763e-06,
+      "loss": 0.072,
+      "step": 6561
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.127809498574147e-06,
+      "loss": 0.0717,
+      "step": 6562
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124978128301809e-06,
+      "loss": 0.0707,
+      "step": 6563
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.0717,
+      "step": 6564
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119317545240978e-06,
+      "loss": 0.0729,
+      "step": 6565
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1164883331452876e-06,
+      "loss": 0.0687,
+      "step": 6566
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1136598411344695e-06,
+      "loss": 0.0711,
+      "step": 6567
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1108320695547065e-06,
+      "loss": 0.0699,
+      "step": 6568
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.108005018752083e-06,
+      "loss": 0.073,
+      "step": 6569
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105178689072605e-06,
+      "loss": 0.0676,
+      "step": 6570
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1023530808621905e-06,
+      "loss": 0.0724,
+      "step": 6571
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.099528194466664e-06,
+      "loss": 0.0706,
+      "step": 6572
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.096704030231767e-06,
+      "loss": 0.0738,
+      "step": 6573
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093880588503147e-06,
+      "loss": 0.0746,
+      "step": 6574
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.091057869626369e-06,
+      "loss": 0.0718,
+      "step": 6575
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0882358739469075e-06,
+      "loss": 0.0673,
+      "step": 6576
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.08541460181014e-06,
+      "loss": 0.0708,
+      "step": 6577
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082594053561369e-06,
+      "loss": 0.0705,
+      "step": 6578
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.079774229545801e-06,
+      "loss": 0.0626,
+      "step": 6579
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076955130108554e-06,
+      "loss": 0.0628,
+      "step": 6580
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07413675559466e-06,
+      "loss": 0.0715,
+      "step": 6581
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0713191063490605e-06,
+      "loss": 0.0765,
+      "step": 6582
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0685021827166125e-06,
+      "loss": 0.0653,
+      "step": 6583
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.065685985042068e-06,
+      "loss": 0.0644,
+      "step": 6584
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062870513670117e-06,
+      "loss": 0.0694,
+      "step": 6585
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060055768945337e-06,
+      "loss": 0.0727,
+      "step": 6586
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.057241751212226e-06,
+      "loss": 0.0782,
+      "step": 6587
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054428460815196e-06,
+      "loss": 0.0692,
+      "step": 6588
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.051615898098564e-06,
+      "loss": 0.0722,
+      "step": 6589
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.048804063406565e-06,
+      "loss": 0.07,
+      "step": 6590
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04599295708333e-06,
+      "loss": 0.067,
+      "step": 6591
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.043182579472924e-06,
+      "loss": 0.074,
+      "step": 6592
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.040372930919298e-06,
+      "loss": 0.0699,
+      "step": 6593
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.037564011766337e-06,
+      "loss": 0.0645,
+      "step": 6594
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034755822357818e-06,
+      "loss": 0.0678,
+      "step": 6595
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031948363037437e-06,
+      "loss": 0.0724,
+      "step": 6596
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029141634148802e-06,
+      "loss": 0.0777,
+      "step": 6597
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026335636035429e-06,
+      "loss": 0.0729,
+      "step": 6598
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0235303690407466e-06,
+      "loss": 0.0718,
+      "step": 6599
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.020725833508088e-06,
+      "loss": 0.072,
+      "step": 6600
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017922029780703e-06,
+      "loss": 0.0716,
+      "step": 6601
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.015118958201752e-06,
+      "loss": 0.077,
+      "step": 6602
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0123166191143e-06,
+      "loss": 0.0698,
+      "step": 6603
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00951501286133e-06,
+      "loss": 0.074,
+      "step": 6604
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0067141397857305e-06,
+      "loss": 0.0667,
+      "step": 6605
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.003914000230299e-06,
+      "loss": 0.0627,
+      "step": 6606
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0011145945377515e-06,
+      "loss": 0.0765,
+      "step": 6607
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9983159230507014e-06,
+      "loss": 0.0722,
+      "step": 6608
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995517986111681e-06,
+      "loss": 0.069,
+      "step": 6609
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.992720784063131e-06,
+      "loss": 0.0675,
+      "step": 6610
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.989924317247402e-06,
+      "loss": 0.0736,
+      "step": 6611
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987128586006757e-06,
+      "loss": 0.0673,
+      "step": 6612
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.984333590683359e-06,
+      "loss": 0.0702,
+      "step": 6613
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.981539331619299e-06,
+      "loss": 0.0721,
+      "step": 6614
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.978745809156556e-06,
+      "loss": 0.0749,
+      "step": 6615
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975953023637042e-06,
+      "loss": 0.0779,
+      "step": 6616
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973160975402552e-06,
+      "loss": 0.0697,
+      "step": 6617
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970369664794823e-06,
+      "loss": 0.0704,
+      "step": 6618
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.967579092155473e-06,
+      "loss": 0.0769,
+      "step": 6619
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.964789257826042e-06,
+      "loss": 0.0673,
+      "step": 6620
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962000162147984e-06,
+      "loss": 0.0763,
+      "step": 6621
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.959211805462647e-06,
+      "loss": 0.0734,
+      "step": 6622
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9564241881113145e-06,
+      "loss": 0.0743,
+      "step": 6623
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9536373104351485e-06,
+      "loss": 0.0735,
+      "step": 6624
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9508511727752444e-06,
+      "loss": 0.0681,
+      "step": 6625
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948065775472596e-06,
+      "loss": 0.0682,
+      "step": 6626
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945281118868109e-06,
+      "loss": 0.0731,
+      "step": 6627
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9424972033026e-06,
+      "loss": 0.0687,
+      "step": 6628
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939714029116791e-06,
+      "loss": 0.072,
+      "step": 6629
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93693159665132e-06,
+      "loss": 0.0763,
+      "step": 6630
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9341499062467235e-06,
+      "loss": 0.069,
+      "step": 6631
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931368958243456e-06,
+      "loss": 0.0722,
+      "step": 6632
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928588752981881e-06,
+      "loss": 0.0716,
+      "step": 6633
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.925809290802265e-06,
+      "loss": 0.0717,
+      "step": 6634
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92303057204479e-06,
+      "loss": 0.0691,
+      "step": 6635
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.920252597049547e-06,
+      "loss": 0.0737,
+      "step": 6636
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9174753661565215e-06,
+      "loss": 0.0772,
+      "step": 6637
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.914698879705635e-06,
+      "loss": 0.0634,
+      "step": 6638
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.911923138036689e-06,
+      "loss": 0.065,
+      "step": 6639
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90914814148942e-06,
+      "loss": 0.0698,
+      "step": 6640
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.906373890403452e-06,
+      "loss": 0.0701,
+      "step": 6641
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903600385118328e-06,
+      "loss": 0.0742,
+      "step": 6642
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900827625973503e-06,
+      "loss": 0.0774,
+      "step": 6643
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898055613308325e-06,
+      "loss": 0.0669,
+      "step": 6644
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895284347462075e-06,
+      "loss": 0.0762,
+      "step": 6645
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892513828773916e-06,
+      "loss": 0.0704,
+      "step": 6646
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.889744057582947e-06,
+      "loss": 0.0835,
+      "step": 6647
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.88697503422815e-06,
+      "loss": 0.075,
+      "step": 6648
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.884206759048429e-06,
+      "loss": 0.0702,
+      "step": 6649
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.881439232382595e-06,
+      "loss": 0.0704,
+      "step": 6650
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.878672454569369e-06,
+      "loss": 0.0689,
+      "step": 6651
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875906425947374e-06,
+      "loss": 0.0746,
+      "step": 6652
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873141146855146e-06,
+      "loss": 0.0789,
+      "step": 6653
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870376617631133e-06,
+      "loss": 0.0687,
+      "step": 6654
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.86761283861368e-06,
+      "loss": 0.0758,
+      "step": 6655
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.864849810141048e-06,
+      "loss": 0.0703,
+      "step": 6656
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862087532551407e-06,
+      "loss": 0.0726,
+      "step": 6657
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859326006182833e-06,
+      "loss": 0.0781,
+      "step": 6658
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85656523137331e-06,
+      "loss": 0.0703,
+      "step": 6659
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.853805208460728e-06,
+      "loss": 0.0689,
+      "step": 6660
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851045937782889e-06,
+      "loss": 0.0785,
+      "step": 6661
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8482874196775055e-06,
+      "loss": 0.0738,
+      "step": 6662
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845529654482184e-06,
+      "loss": 0.0739,
+      "step": 6663
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.842772642534454e-06,
+      "loss": 0.0662,
+      "step": 6664
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840016384171744e-06,
+      "loss": 0.0714,
+      "step": 6665
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8372608797313974e-06,
+      "loss": 0.0689,
+      "step": 6666
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.834506129550661e-06,
+      "loss": 0.075,
+      "step": 6667
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831752133966682e-06,
+      "loss": 0.0691,
+      "step": 6668
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828998893316535e-06,
+      "loss": 0.0768,
+      "step": 6669
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.826246407937176e-06,
+      "loss": 0.0727,
+      "step": 6670
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823494678165498e-06,
+      "loss": 0.0788,
+      "step": 6671
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820743704338274e-06,
+      "loss": 0.0743,
+      "step": 6672
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817993486792202e-06,
+      "loss": 0.0701,
+      "step": 6673
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.81524402586388e-06,
+      "loss": 0.0674,
+      "step": 6674
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.812495321889815e-06,
+      "loss": 0.0677,
+      "step": 6675
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8097473752064284e-06,
+      "loss": 0.07,
+      "step": 6676
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8070001861500284e-06,
+      "loss": 0.065,
+      "step": 6677
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8042537550568603e-06,
+      "loss": 0.0706,
+      "step": 6678
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.80150808226305e-06,
+      "loss": 0.0641,
+      "step": 6679
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7987631681046435e-06,
+      "loss": 0.0679,
+      "step": 6680
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7960190129175934e-06,
+      "loss": 0.0719,
+      "step": 6681
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.793275617037756e-06,
+      "loss": 0.0722,
+      "step": 6682
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.790532980800898e-06,
+      "loss": 0.076,
+      "step": 6683
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.787791104542691e-06,
+      "loss": 0.0718,
+      "step": 6684
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.785049988598717e-06,
+      "loss": 0.0615,
+      "step": 6685
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7823096333044553e-06,
+      "loss": 0.0678,
+      "step": 6686
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779570038995303e-06,
+      "loss": 0.0655,
+      "step": 6687
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7768312060065593e-06,
+      "loss": 0.0688,
+      "step": 6688
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.774093134673431e-06,
+      "loss": 0.0687,
+      "step": 6689
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.771355825331032e-06,
+      "loss": 0.0618,
+      "step": 6690
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.768619278314385e-06,
+      "loss": 0.0733,
+      "step": 6691
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7658834939584076e-06,
+      "loss": 0.0658,
+      "step": 6692
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.763148472597946e-06,
+      "loss": 0.071,
+      "step": 6693
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7604142145677303e-06,
+      "loss": 0.0669,
+      "step": 6694
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.757680720202411e-06,
+      "loss": 0.0749,
+      "step": 6695
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7549479898365413e-06,
+      "loss": 0.0744,
+      "step": 6696
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7522160238045814e-06,
+      "loss": 0.0699,
+      "step": 6697
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7494848224408998e-06,
+      "loss": 0.0805,
+      "step": 6698
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7467543860797594e-06,
+      "loss": 0.0705,
+      "step": 6699
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7440247150553532e-06,
+      "loss": 0.0699,
+      "step": 6700
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741295809701753e-06,
+      "loss": 0.0708,
+      "step": 6701
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7385676703529628e-06,
+      "loss": 0.0759,
+      "step": 6702
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7358402973428707e-06,
+      "loss": 0.0674,
+      "step": 6703
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7331136910052847e-06,
+      "loss": 0.0688,
+      "step": 6704
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7303878516739155e-06,
+      "loss": 0.0725,
+      "step": 6705
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.727662779682377e-06,
+      "loss": 0.0679,
+      "step": 6706
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7249384753641984e-06,
+      "loss": 0.0684,
+      "step": 6707
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.722214939052796e-06,
+      "loss": 0.0647,
+      "step": 6708
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.719492171081518e-06,
+      "loss": 0.074,
+      "step": 6709
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716770171783595e-06,
+      "loss": 0.0684,
+      "step": 6710
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7140489414921756e-06,
+      "loss": 0.0707,
+      "step": 6711
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7113284805403138e-06,
+      "loss": 0.068,
+      "step": 6712
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.708608789260967e-06,
+      "loss": 0.0745,
+      "step": 6713
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7058898679869994e-06,
+      "loss": 0.0723,
+      "step": 6714
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7031717170511826e-06,
+      "loss": 0.0706,
+      "step": 6715
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7004543367861867e-06,
+      "loss": 0.0707,
+      "step": 6716
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6977377275245962e-06,
+      "loss": 0.0704,
+      "step": 6717
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6950218895988966e-06,
+      "loss": 0.0759,
+      "step": 6718
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6923068233414806e-06,
+      "loss": 0.0751,
+      "step": 6719
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6895925290846467e-06,
+      "loss": 0.0714,
+      "step": 6720
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6868790071605974e-06,
+      "loss": 0.0674,
+      "step": 6721
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.684166257901446e-06,
+      "loss": 0.0774,
+      "step": 6722
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.681454281639195e-06,
+      "loss": 0.0771,
+      "step": 6723
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6787430787057786e-06,
+      "loss": 0.069,
+      "step": 6724
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6760326494330123e-06,
+      "loss": 0.07,
+      "step": 6725
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6733229941526284e-06,
+      "loss": 0.0649,
+      "step": 6726
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6706141131962625e-06,
+      "loss": 0.0682,
+      "step": 6727
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6679060068954566e-06,
+      "loss": 0.0781,
+      "step": 6728
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6651986755816593e-06,
+      "loss": 0.0698,
+      "step": 6729
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6624921195862116e-06,
+      "loss": 0.0704,
+      "step": 6730
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659786339240383e-06,
+      "loss": 0.0654,
+      "step": 6731
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.657081334875322e-06,
+      "loss": 0.0702,
+      "step": 6732
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.654377106822109e-06,
+      "loss": 0.0728,
+      "step": 6733
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6516736554117037e-06,
+      "loss": 0.0796,
+      "step": 6734
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.648970980974986e-06,
+      "loss": 0.0734,
+      "step": 6735
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6462690838427384e-06,
+      "loss": 0.0782,
+      "step": 6736
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6435679643456445e-06,
+      "loss": 0.0706,
+      "step": 6737
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6408676228143013e-06,
+      "loss": 0.0702,
+      "step": 6738
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6381680595791935e-06,
+      "loss": 0.0722,
+      "step": 6739
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6354692749707344e-06,
+      "loss": 0.0736,
+      "step": 6740
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6327712693192184e-06,
+      "loss": 0.0727,
+      "step": 6741
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6300740429548597e-06,
+      "loss": 0.0766,
+      "step": 6742
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6273775962077728e-06,
+      "loss": 0.0788,
+      "step": 6743
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.624681929407977e-06,
+      "loss": 0.0731,
+      "step": 6744
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6219870428853942e-06,
+      "loss": 0.0713,
+      "step": 6745
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.619292936969857e-06,
+      "loss": 0.07,
+      "step": 6746
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.616599611991092e-06,
+      "loss": 0.0686,
+      "step": 6747
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.613907068278737e-06,
+      "loss": 0.0704,
+      "step": 6748
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6112153061623344e-06,
+      "loss": 0.0709,
+      "step": 6749
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.608524325971331e-06,
+      "loss": 0.0685,
+      "step": 6750
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.605834128035076e-06,
+      "loss": 0.0739,
+      "step": 6751
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6031447126828223e-06,
+      "loss": 0.0703,
+      "step": 6752
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6004560802437337e-06,
+      "loss": 0.0742,
+      "step": 6753
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.597768231046862e-06,
+      "loss": 0.0666,
+      "step": 6754
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5950811654211857e-06,
+      "loss": 0.065,
+      "step": 6755
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.592394883695569e-06,
+      "loss": 0.0647,
+      "step": 6756
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5897093861987864e-06,
+      "loss": 0.0739,
+      "step": 6757
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.587024673259519e-06,
+      "loss": 0.0669,
+      "step": 6758
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5843407452063495e-06,
+      "loss": 0.073,
+      "step": 6759
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5816576023677687e-06,
+      "loss": 0.0667,
+      "step": 6760
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5789752450721548e-06,
+      "loss": 0.0737,
+      "step": 6761
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.576293673647817e-06,
+      "loss": 0.0757,
+      "step": 6762
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5736128884229415e-06,
+      "loss": 0.0766,
+      "step": 6763
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5709328897256423e-06,
+      "loss": 0.0803,
+      "step": 6764
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.568253677883916e-06,
+      "loss": 0.0739,
+      "step": 6765
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.565575253225676e-06,
+      "loss": 0.0686,
+      "step": 6766
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5628976160787343e-06,
+      "loss": 0.0757,
+      "step": 6767
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.560220766770809e-06,
+      "loss": 0.0741,
+      "step": 6768
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.557544705629523e-06,
+      "loss": 0.0737,
+      "step": 6769
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.554869432982394e-06,
+      "loss": 0.0654,
+      "step": 6770
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.552194949156852e-06,
+      "loss": 0.0737,
+      "step": 6771
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5495212544802294e-06,
+      "loss": 0.0714,
+      "step": 6772
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.54684834927976e-06,
+      "loss": 0.0676,
+      "step": 6773
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5441762338825803e-06,
+      "loss": 0.0702,
+      "step": 6774
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5415049086157327e-06,
+      "loss": 0.0751,
+      "step": 6775
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5388343738061615e-06,
+      "loss": 0.0687,
+      "step": 6776
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.536164629780717e-06,
+      "loss": 0.07,
+      "step": 6777
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.533495676866141e-06,
+      "loss": 0.071,
+      "step": 6778
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5308275153891004e-06,
+      "loss": 0.0666,
+      "step": 6779
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5281601456761404e-06,
+      "loss": 0.0767,
+      "step": 6780
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5254935680537285e-06,
+      "loss": 0.073,
+      "step": 6781
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5228277828482283e-06,
+      "loss": 0.0689,
+      "step": 6782
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.520162790385896e-06,
+      "loss": 0.0666,
+      "step": 6783
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517498590992916e-06,
+      "loss": 0.0653,
+      "step": 6784
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5148351849953467e-06,
+      "loss": 0.0713,
+      "step": 6785
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5121725727191746e-06,
+      "loss": 0.0767,
+      "step": 6786
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509510754490267e-06,
+      "loss": 0.0667,
+      "step": 6787
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5068497306344173e-06,
+      "loss": 0.0728,
+      "step": 6788
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5041895014772973e-06,
+      "loss": 0.0711,
+      "step": 6789
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5015300673444995e-06,
+      "loss": 0.0747,
+      "step": 6790
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4988714285615144e-06,
+      "loss": 0.0781,
+      "step": 6791
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.496213585453724e-06,
+      "loss": 0.0813,
+      "step": 6792
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493556538346435e-06,
+      "loss": 0.0671,
+      "step": 6793
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4909002875648356e-06,
+      "loss": 0.0754,
+      "step": 6794
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.488244833434028e-06,
+      "loss": 0.0659,
+      "step": 6795
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4855901762790144e-06,
+      "loss": 0.0789,
+      "step": 6796
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4829363164246987e-06,
+      "loss": 0.068,
+      "step": 6797
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4802832541958886e-06,
+      "loss": 0.0668,
+      "step": 6798
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.477630989917291e-06,
+      "loss": 0.0714,
+      "step": 6799
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4749795239135197e-06,
+      "loss": 0.0695,
+      "step": 6800
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4723288565090908e-06,
+      "loss": 0.0713,
+      "step": 6801
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4696789880284133e-06,
+      "loss": 0.0668,
+      "step": 6802
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4670299187958102e-06,
+      "loss": 0.0751,
+      "step": 6803
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4643816491355e-06,
+      "loss": 0.0718,
+      "step": 6804
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4617341793716063e-06,
+      "loss": 0.0727,
+      "step": 6805
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.459087509828154e-06,
+      "loss": 0.0681,
+      "step": 6806
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4564416408290703e-06,
+      "loss": 0.08,
+      "step": 6807
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.453796572698186e-06,
+      "loss": 0.0675,
+      "step": 6808
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.451152305759222e-06,
+      "loss": 0.0708,
+      "step": 6809
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.448508840335825e-06,
+      "loss": 0.0676,
+      "step": 6810
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.445866176751519e-06,
+      "loss": 0.0735,
+      "step": 6811
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.443224315329744e-06,
+      "loss": 0.075,
+      "step": 6812
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4405832563938423e-06,
+      "loss": 0.0697,
+      "step": 6813
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4379430002670422e-06,
+      "loss": 0.0716,
+      "step": 6814
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4353035472725005e-06,
+      "loss": 0.0655,
+      "step": 6815
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4326648977332466e-06,
+      "loss": 0.071,
+      "step": 6816
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4300270519722403e-06,
+      "loss": 0.0721,
+      "step": 6817
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4273900103123125e-06,
+      "loss": 0.0643,
+      "step": 6818
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.424753773076227e-06,
+      "loss": 0.0694,
+      "step": 6819
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4221183405866223e-06,
+      "loss": 0.0736,
+      "step": 6820
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419483713166054e-06,
+      "loss": 0.0695,
+      "step": 6821
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4168498911369753e-06,
+      "loss": 0.0712,
+      "step": 6822
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.414216874821741e-06,
+      "loss": 0.07,
+      "step": 6823
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4115846645426075e-06,
+      "loss": 0.0747,
+      "step": 6824
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.408953260621728e-06,
+      "loss": 0.0689,
+      "step": 6825
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.406322663381163e-06,
+      "loss": 0.0676,
+      "step": 6826
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.403692873142872e-06,
+      "loss": 0.0709,
+      "step": 6827
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4010638902287163e-06,
+      "loss": 0.0677,
+      "step": 6828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3984357149604575e-06,
+      "loss": 0.0662,
+      "step": 6829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395808347659759e-06,
+      "loss": 0.0658,
+      "step": 6830
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3931817886481856e-06,
+      "loss": 0.0625,
+      "step": 6831
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.390556038247206e-06,
+      "loss": 0.0747,
+      "step": 6832
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.387931096778179e-06,
+      "loss": 0.0726,
+      "step": 6833
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.385306964562376e-06,
+      "loss": 0.0695,
+      "step": 6834
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382683641920965e-06,
+      "loss": 0.0722,
+      "step": 6835
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3800611291750165e-06,
+      "loss": 0.075,
+      "step": 6836
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3774394266455023e-06,
+      "loss": 0.0743,
+      "step": 6837
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3748185346532847e-06,
+      "loss": 0.0697,
+      "step": 6838
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.372198453519149e-06,
+      "loss": 0.068,
+      "step": 6839
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3695791835637524e-06,
+      "loss": 0.0653,
+      "step": 6840
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.366960725107684e-06,
+      "loss": 0.071,
+      "step": 6841
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.364343078471407e-06,
+      "loss": 0.0717,
+      "step": 6842
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3617262439752982e-06,
+      "loss": 0.0739,
+      "step": 6843
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3591102219396343e-06,
+      "loss": 0.0761,
+      "step": 6844
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3564950126845895e-06,
+      "loss": 0.0714,
+      "step": 6845
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.353880616530245e-06,
+      "loss": 0.0739,
+      "step": 6846
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.351267033796567e-06,
+      "loss": 0.0716,
+      "step": 6847
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348654264803446e-06,
+      "loss": 0.0751,
+      "step": 6848
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.34604230987065e-06,
+      "loss": 0.0697,
+      "step": 6849
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3434311693178588e-06,
+      "loss": 0.0772,
+      "step": 6850
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3408208434646536e-06,
+      "loss": 0.0735,
+      "step": 6851
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.33821133263051e-06,
+      "loss": 0.0685,
+      "step": 6852
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3356026371348083e-06,
+      "loss": 0.0669,
+      "step": 6853
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.332994757296828e-06,
+      "loss": 0.0698,
+      "step": 6854
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3303876934357502e-06,
+      "loss": 0.0696,
+      "step": 6855
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3277814458706504e-06,
+      "loss": 0.0721,
+      "step": 6856
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.325176014920508e-06,
+      "loss": 0.0742,
+      "step": 6857
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3225714009042045e-06,
+      "loss": 0.0672,
+      "step": 6858
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3199676041405183e-06,
+      "loss": 0.0701,
+      "step": 6859
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3173646249481308e-06,
+      "loss": 0.0744,
+      "step": 6860
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3147624636456232e-06,
+      "loss": 0.0727,
+      "step": 6861
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312161120551465e-06,
+      "loss": 0.0776,
+      "step": 6862
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309560595984048e-06,
+      "loss": 0.0704,
+      "step": 6863
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3069608902616447e-06,
+      "loss": 0.0779,
+      "step": 6864
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3043620037024326e-06,
+      "loss": 0.0749,
+      "step": 6865
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3017639366244936e-06,
+      "loss": 0.0663,
+      "step": 6866
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299166689345804e-06,
+      "loss": 0.0756,
+      "step": 6867
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2965702621842464e-06,
+      "loss": 0.0684,
+      "step": 6868
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293974655457588e-06,
+      "loss": 0.067,
+      "step": 6869
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2913798694835187e-06,
+      "loss": 0.0764,
+      "step": 6870
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2887859045796033e-06,
+      "loss": 0.0688,
+      "step": 6871
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.286192761063329e-06,
+      "loss": 0.064,
+      "step": 6872
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.283600439252063e-06,
+      "loss": 0.0737,
+      "step": 6873
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.281008939463084e-06,
+      "loss": 0.0709,
+      "step": 6874
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.278418262013565e-06,
+      "loss": 0.0687,
+      "step": 6875
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2758284072205816e-06,
+      "loss": 0.078,
+      "step": 6876
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2732393754011083e-06,
+      "loss": 0.0638,
+      "step": 6877
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.27065116687201e-06,
+      "loss": 0.0658,
+      "step": 6878
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2680637819500693e-06,
+      "loss": 0.0691,
+      "step": 6879
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2654772209519493e-06,
+      "loss": 0.076,
+      "step": 6880
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2628914841942214e-06,
+      "loss": 0.0662,
+      "step": 6881
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260306571993356e-06,
+      "loss": 0.0716,
+      "step": 6882
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.257722484665721e-06,
+      "loss": 0.0713,
+      "step": 6883
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2551392225275835e-06,
+      "loss": 0.0693,
+      "step": 6884
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.252556785895109e-06,
+      "loss": 0.0723,
+      "step": 6885
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.249975175084368e-06,
+      "loss": 0.0634,
+      "step": 6886
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.247394390411317e-06,
+      "loss": 0.0716,
+      "step": 6887
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2448144321918217e-06,
+      "loss": 0.0707,
+      "step": 6888
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242235300741644e-06,
+      "loss": 0.0717,
+      "step": 6889
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239656996376447e-06,
+      "loss": 0.0695,
+      "step": 6890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237079519411788e-06,
+      "loss": 0.0707,
+      "step": 6891
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2345028701631277e-06,
+      "loss": 0.0796,
+      "step": 6892
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2319270489458154e-06,
+      "loss": 0.0653,
+      "step": 6893
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.229352056075118e-06,
+      "loss": 0.0755,
+      "step": 6894
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.226777891866182e-06,
+      "loss": 0.071,
+      "step": 6895
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.224204556634061e-06,
+      "loss": 0.0696,
+      "step": 6896
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2216320506937083e-06,
+      "loss": 0.065,
+      "step": 6897
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2190603743599723e-06,
+      "loss": 0.0725,
+      "step": 6898
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.216489527947606e-06,
+      "loss": 0.0616,
+      "step": 6899
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.213919511771245e-06,
+      "loss": 0.0784,
+      "step": 6900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.211350326145448e-06,
+      "loss": 0.0691,
+      "step": 6901
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2087819713846445e-06,
+      "loss": 0.0735,
+      "step": 6902
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.206214447803191e-06,
+      "loss": 0.0621,
+      "step": 6903
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2036477557153166e-06,
+      "loss": 0.0652,
+      "step": 6904
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2010818954351642e-06,
+      "loss": 0.0689,
+      "step": 6905
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1985168672767674e-06,
+      "loss": 0.0767,
+      "step": 6906
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1959526715540632e-06,
+      "loss": 0.0688,
+      "step": 6907
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1933893085808875e-06,
+      "loss": 0.0664,
+      "step": 6908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1908267786709614e-06,
+      "loss": 0.0721,
+      "step": 6909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1882650821379247e-06,
+      "loss": 0.0663,
+      "step": 6910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185704219295297e-06,
+      "loss": 0.0722,
+      "step": 6911
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1831441904565042e-06,
+      "loss": 0.0719,
+      "step": 6912
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.180584995934871e-06,
+      "loss": 0.0711,
+      "step": 6913
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1780266360436175e-06,
+      "loss": 0.0633,
+      "step": 6914
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1754691110958615e-06,
+      "loss": 0.0755,
+      "step": 6915
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1729124214046236e-06,
+      "loss": 0.0773,
+      "step": 6916
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1703565672828098e-06,
+      "loss": 0.0655,
+      "step": 6917
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1678015490432368e-06,
+      "loss": 0.0734,
+      "step": 6918
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1652473669986127e-06,
+      "loss": 0.0727,
+      "step": 6919
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1626940214615466e-06,
+      "loss": 0.0701,
+      "step": 6920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1601415127445414e-06,
+      "loss": 0.069,
+      "step": 6921
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1575898411599994e-06,
+      "loss": 0.071,
+      "step": 6922
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1550390070202255e-06,
+      "loss": 0.0672,
+      "step": 6923
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.152489010637405e-06,
+      "loss": 0.0662,
+      "step": 6924
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.149939852323648e-06,
+      "loss": 0.0696,
+      "step": 6925
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1473915323909323e-06,
+      "loss": 0.0763,
+      "step": 6926
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1448440511511613e-06,
+      "loss": 0.0626,
+      "step": 6927
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.142297408916113e-06,
+      "loss": 0.0739,
+      "step": 6928
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1397516059974733e-06,
+      "loss": 0.0726,
+      "step": 6929
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.137206642706828e-06,
+      "loss": 0.0817,
+      "step": 6930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1346625193556458e-06,
+      "loss": 0.0711,
+      "step": 6931
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1321192362553164e-06,
+      "loss": 0.0748,
+      "step": 6932
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1295767937171007e-06,
+      "loss": 0.0666,
+      "step": 6933
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1270351920521803e-06,
+      "loss": 0.0705,
+      "step": 6934
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.124494431571613e-06,
+      "loss": 0.0754,
+      "step": 6935
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.121954512586368e-06,
+      "loss": 0.0735,
+      "step": 6936
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.119415435407306e-06,
+      "loss": 0.0722,
+      "step": 6937
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1168772003451855e-06,
+      "loss": 0.072,
+      "step": 6938
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1143398077106645e-06,
+      "loss": 0.0725,
+      "step": 6939
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.11180325781429e-06,
+      "loss": 0.0737,
+      "step": 6940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.109267550966514e-06,
+      "loss": 0.0678,
+      "step": 6941
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1067326874776816e-06,
+      "loss": 0.0712,
+      "step": 6942
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1041986676580372e-06,
+      "loss": 0.0707,
+      "step": 6943
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.101665491817719e-06,
+      "loss": 0.0706,
+      "step": 6944
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099133160266763e-06,
+      "loss": 0.0733,
+      "step": 6945
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0966016733151028e-06,
+      "loss": 0.0682,
+      "step": 6946
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.094071031272572e-06,
+      "loss": 0.0704,
+      "step": 6947
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091541234448885e-06,
+      "loss": 0.0752,
+      "step": 6948
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0890122831536782e-06,
+      "loss": 0.0673,
+      "step": 6949
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0864841776964615e-06,
+      "loss": 0.0635,
+      "step": 6950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0839569183866534e-06,
+      "loss": 0.0726,
+      "step": 6951
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.081430505533566e-06,
+      "loss": 0.0705,
+      "step": 6952
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0789049394464076e-06,
+      "loss": 0.0761,
+      "step": 6953
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.076380220434286e-06,
+      "loss": 0.0807,
+      "step": 6954
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0738563488061936e-06,
+      "loss": 0.0716,
+      "step": 6955
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0713333248710407e-06,
+      "loss": 0.0743,
+      "step": 6956
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.068811148937607e-06,
+      "loss": 0.073,
+      "step": 6957
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0662898213145965e-06,
+      "loss": 0.0682,
+      "step": 6958
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.063769342310585e-06,
+      "loss": 0.0658,
+      "step": 6959
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.061249712234058e-06,
+      "loss": 0.0734,
+      "step": 6960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0587309313933944e-06,
+      "loss": 0.0709,
+      "step": 6961
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.056213000096868e-06,
+      "loss": 0.0735,
+      "step": 6962
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0536959186526515e-06,
+      "loss": 0.0697,
+      "step": 6963
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.051179687368804e-06,
+      "loss": 0.0698,
+      "step": 6964
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0486643065532985e-06,
+      "loss": 0.0732,
+      "step": 6965
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0461497765139848e-06,
+      "loss": 0.0623,
+      "step": 6966
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.04363609755862e-06,
+      "loss": 0.0631,
+      "step": 6967
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0411232699948546e-06,
+      "loss": 0.0685,
+      "step": 6968
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0386112941302327e-06,
+      "loss": 0.0726,
+      "step": 6969
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0361001702721983e-06,
+      "loss": 0.0729,
+      "step": 6970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.033589898728089e-06,
+      "loss": 0.0673,
+      "step": 6971
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031080479805134e-06,
+      "loss": 0.0703,
+      "step": 6972
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.028571913810463e-06,
+      "loss": 0.0727,
+      "step": 6973
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.026064201051101e-06,
+      "loss": 0.0762,
+      "step": 6974
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.023557341833969e-06,
+      "loss": 0.0705,
+      "step": 6975
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02105133646588e-06,
+      "loss": 0.0668,
+      "step": 6976
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0185461852535467e-06,
+      "loss": 0.0785,
+      "step": 6977
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.016041888503578e-06,
+      "loss": 0.0719,
+      "step": 6978
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0135384465224647e-06,
+      "loss": 0.0714,
+      "step": 6979
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011035859616619e-06,
+      "loss": 0.0756,
+      "step": 6980
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0085341280923217e-06,
+      "loss": 0.0685,
+      "step": 6981
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0060332522557646e-06,
+      "loss": 0.0714,
+      "step": 6982
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0035332324130306e-06,
+      "loss": 0.0636,
+      "step": 6983
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0010340688700977e-06,
+      "loss": 0.0706,
+      "step": 6984
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9985357619328435e-06,
+      "loss": 0.0653,
+      "step": 6985
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9960383119070257e-06,
+      "loss": 0.073,
+      "step": 6986
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.993541719098322e-06,
+      "loss": 0.0722,
+      "step": 6987
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.991045983812276e-06,
+      "loss": 0.0667,
+      "step": 6988
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.988551106354357e-06,
+      "loss": 0.0696,
+      "step": 6989
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9860570870299045e-06,
+      "loss": 0.0713,
+      "step": 6990
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.983563926144163e-06,
+      "loss": 0.0704,
+      "step": 6991
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.981071624002272e-06,
+      "loss": 0.0675,
+      "step": 6992
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9785801809092673e-06,
+      "loss": 0.0741,
+      "step": 6993
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9760895971700775e-06,
+      "loss": 0.0717,
+      "step": 6994
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.973599873089522e-06,
+      "loss": 0.0701,
+      "step": 6995
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9711110089723218e-06,
+      "loss": 0.0724,
+      "step": 6996
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.968623005123088e-06,
+      "loss": 0.0713,
+      "step": 6997
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9661358618463297e-06,
+      "loss": 0.073,
+      "step": 6998
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9636495794464493e-06,
+      "loss": 0.0693,
+      "step": 6999
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.961164158227743e-06,
+      "loss": 0.0732,
+      "step": 7000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9586795984944017e-06,
+      "loss": 0.0738,
+      "step": 7001
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.956195900550516e-06,
+      "loss": 0.0751,
+      "step": 7002
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.95371306470006e-06,
+      "loss": 0.0734,
+      "step": 7003
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.951231091246911e-06,
+      "loss": 0.0678,
+      "step": 7004
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9487499804948395e-06,
+      "loss": 0.0811,
+      "step": 7005
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9462697327475087e-06,
+      "loss": 0.0704,
+      "step": 7006
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9437903483084807e-06,
+      "loss": 0.0707,
+      "step": 7007
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.941311827481198e-06,
+      "loss": 0.0738,
+      "step": 7008
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9388341705690204e-06,
+      "loss": 0.0644,
+      "step": 7009
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9363573778751776e-06,
+      "loss": 0.0736,
+      "step": 7010
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9338814497028158e-06,
+      "loss": 0.0768,
+      "step": 7011
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9314063863549556e-06,
+      "loss": 0.0788,
+      "step": 7012
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.0712,
+      "step": 7013
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.926458855344342e-06,
+      "loss": 0.069,
+      "step": 7014
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.923986388287118e-06,
+      "loss": 0.0715,
+      "step": 7015
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9215147872654604e-06,
+      "loss": 0.0758,
+      "step": 7016
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.919044052581863e-06,
+      "loss": 0.0651,
+      "step": 7017
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.916574184538731e-06,
+      "loss": 0.0719,
+      "step": 7018
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9141051834383383e-06,
+      "loss": 0.0712,
+      "step": 7019
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9116370495828817e-06,
+      "loss": 0.0699,
+      "step": 7020
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9091697832744258e-06,
+      "loss": 0.0644,
+      "step": 7021
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906703384814944e-06,
+      "loss": 0.0701,
+      "step": 7022
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9042378545063e-06,
+      "loss": 0.0704,
+      "step": 7023
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9017731926502503e-06,
+      "loss": 0.0704,
+      "step": 7024
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8993093995484477e-06,
+      "loss": 0.0733,
+      "step": 7025
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896846475502433e-06,
+      "loss": 0.0675,
+      "step": 7026
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.894384420813645e-06,
+      "loss": 0.0691,
+      "step": 7027
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.891923235783416e-06,
+      "loss": 0.072,
+      "step": 7028
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889462920712972e-06,
+      "loss": 0.0691,
+      "step": 7029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8870034759034306e-06,
+      "loss": 0.0688,
+      "step": 7030
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.884544901655805e-06,
+      "loss": 0.07,
+      "step": 7031
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882087198271001e-06,
+      "loss": 0.0706,
+      "step": 7032
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8796303660498193e-06,
+      "loss": 0.0675,
+      "step": 7033
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877174405292947e-06,
+      "loss": 0.0648,
+      "step": 7034
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8747193163009745e-06,
+      "loss": 0.0706,
+      "step": 7035
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8722650993743785e-06,
+      "loss": 0.0701,
+      "step": 7036
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.869811754813533e-06,
+      "loss": 0.0841,
+      "step": 7037
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8673592829187057e-06,
+      "loss": 0.0688,
+      "step": 7038
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8649076839900467e-06,
+      "loss": 0.0738,
+      "step": 7039
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.862456958327622e-06,
+      "loss": 0.0707,
+      "step": 7040
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8600071062313606e-06,
+      "loss": 0.0697,
+      "step": 7041
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8575581280011157e-06,
+      "loss": 0.0697,
+      "step": 7042
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8551100239366093e-06,
+      "loss": 0.071,
+      "step": 7043
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8526627943374674e-06,
+      "loss": 0.0668,
+      "step": 7044
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8502164395032086e-06,
+      "loss": 0.0697,
+      "step": 7045
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8477709597332416e-06,
+      "loss": 0.0688,
+      "step": 7046
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8453263553268726e-06,
+      "loss": 0.0747,
+      "step": 7047
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.84288262658329e-06,
+      "loss": 0.0742,
+      "step": 7048
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8404397738015944e-06,
+      "loss": 0.0738,
+      "step": 7049
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.837997797280756e-06,
+      "loss": 0.0744,
+      "step": 7050
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8355566973196545e-06,
+      "loss": 0.0736,
+      "step": 7051
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.833116474217056e-06,
+      "loss": 0.069,
+      "step": 7052
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8306771282716205e-06,
+      "loss": 0.0694,
+      "step": 7053
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.828238659781899e-06,
+      "loss": 0.0653,
+      "step": 7054
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8258010690463387e-06,
+      "loss": 0.0737,
+      "step": 7055
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8233643563632783e-06,
+      "loss": 0.0645,
+      "step": 7056
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8209285220309424e-06,
+      "loss": 0.0699,
+      "step": 7057
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8184935663474567e-06,
+      "loss": 0.0685,
+      "step": 7058
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.816059489610836e-06,
+      "loss": 0.0659,
+      "step": 7059
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8136262921189884e-06,
+      "loss": 0.0668,
+      "step": 7060
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8111939741697135e-06,
+      "loss": 0.0649,
+      "step": 7061
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8087625360607053e-06,
+      "loss": 0.0698,
+      "step": 7062
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8063319780895406e-06,
+      "loss": 0.0772,
+      "step": 7063
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.803902300553707e-06,
+      "loss": 0.0698,
+      "step": 7064
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.801473503750566e-06,
+      "loss": 0.069,
+      "step": 7065
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7990455879773804e-06,
+      "loss": 0.0679,
+      "step": 7066
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.796618553531304e-06,
+      "loss": 0.071,
+      "step": 7067
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7941924007093823e-06,
+      "loss": 0.0721,
+      "step": 7068
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.791767129808557e-06,
+      "loss": 0.0694,
+      "step": 7069
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.789342741125648e-06,
+      "loss": 0.0727,
+      "step": 7070
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7869192349573883e-06,
+      "loss": 0.0727,
+      "step": 7071
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.78449661160038e-06,
+      "loss": 0.0715,
+      "step": 7072
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7820748713511413e-06,
+      "loss": 0.0674,
+      "step": 7073
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.779654014506061e-06,
+      "loss": 0.0688,
+      "step": 7074
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7772340413614306e-06,
+      "loss": 0.0678,
+      "step": 7075
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7748149522134317e-06,
+      "loss": 0.0699,
+      "step": 7076
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.772396747358137e-06,
+      "loss": 0.0733,
+      "step": 7077
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.769979427091515e-06,
+      "loss": 0.0695,
+      "step": 7078
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.767562991709414e-06,
+      "loss": 0.0652,
+      "step": 7079
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7651474415075917e-06,
+      "loss": 0.0709,
+      "step": 7080
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7627327767816824e-06,
+      "loss": 0.0704,
+      "step": 7081
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.760318997827218e-06,
+      "loss": 0.0771,
+      "step": 7082
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7579061049396225e-06,
+      "loss": 0.0709,
+      "step": 7083
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.755494098414212e-06,
+      "loss": 0.0691,
+      "step": 7084
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.753082978546191e-06,
+      "loss": 0.0702,
+      "step": 7085
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7506727456306583e-06,
+      "loss": 0.0703,
+      "step": 7086
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.748263399962602e-06,
+      "loss": 0.0778,
+      "step": 7087
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7458549418369064e-06,
+      "loss": 0.0696,
+      "step": 7088
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.743447371548337e-06,
+      "loss": 0.0744,
+      "step": 7089
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7410406893915607e-06,
+      "loss": 0.0646,
+      "step": 7090
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.738634895661132e-06,
+      "loss": 0.0689,
+      "step": 7091
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7362299906514966e-06,
+      "loss": 0.0741,
+      "step": 7092
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.733825974656995e-06,
+      "loss": 0.0693,
+      "step": 7093
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.731422847971845e-06,
+      "loss": 0.0633,
+      "step": 7094
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.729020610890181e-06,
+      "loss": 0.0696,
+      "step": 7095
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7266192637059973e-06,
+      "loss": 0.0744,
+      "step": 7096
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7242188067132114e-06,
+      "loss": 0.0738,
+      "step": 7097
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7218192402056052e-06,
+      "loss": 0.0684,
+      "step": 7098
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.719420564476866e-06,
+      "loss": 0.0687,
+      "step": 7099
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7170227798205705e-06,
+      "loss": 0.0657,
+      "step": 7100
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7146258865301765e-06,
+      "loss": 0.0728,
+      "step": 7101
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7122298848990527e-06,
+      "loss": 0.0766,
+      "step": 7102
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7098347752204335e-06,
+      "loss": 0.0735,
+      "step": 7103
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7074405577874698e-06,
+      "loss": 0.0728,
+      "step": 7104
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7050472328931807e-06,
+      "loss": 0.0703,
+      "step": 7105
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.70265480083049e-06,
+      "loss": 0.0708,
+      "step": 7106
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.700263261892209e-06,
+      "loss": 0.0674,
+      "step": 7107
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6978726163710354e-06,
+      "loss": 0.072,
+      "step": 7108
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6954828645595656e-06,
+      "loss": 0.0708,
+      "step": 7109
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6930940067502787e-06,
+      "loss": 0.0677,
+      "step": 7110
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6907060432355526e-06,
+      "loss": 0.0711,
+      "step": 7111
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6883189743076442e-06,
+      "loss": 0.0717,
+      "step": 7112
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6859328002587113e-06,
+      "loss": 0.0716,
+      "step": 7113
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.683547521380797e-06,
+      "loss": 0.0746,
+      "step": 7114
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.681163137965839e-06,
+      "loss": 0.0683,
+      "step": 7115
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6787796503056596e-06,
+      "loss": 0.0652,
+      "step": 7116
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6763970586919798e-06,
+      "loss": 0.0734,
+      "step": 7117
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6740153634163955e-06,
+      "loss": 0.073,
+      "step": 7118
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6716345647704154e-06,
+      "loss": 0.064,
+      "step": 7119
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6692546630454187e-06,
+      "loss": 0.0652,
+      "step": 7120
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6668756585326828e-06,
+      "loss": 0.0667,
+      "step": 7121
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.664497551523377e-06,
+      "loss": 0.0713,
+      "step": 7122
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6621203423085574e-06,
+      "loss": 0.0721,
+      "step": 7123
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6597440311791757e-06,
+      "loss": 0.0716,
+      "step": 7124
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6573686184260585e-06,
+      "loss": 0.0699,
+      "step": 7125
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6549941043399463e-06,
+      "loss": 0.0666,
+      "step": 7126
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.652620489211444e-06,
+      "loss": 0.078,
+      "step": 7127
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6502477733310726e-06,
+      "loss": 0.0661,
+      "step": 7128
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6478759569892186e-06,
+      "loss": 0.0703,
+      "step": 7129
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.645505040476174e-06,
+      "loss": 0.0712,
+      "step": 7130
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.643135024082116e-06,
+      "loss": 0.0705,
+      "step": 7131
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6407659080971104e-06,
+      "loss": 0.0755,
+      "step": 7132
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6383976928111176e-06,
+      "loss": 0.0687,
+      "step": 7133
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6360303785139763e-06,
+      "loss": 0.0675,
+      "step": 7134
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6336639654954333e-06,
+      "loss": 0.0686,
+      "step": 7135
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.631298454045106e-06,
+      "loss": 0.0627,
+      "step": 7136
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6289338444525137e-06,
+      "loss": 0.0697,
+      "step": 7137
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.626570137007061e-06,
+      "loss": 0.0728,
+      "step": 7138
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.624207331998043e-06,
+      "loss": 0.07,
+      "step": 7139
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6218454297146455e-06,
+      "loss": 0.0709,
+      "step": 7140
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.619484430445943e-06,
+      "loss": 0.0719,
+      "step": 7141
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6171243344808948e-06,
+      "loss": 0.0679,
+      "step": 7142
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6147651421083574e-06,
+      "loss": 0.0678,
+      "step": 7143
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612406853617071e-06,
+      "loss": 0.0647,
+      "step": 7144
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.610049469295668e-06,
+      "loss": 0.0681,
+      "step": 7145
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.60769298943267e-06,
+      "loss": 0.0694,
+      "step": 7146
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6053374143164865e-06,
+      "loss": 0.0681,
+      "step": 7147
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6029827442354218e-06,
+      "loss": 0.0784,
+      "step": 7148
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.600628979477654e-06,
+      "loss": 0.0699,
+      "step": 7149
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5982761203312733e-06,
+      "loss": 0.0682,
+      "step": 7150
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5959241670842384e-06,
+      "loss": 0.0725,
+      "step": 7151
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5935731200244085e-06,
+      "loss": 0.0714,
+      "step": 7152
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5912229794395282e-06,
+      "loss": 0.0692,
+      "step": 7153
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.588873745617234e-06,
+      "loss": 0.0771,
+      "step": 7154
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5865254188450507e-06,
+      "loss": 0.0702,
+      "step": 7155
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.584177999410381e-06,
+      "loss": 0.0782,
+      "step": 7156
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5818314876005413e-06,
+      "loss": 0.0672,
+      "step": 7157
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.579485883702707e-06,
+      "loss": 0.0614,
+      "step": 7158
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57714118800397e-06,
+      "loss": 0.0684,
+      "step": 7159
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5747974007912903e-06,
+      "loss": 0.0706,
+      "step": 7160
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5724545223515275e-06,
+      "loss": 0.0727,
+      "step": 7161
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5701125529714265e-06,
+      "loss": 0.0696,
+      "step": 7162
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.567771492937622e-06,
+      "loss": 0.0716,
+      "step": 7163
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5654313425366407e-06,
+      "loss": 0.0713,
+      "step": 7164
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5630921020548848e-06,
+      "loss": 0.069,
+      "step": 7165
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5607537717786666e-06,
+      "loss": 0.0684,
+      "step": 7166
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5584163519941664e-06,
+      "loss": 0.0688,
+      "step": 7167
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5560798429874646e-06,
+      "loss": 0.0675,
+      "step": 7168
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.553744245044527e-06,
+      "loss": 0.0707,
+      "step": 7169
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551409558451209e-06,
+      "loss": 0.0698,
+      "step": 7170
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.549075783493252e-06,
+      "loss": 0.0703,
+      "step": 7171
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.546742920456292e-06,
+      "loss": 0.0723,
+      "step": 7172
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5444109696258434e-06,
+      "loss": 0.0709,
+      "step": 7173
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5420799312873156e-06,
+      "loss": 0.0714,
+      "step": 7174
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.539749805726006e-06,
+      "loss": 0.0763,
+      "step": 7175
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5374205932271002e-06,
+      "loss": 0.0712,
+      "step": 7176
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5350922940756696e-06,
+      "loss": 0.07,
+      "step": 7177
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5327649085566763e-06,
+      "loss": 0.0696,
+      "step": 7178
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5304384369549717e-06,
+      "loss": 0.0731,
+      "step": 7179
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5281128795552866e-06,
+      "loss": 0.0698,
+      "step": 7180
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525788236642258e-06,
+      "loss": 0.0732,
+      "step": 7181
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5234645085003894e-06,
+      "loss": 0.073,
+      "step": 7182
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5211416954140878e-06,
+      "loss": 0.0732,
+      "step": 7183
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.51881979766764e-06,
+      "loss": 0.0672,
+      "step": 7184
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5164988155452262e-06,
+      "loss": 0.0624,
+      "step": 7185
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5141787493309158e-06,
+      "loss": 0.0705,
+      "step": 7186
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5118595993086505e-06,
+      "loss": 0.0751,
+      "step": 7187
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.509541365762286e-06,
+      "loss": 0.0698,
+      "step": 7188
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5072240489755383e-06,
+      "loss": 0.0745,
+      "step": 7189
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.504907649232038e-06,
+      "loss": 0.068,
+      "step": 7190
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502592166815281e-06,
+      "loss": 0.0706,
+      "step": 7191
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5002776020086604e-06,
+      "loss": 0.0721,
+      "step": 7192
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4979639550954605e-06,
+      "loss": 0.0684,
+      "step": 7193
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.495651226358845e-06,
+      "loss": 0.0645,
+      "step": 7194
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4933394160818756e-06,
+      "loss": 0.0669,
+      "step": 7195
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4910285245474896e-06,
+      "loss": 0.0787,
+      "step": 7196
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.488718552038519e-06,
+      "loss": 0.0718,
+      "step": 7197
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.486409498837683e-06,
+      "loss": 0.076,
+      "step": 7198
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4841013652275857e-06,
+      "loss": 0.0719,
+      "step": 7199
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.481794151490724e-06,
+      "loss": 0.0722,
+      "step": 7200
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.479487857909475e-06,
+      "loss": 0.0691,
+      "step": 7201
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4771824847661076e-06,
+      "loss": 0.0702,
+      "step": 7202
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4748780323427813e-06,
+      "loss": 0.0704,
+      "step": 7203
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4725745009215317e-06,
+      "loss": 0.0763,
+      "step": 7204
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4702718907842926e-06,
+      "loss": 0.0722,
+      "step": 7205
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4679702022128806e-06,
+      "loss": 0.072,
+      "step": 7206
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4656694354889986e-06,
+      "loss": 0.0666,
+      "step": 7207
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4633695908942445e-06,
+      "loss": 0.0739,
+      "step": 7208
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461070668710086e-06,
+      "loss": 0.0699,
+      "step": 7209
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4587726692179004e-06,
+      "loss": 0.07,
+      "step": 7210
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4564755926989293e-06,
+      "loss": 0.0675,
+      "step": 7211
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4541794394343257e-06,
+      "loss": 0.0717,
+      "step": 7212
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4518842097051032e-06,
+      "loss": 0.0762,
+      "step": 7213
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4495899037921877e-06,
+      "loss": 0.0701,
+      "step": 7214
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4472965219763713e-06,
+      "loss": 0.0762,
+      "step": 7215
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4450040645383443e-06,
+      "loss": 0.0712,
+      "step": 7216
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4427125317586855e-06,
+      "loss": 0.0719,
+      "step": 7217
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.440421923917846e-06,
+      "loss": 0.0703,
+      "step": 7218
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.438132241296187e-06,
+      "loss": 0.0695,
+      "step": 7219
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.435843484173934e-06,
+      "loss": 0.0727,
+      "step": 7220
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4335556528312133e-06,
+      "loss": 0.0694,
+      "step": 7221
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.43126874754803e-06,
+      "loss": 0.0722,
+      "step": 7222
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.428982768604281e-06,
+      "loss": 0.0675,
+      "step": 7223
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.426697716279749e-06,
+      "loss": 0.0743,
+      "step": 7224
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424413590854101e-06,
+      "loss": 0.0637,
+      "step": 7225
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.422130392606895e-06,
+      "loss": 0.0695,
+      "step": 7226
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4198481218175664e-06,
+      "loss": 0.078,
+      "step": 7227
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4175667787654454e-06,
+      "loss": 0.0701,
+      "step": 7228
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4152863637297487e-06,
+      "loss": 0.0729,
+      "step": 7229
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.413006876989574e-06,
+      "loss": 0.0719,
+      "step": 7230
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4107283188239107e-06,
+      "loss": 0.0675,
+      "step": 7231
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.40845068951163e-06,
+      "loss": 0.0656,
+      "step": 7232
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406173989331494e-06,
+      "loss": 0.0717,
+      "step": 7233
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.403898218562151e-06,
+      "loss": 0.0723,
+      "step": 7234
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.401623377482124e-06,
+      "loss": 0.0674,
+      "step": 7235
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3993494663698422e-06,
+      "loss": 0.0693,
+      "step": 7236
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3970764855036043e-06,
+      "loss": 0.0709,
+      "step": 7237
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.394804435161603e-06,
+      "loss": 0.0744,
+      "step": 7238
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.392533315621917e-06,
+      "loss": 0.0696,
+      "step": 7239
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390263127162501e-06,
+      "loss": 0.0716,
+      "step": 7240
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.387993870061216e-06,
+      "loss": 0.0648,
+      "step": 7241
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3857255445957862e-06,
+      "loss": 0.0697,
+      "step": 7242
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3834581510438425e-06,
+      "loss": 0.0716,
+      "step": 7243
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.381191689682882e-06,
+      "loss": 0.0703,
+      "step": 7244
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.378926160790309e-06,
+      "loss": 0.0663,
+      "step": 7245
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3766615646433943e-06,
+      "loss": 0.0713,
+      "step": 7246
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.374397901519303e-06,
+      "loss": 0.0734,
+      "step": 7247
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3721351716950904e-06,
+      "loss": 0.0714,
+      "step": 7248
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3698733754476832e-06,
+      "loss": 0.0679,
+      "step": 7249
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.367612513053916e-06,
+      "loss": 0.0738,
+      "step": 7250
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365352584790487e-06,
+      "loss": 0.0778,
+      "step": 7251
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.363093590933991e-06,
+      "loss": 0.0692,
+      "step": 7252
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3608355317609098e-06,
+      "loss": 0.0755,
+      "step": 7253
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358578407547607e-06,
+      "loss": 0.0677,
+      "step": 7254
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.356322218570333e-06,
+      "loss": 0.0692,
+      "step": 7255
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3540669651052205e-06,
+      "loss": 0.0698,
+      "step": 7256
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3518126474282953e-06,
+      "loss": 0.0722,
+      "step": 7257
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349559265815463e-06,
+      "loss": 0.0668,
+      "step": 7258
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3473068205425133e-06,
+      "loss": 0.0672,
+      "step": 7259
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345055311885124e-06,
+      "loss": 0.0749,
+      "step": 7260
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342804740118859e-06,
+      "loss": 0.0732,
+      "step": 7261
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3405551055191666e-06,
+      "loss": 0.0738,
+      "step": 7262
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3383064083613837e-06,
+      "loss": 0.0759,
+      "step": 7263
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3360586489207183e-06,
+      "loss": 0.0664,
+      "step": 7264
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333811827472289e-06,
+      "loss": 0.0675,
+      "step": 7265
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3315659442910697e-06,
+      "loss": 0.0714,
+      "step": 7266
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3293209996519493e-06,
+      "loss": 0.0748,
+      "step": 7267
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.327076993829679e-06,
+      "loss": 0.0721,
+      "step": 7268
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.324833927098904e-06,
+      "loss": 0.0747,
+      "step": 7269
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.322591799734154e-06,
+      "loss": 0.0734,
+      "step": 7270
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.320350612009846e-06,
+      "loss": 0.0608,
+      "step": 7271
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318110364200281e-06,
+      "loss": 0.0664,
+      "step": 7272
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315871056579635e-06,
+      "loss": 0.0722,
+      "step": 7273
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3136326894219885e-06,
+      "loss": 0.0695,
+      "step": 7274
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.311395263001289e-06,
+      "loss": 0.0679,
+      "step": 7275
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.309158777591378e-06,
+      "loss": 0.0696,
+      "step": 7276
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3069232334659797e-06,
+      "loss": 0.0757,
+      "step": 7277
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3046886308987017e-06,
+      "loss": 0.076,
+      "step": 7278
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3024549701630393e-06,
+      "loss": 0.0683,
+      "step": 7279
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.300222251532371e-06,
+      "loss": 0.0691,
+      "step": 7280
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.297990475279962e-06,
+      "loss": 0.0747,
+      "step": 7281
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295759641678955e-06,
+      "loss": 0.0684,
+      "step": 7282
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.293529751002386e-06,
+      "loss": 0.073,
+      "step": 7283
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.29130080352317e-06,
+      "loss": 0.0645,
+      "step": 7284
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.289072799514109e-06,
+      "loss": 0.0655,
+      "step": 7285
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2868457392478904e-06,
+      "loss": 0.0732,
+      "step": 7286
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2846196229970873e-06,
+      "loss": 0.0684,
+      "step": 7287
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2823944510341454e-06,
+      "loss": 0.073,
+      "step": 7288
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.280170223631417e-06,
+      "loss": 0.0715,
+      "step": 7289
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2779469410611166e-06,
+      "loss": 0.0679,
+      "step": 7290
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275724603595354e-06,
+      "loss": 0.0768,
+      "step": 7291
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.273503211506124e-06,
+      "loss": 0.0656,
+      "step": 7292
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2712827650653035e-06,
+      "loss": 0.0683,
+      "step": 7293
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.269063264544654e-06,
+      "loss": 0.0681,
+      "step": 7294
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2668447102158153e-06,
+      "loss": 0.073,
+      "step": 7295
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264627102350326e-06,
+      "loss": 0.0664,
+      "step": 7296
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26241044121959e-06,
+      "loss": 0.0701,
+      "step": 7297
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2601947270949164e-06,
+      "loss": 0.0715,
+      "step": 7298
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2579799602474784e-06,
+      "loss": 0.0686,
+      "step": 7299
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2557661409483432e-06,
+      "loss": 0.0682,
+      "step": 7300
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2535532694684635e-06,
+      "loss": 0.0683,
+      "step": 7301
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2513413460786726e-06,
+      "loss": 0.0703,
+      "step": 7302
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2491303710496903e-06,
+      "loss": 0.0694,
+      "step": 7303
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2469203446521105e-06,
+      "loss": 0.0638,
+      "step": 7304
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2447112671564307e-06,
+      "loss": 0.069,
+      "step": 7305
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242503138833012e-06,
+      "loss": 0.0685,
+      "step": 7306
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.24029595995211e-06,
+      "loss": 0.0644,
+      "step": 7307
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2380897307838646e-06,
+      "loss": 0.0639,
+      "step": 7308
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2358844515982938e-06,
+      "loss": 0.0685,
+      "step": 7309
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2336801226653037e-06,
+      "loss": 0.0656,
+      "step": 7310
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2314767442546825e-06,
+      "loss": 0.0689,
+      "step": 7311
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2292743166361065e-06,
+      "loss": 0.0681,
+      "step": 7312
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.227072840079123e-06,
+      "loss": 0.0692,
+      "step": 7313
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224872314853178e-06,
+      "loss": 0.0639,
+      "step": 7314
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.222672741227592e-06,
+      "loss": 0.0773,
+      "step": 7315
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2204741194715716e-06,
+      "loss": 0.0719,
+      "step": 7316
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2182764498542073e-06,
+      "loss": 0.0688,
+      "step": 7317
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.216079732644476e-06,
+      "loss": 0.0641,
+      "step": 7318
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2138839681112256e-06,
+      "loss": 0.0604,
+      "step": 7319
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211689156523208e-06,
+      "loss": 0.0728,
+      "step": 7320
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2094952981490383e-06,
+      "loss": 0.0685,
+      "step": 7321
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2073023932572246e-06,
+      "loss": 0.0711,
+      "step": 7322
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2051104421161607e-06,
+      "loss": 0.0733,
+      "step": 7323
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.202919444994117e-06,
+      "loss": 0.0766,
+      "step": 7324
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.200729402159255e-06,
+      "loss": 0.0668,
+      "step": 7325
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.198540313879606e-06,
+      "loss": 0.0655,
+      "step": 7326
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1963521804231036e-06,
+      "loss": 0.0686,
+      "step": 7327
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.194165002057543e-06,
+      "loss": 0.0713,
+      "step": 7328
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1919787790506265e-06,
+      "loss": 0.0718,
+      "step": 7329
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1897935116699163e-06,
+      "loss": 0.0679,
+      "step": 7330
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.187609200182873e-06,
+      "loss": 0.0661,
+      "step": 7331
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.185425844856833e-06,
+      "loss": 0.0736,
+      "step": 7332
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.183243445959018e-06,
+      "loss": 0.0718,
+      "step": 7333
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1810620037565365e-06,
+      "loss": 0.0697,
+      "step": 7334
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.178881518516367e-06,
+      "loss": 0.0788,
+      "step": 7335
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.176701990505391e-06,
+      "loss": 0.0619,
+      "step": 7336
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.174523419990353e-06,
+      "loss": 0.0729,
+      "step": 7337
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1723458072378924e-06,
+      "loss": 0.0722,
+      "step": 7338
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1701691525145275e-06,
+      "loss": 0.0732,
+      "step": 7339
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.167993456086659e-06,
+      "loss": 0.0665,
+      "step": 7340
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1658187182205716e-06,
+      "loss": 0.0646,
+      "step": 7341
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1636449391824356e-06,
+      "loss": 0.0736,
+      "step": 7342
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1614721192382947e-06,
+      "loss": 0.0713,
+      "step": 7343
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1593002586540836e-06,
+      "loss": 0.0706,
+      "step": 7344
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1571293576956166e-06,
+      "loss": 0.0696,
+      "step": 7345
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1549594166285915e-06,
+      "loss": 0.0704,
+      "step": 7346
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.152790435718587e-06,
+      "loss": 0.0721,
+      "step": 7347
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.150622415231067e-06,
+      "loss": 0.0709,
+      "step": 7348
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1484553554313793e-06,
+      "loss": 0.071,
+      "step": 7349
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1462892565847404e-06,
+      "loss": 0.0735,
+      "step": 7350
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1441241189562722e-06,
+      "loss": 0.0784,
+      "step": 7351
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1419599428109592e-06,
+      "loss": 0.0697,
+      "step": 7352
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1397967284136767e-06,
+      "loss": 0.0709,
+      "step": 7353
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.137634476029182e-06,
+      "loss": 0.065,
+      "step": 7354
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1354731859221145e-06,
+      "loss": 0.0725,
+      "step": 7355
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.133312858356997e-06,
+      "loss": 0.0649,
+      "step": 7356
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131153493598225e-06,
+      "loss": 0.0723,
+      "step": 7357
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1289950919100932e-06,
+      "loss": 0.0736,
+      "step": 7358
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1268376535567613e-06,
+      "loss": 0.0702,
+      "step": 7359
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.124681178802287e-06,
+      "loss": 0.0719,
+      "step": 7360
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1225256679105943e-06,
+      "loss": 0.0718,
+      "step": 7361
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1203711211455e-06,
+      "loss": 0.0712,
+      "step": 7362
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1182175387707007e-06,
+      "loss": 0.0642,
+      "step": 7363
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.116064921049773e-06,
+      "loss": 0.0695,
+      "step": 7364
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1139132682461783e-06,
+      "loss": 0.07,
+      "step": 7365
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1117625806232545e-06,
+      "loss": 0.071,
+      "step": 7366
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.109612858444227e-06,
+      "loss": 0.0742,
+      "step": 7367
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1074641019721997e-06,
+      "loss": 0.0734,
+      "step": 7368
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1053163114701604e-06,
+      "loss": 0.0683,
+      "step": 7369
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.103169487200979e-06,
+      "loss": 0.0651,
+      "step": 7370
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1010236294274056e-06,
+      "loss": 0.0666,
+      "step": 7371
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.098878738412071e-06,
+      "loss": 0.0708,
+      "step": 7372
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0967348144174924e-06,
+      "loss": 0.0729,
+      "step": 7373
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.094591857706061e-06,
+      "loss": 0.0714,
+      "step": 7374
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0924498685400565e-06,
+      "loss": 0.0738,
+      "step": 7375
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0903088471816358e-06,
+      "loss": 0.0653,
+      "step": 7376
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0881687938928417e-06,
+      "loss": 0.0726,
+      "step": 7377
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0860297089355943e-06,
+      "loss": 0.0705,
+      "step": 7378
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.083891592571696e-06,
+      "loss": 0.0733,
+      "step": 7379
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.081754445062838e-06,
+      "loss": 0.068,
+      "step": 7380
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0796182666705743e-06,
+      "loss": 0.0685,
+      "step": 7381
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0774830576563653e-06,
+      "loss": 0.0692,
+      "step": 7382
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0753488182815297e-06,
+      "loss": 0.0696,
+      "step": 7383
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.073215548807286e-06,
+      "loss": 0.0781,
+      "step": 7384
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.07108324949472e-06,
+      "loss": 0.0708,
+      "step": 7385
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0689519206048058e-06,
+      "loss": 0.0675,
+      "step": 7386
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.066821562398401e-06,
+      "loss": 0.079,
+      "step": 7387
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.064692175136234e-06,
+      "loss": 0.0687,
+      "step": 7388
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.062563759078928e-06,
+      "loss": 0.0774,
+      "step": 7389
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0604363144869732e-06,
+      "loss": 0.0784,
+      "step": 7390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.058309841620758e-06,
+      "loss": 0.0687,
+      "step": 7391
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0561843407405345e-06,
+      "loss": 0.0685,
+      "step": 7392
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.054059812106445e-06,
+      "loss": 0.0764,
+      "step": 7393
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0519362559785116e-06,
+      "loss": 0.0734,
+      "step": 7394
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0498136726166386e-06,
+      "loss": 0.0675,
+      "step": 7395
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047692062280606e-06,
+      "loss": 0.0651,
+      "step": 7396
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045571425230085e-06,
+      "loss": 0.0675,
+      "step": 7397
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.043451761724614e-06,
+      "loss": 0.0682,
+      "step": 7398
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.041333072023621e-06,
+      "loss": 0.0711,
+      "step": 7399
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0392153563864157e-06,
+      "loss": 0.0714,
+      "step": 7400
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.037098615072184e-06,
+      "loss": 0.0722,
+      "step": 7401
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0349828483399936e-06,
+      "loss": 0.073,
+      "step": 7402
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.032868056448797e-06,
+      "loss": 0.0708,
+      "step": 7403
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.030754239657423e-06,
+      "loss": 0.0679,
+      "step": 7404
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0286413982245776e-06,
+      "loss": 0.073,
+      "step": 7405
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.026529532408862e-06,
+      "loss": 0.0742,
+      "step": 7406
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0244186424687397e-06,
+      "loss": 0.0685,
+      "step": 7407
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0223087286625665e-06,
+      "loss": 0.0638,
+      "step": 7408
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.020199791248576e-06,
+      "loss": 0.0734,
+      "step": 7409
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0180918304848764e-06,
+      "loss": 0.0699,
+      "step": 7410
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0159848466294706e-06,
+      "loss": 0.0635,
+      "step": 7411
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.013878839940223e-06,
+      "loss": 0.0688,
+      "step": 7412
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0117738106748998e-06,
+      "loss": 0.0734,
+      "step": 7413
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009669759091124e-06,
+      "loss": 0.0657,
+      "step": 7414
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.007566685446424e-06,
+      "loss": 0.0698,
+      "step": 7415
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.005464589998185e-06,
+      "loss": 0.0678,
+      "step": 7416
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0033634730036877e-06,
+      "loss": 0.0663,
+      "step": 7417
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0012633347200873e-06,
+      "loss": 0.069,
+      "step": 7418
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.999164175404421e-06,
+      "loss": 0.0662,
+      "step": 7419
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9970659953136083e-06,
+      "loss": 0.0711,
+      "step": 7420
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9949687947044405e-06,
+      "loss": 0.0733,
+      "step": 7421
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9928725738335985e-06,
+      "loss": 0.0665,
+      "step": 7422
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9907773329576372e-06,
+      "loss": 0.071,
+      "step": 7423
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9886830723329963e-06,
+      "loss": 0.0759,
+      "step": 7424
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9865897922159904e-06,
+      "loss": 0.0749,
+      "step": 7425
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.984497492862819e-06,
+      "loss": 0.0638,
+      "step": 7426
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9824061745295587e-06,
+      "loss": 0.0706,
+      "step": 7427
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.980315837472169e-06,
+      "loss": 0.0725,
+      "step": 7428
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9782264819464803e-06,
+      "loss": 0.0694,
+      "step": 7429
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9761381082082144e-06,
+      "loss": 0.0696,
+      "step": 7430
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9740507165129676e-06,
+      "loss": 0.0647,
+      "step": 7431
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9719643071162143e-06,
+      "loss": 0.0713,
+      "step": 7432
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.969878880273316e-06,
+      "loss": 0.0665,
+      "step": 7433
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9677944362394996e-06,
+      "loss": 0.0667,
+      "step": 7434
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.965710975269892e-06,
+      "loss": 0.0723,
+      "step": 7435
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9636284976194776e-06,
+      "loss": 0.0713,
+      "step": 7436
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9615470035431426e-06,
+      "loss": 0.0714,
+      "step": 7437
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.959466493295633e-06,
+      "loss": 0.0688,
+      "step": 7438
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.957386967131587e-06,
+      "loss": 0.0684,
+      "step": 7439
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9553084253055164e-06,
+      "loss": 0.063,
+      "step": 7440
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953230868071817e-06,
+      "loss": 0.0647,
+      "step": 7441
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.951154295684763e-06,
+      "loss": 0.0681,
+      "step": 7442
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9490787083984985e-06,
+      "loss": 0.0706,
+      "step": 7443
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9470041064670675e-06,
+      "loss": 0.0685,
+      "step": 7444
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944930490144371e-06,
+      "loss": 0.0735,
+      "step": 7445
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.942857859684204e-06,
+      "loss": 0.0723,
+      "step": 7446
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9407862153402355e-06,
+      "loss": 0.0694,
+      "step": 7447
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9387155573660154e-06,
+      "loss": 0.0724,
+      "step": 7448
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9366458860149718e-06,
+      "loss": 0.0664,
+      "step": 7449
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.934577201540412e-06,
+      "loss": 0.0685,
+      "step": 7450
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9325095041955264e-06,
+      "loss": 0.0669,
+      "step": 7451
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.930442794233376e-06,
+      "loss": 0.0731,
+      "step": 7452
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.928377071906907e-06,
+      "loss": 0.0722,
+      "step": 7453
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9263123374689465e-06,
+      "loss": 0.0652,
+      "step": 7454
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9242485911721954e-06,
+      "loss": 0.0694,
+      "step": 7455
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9221858332692376e-06,
+      "loss": 0.0705,
+      "step": 7456
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.920124064012534e-06,
+      "loss": 0.0731,
+      "step": 7457
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9180632836544268e-06,
+      "loss": 0.0689,
+      "step": 7458
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.916003492447136e-06,
+      "loss": 0.0707,
+      "step": 7459
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.913944690642757e-06,
+      "loss": 0.0714,
+      "step": 7460
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9118868784932674e-06,
+      "loss": 0.072,
+      "step": 7461
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.0713,
+      "step": 7462
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9077742241662665e-06,
+      "loss": 0.0696,
+      "step": 7463
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.905719382492106e-06,
+      "loss": 0.0704,
+      "step": 7464
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9036655314795294e-06,
+      "loss": 0.0745,
+      "step": 7465
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9016126713799176e-06,
+      "loss": 0.0686,
+      "step": 7466
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8995608024445122e-06,
+      "loss": 0.0722,
+      "step": 7467
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.89750992492445e-06,
+      "loss": 0.0713,
+      "step": 7468
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8954600390707335e-06,
+      "loss": 0.0689,
+      "step": 7469
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8934111451342497e-06,
+      "loss": 0.0652,
+      "step": 7470
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8913632433657636e-06,
+      "loss": 0.0661,
+      "step": 7471
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8893163340159194e-06,
+      "loss": 0.0653,
+      "step": 7472
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8872704173352407e-06,
+      "loss": 0.0671,
+      "step": 7473
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8852254935741198e-06,
+      "loss": 0.0646,
+      "step": 7474
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8831815629828475e-06,
+      "loss": 0.0685,
+      "step": 7475
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8811386258115716e-06,
+      "loss": 0.075,
+      "step": 7476
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8790966823103319e-06,
+      "loss": 0.0733,
+      "step": 7477
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8770557327290418e-06,
+      "loss": 0.0676,
+      "step": 7478
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8750157773174938e-06,
+      "loss": 0.0673,
+      "step": 7479
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8729768163253581e-06,
+      "loss": 0.071,
+      "step": 7480
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8709388500021852e-06,
+      "loss": 0.0628,
+      "step": 7481
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8689018785974046e-06,
+      "loss": 0.0623,
+      "step": 7482
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8668659023603154e-06,
+      "loss": 0.066,
+      "step": 7483
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.864830921540105e-06,
+      "loss": 0.0686,
+      "step": 7484
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8627969363858357e-06,
+      "loss": 0.077,
+      "step": 7485
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8607639471464467e-06,
+      "loss": 0.07,
+      "step": 7486
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.858731954070756e-06,
+      "loss": 0.0714,
+      "step": 7487
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8567009574074645e-06,
+      "loss": 0.0682,
+      "step": 7488
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.854670957405137e-06,
+      "loss": 0.0689,
+      "step": 7489
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8526419543122365e-06,
+      "loss": 0.0689,
+      "step": 7490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8506139483770858e-06,
+      "loss": 0.0725,
+      "step": 7491
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8485869398478951e-06,
+      "loss": 0.0667,
+      "step": 7492
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8465609289727514e-06,
+      "loss": 0.0686,
+      "step": 7493
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.844535915999619e-06,
+      "loss": 0.0687,
+      "step": 7494
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8425119011763415e-06,
+      "loss": 0.0691,
+      "step": 7495
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8404888847506318e-06,
+      "loss": 0.0666,
+      "step": 7496
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8384668669700979e-06,
+      "loss": 0.0749,
+      "step": 7497
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8364458480822045e-06,
+      "loss": 0.0662,
+      "step": 7498
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8344258283343142e-06,
+      "loss": 0.0647,
+      "step": 7499
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8324068079736523e-06,
+      "loss": 0.0743,
+      "step": 7500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.830388787247328e-06,
+      "loss": 0.0745,
+      "step": 7501
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8283717664023282e-06,
+      "loss": 0.0661,
+      "step": 7502
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8263557456855163e-06,
+      "loss": 0.0663,
+      "step": 7503
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8243407253436373e-06,
+      "loss": 0.0654,
+      "step": 7504
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8223267056233018e-06,
+      "loss": 0.0661,
+      "step": 7505
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8203136867710169e-06,
+      "loss": 0.0738,
+      "step": 7506
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8183016690331478e-06,
+      "loss": 0.0695,
+      "step": 7507
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8162906526559498e-06,
+      "loss": 0.072,
+      "step": 7508
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8142806378855515e-06,
+      "loss": 0.0677,
+      "step": 7509
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8122716249679596e-06,
+      "loss": 0.0695,
+      "step": 7510
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.810263614149057e-06,
+      "loss": 0.0666,
+      "step": 7511
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8082566056746087e-06,
+      "loss": 0.0754,
+      "step": 7512
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8062505997902458e-06,
+      "loss": 0.0623,
+      "step": 7513
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8042455967414884e-06,
+      "loss": 0.0687,
+      "step": 7514
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8022415967737294e-06,
+      "loss": 0.0763,
+      "step": 7515
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8002386001322381e-06,
+      "loss": 0.0701,
+      "step": 7516
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7982366070621626e-06,
+      "loss": 0.0678,
+      "step": 7517
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.796235617808526e-06,
+      "loss": 0.0674,
+      "step": 7518
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7942356326162347e-06,
+      "loss": 0.0684,
+      "step": 7519
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7922366517300594e-06,
+      "loss": 0.0741,
+      "step": 7520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7902386753946642e-06,
+      "loss": 0.0684,
+      "step": 7521
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7882417038545752e-06,
+      "loss": 0.0672,
+      "step": 7522
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7862457373542096e-06,
+      "loss": 0.0684,
+      "step": 7523
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.784250776137848e-06,
+      "loss": 0.0716,
+      "step": 7524
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7822568204496571e-06,
+      "loss": 0.068,
+      "step": 7525
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7802638705336794e-06,
+      "loss": 0.0708,
+      "step": 7526
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7782719266338267e-06,
+      "loss": 0.0683,
+      "step": 7527
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7762809889939014e-06,
+      "loss": 0.0767,
+      "step": 7528
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7742910578575679e-06,
+      "loss": 0.069,
+      "step": 7529
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.772302133468381e-06,
+      "loss": 0.065,
+      "step": 7530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7703142160697618e-06,
+      "loss": 0.0771,
+      "step": 7531
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.768327305905012e-06,
+      "loss": 0.0705,
+      "step": 7532
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7663414032173121e-06,
+      "loss": 0.0676,
+      "step": 7533
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.764356508249715e-06,
+      "loss": 0.0672,
+      "step": 7534
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7623726212451575e-06,
+      "loss": 0.0715,
+      "step": 7535
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.76038974244644e-06,
+      "loss": 0.0704,
+      "step": 7536
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7584078720962572e-06,
+      "loss": 0.0703,
+      "step": 7537
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7564270104371638e-06,
+      "loss": 0.0685,
+      "step": 7538
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7544471577115995e-06,
+      "loss": 0.0627,
+      "step": 7539
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7524683141618815e-06,
+      "loss": 0.0741,
+      "step": 7540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750490480030198e-06,
+      "loss": 0.0715,
+      "step": 7541
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.748513655558619e-06,
+      "loss": 0.0689,
+      "step": 7542
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7465378409890899e-06,
+      "loss": 0.0651,
+      "step": 7543
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7445630365634247e-06,
+      "loss": 0.0694,
+      "step": 7544
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7425892425233304e-06,
+      "loss": 0.0721,
+      "step": 7545
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7406164591103724e-06,
+      "loss": 0.0705,
+      "step": 7546
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7386446865660023e-06,
+      "loss": 0.0762,
+      "step": 7547
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7366739251315468e-06,
+      "loss": 0.0645,
+      "step": 7548
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7347041750482075e-06,
+      "loss": 0.0776,
+      "step": 7549
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7327354365570647e-06,
+      "loss": 0.0722,
+      "step": 7550
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7307677098990671e-06,
+      "loss": 0.0659,
+      "step": 7551
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.728800995315054e-06,
+      "loss": 0.0677,
+      "step": 7552
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7268352930457232e-06,
+      "loss": 0.0661,
+      "step": 7553
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7248706033316665e-06,
+      "loss": 0.0651,
+      "step": 7554
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7229069264133368e-06,
+      "loss": 0.0732,
+      "step": 7555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7209442625310701e-06,
+      "loss": 0.0804,
+      "step": 7556
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7189826119250808e-06,
+      "loss": 0.0632,
+      "step": 7557
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7170219748354488e-06,
+      "loss": 0.067,
+      "step": 7558
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.715062351502146e-06,
+      "loss": 0.0707,
+      "step": 7559
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.713103742165002e-06,
+      "loss": 0.0705,
+      "step": 7560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7111461470637414e-06,
+      "loss": 0.0721,
+      "step": 7561
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7091895664379488e-06,
+      "loss": 0.0733,
+      "step": 7562
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7072340005270904e-06,
+      "loss": 0.0697,
+      "step": 7563
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7052794495705115e-06,
+      "loss": 0.0698,
+      "step": 7564
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7033259138074275e-06,
+      "loss": 0.0648,
+      "step": 7565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7013733934769349e-06,
+      "loss": 0.0701,
+      "step": 7566
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6994218888180036e-06,
+      "loss": 0.0765,
+      "step": 7567
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6974714000694758e-06,
+      "loss": 0.0709,
+      "step": 7568
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.695521927470073e-06,
+      "loss": 0.0649,
+      "step": 7569
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6935734712583929e-06,
+      "loss": 0.0655,
+      "step": 7570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6916260316729072e-06,
+      "loss": 0.0698,
+      "step": 7571
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6896796089519651e-06,
+      "loss": 0.0709,
+      "step": 7572
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6877342033337873e-06,
+      "loss": 0.0747,
+      "step": 7573
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6857898150564779e-06,
+      "loss": 0.0698,
+      "step": 7574
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6838464443580006e-06,
+      "loss": 0.0669,
+      "step": 7575
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6819040914762185e-06,
+      "loss": 0.0792,
+      "step": 7576
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6799627566488474e-06,
+      "loss": 0.0736,
+      "step": 7577
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6780224401134903e-06,
+      "loss": 0.066,
+      "step": 7578
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.676083142107623e-06,
+      "loss": 0.0668,
+      "step": 7579
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6741448628685985e-06,
+      "loss": 0.07,
+      "step": 7580
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6722076026336443e-06,
+      "loss": 0.0648,
+      "step": 7581
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6702713616398547e-06,
+      "loss": 0.0678,
+      "step": 7582
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6683361401242182e-06,
+      "loss": 0.0656,
+      "step": 7583
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6664019383235764e-06,
+      "loss": 0.0722,
+      "step": 7584
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6644687564746665e-06,
+      "loss": 0.0709,
+      "step": 7585
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6625365948140837e-06,
+      "loss": 0.0655,
+      "step": 7586
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.660605453578309e-06,
+      "loss": 0.0739,
+      "step": 7587
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6586753330036963e-06,
+      "loss": 0.0751,
+      "step": 7588
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6567462333264718e-06,
+      "loss": 0.0724,
+      "step": 7589
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6548181547827413e-06,
+      "loss": 0.0605,
+      "step": 7590
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.652891097608479e-06,
+      "loss": 0.0695,
+      "step": 7591
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6509650620395402e-06,
+      "loss": 0.0685,
+      "step": 7592
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.649040048311653e-06,
+      "loss": 0.0682,
+      "step": 7593
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6471160566604194e-06,
+      "loss": 0.0649,
+      "step": 7594
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6451930873213196e-06,
+      "loss": 0.0638,
+      "step": 7595
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6432711405297043e-06,
+      "loss": 0.065,
+      "step": 7596
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6413502165208028e-06,
+      "loss": 0.0642,
+      "step": 7597
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6394303155297186e-06,
+      "loss": 0.0734,
+      "step": 7598
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6375114377914247e-06,
+      "loss": 0.0708,
+      "step": 7599
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6355935835407755e-06,
+      "loss": 0.064,
+      "step": 7600
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6336767530124986e-06,
+      "loss": 0.0731,
+      "step": 7601
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6317609464411942e-06,
+      "loss": 0.0711,
+      "step": 7602
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6298461640613394e-06,
+      "loss": 0.0707,
+      "step": 7603
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6279324061072843e-06,
+      "loss": 0.0744,
+      "step": 7604
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6260196728132583e-06,
+      "loss": 0.0737,
+      "step": 7605
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6241079644133518e-06,
+      "loss": 0.0697,
+      "step": 7606
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.622197281141551e-06,
+      "loss": 0.0724,
+      "step": 7607
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6202876232316977e-06,
+      "loss": 0.0701,
+      "step": 7608
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6183789909175163e-06,
+      "loss": 0.0732,
+      "step": 7609
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6164713844326075e-06,
+      "loss": 0.0655,
+      "step": 7610
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6145648040104411e-06,
+      "loss": 0.0708,
+      "step": 7611
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6126592498843686e-06,
+      "loss": 0.0723,
+      "step": 7612
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.610754722287603e-06,
+      "loss": 0.0638,
+      "step": 7613
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6088512214532492e-06,
+      "loss": 0.0668,
+      "step": 7614
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6069487476142697e-06,
+      "loss": 0.0734,
+      "step": 7615
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.605047301003516e-06,
+      "loss": 0.0702,
+      "step": 7616
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6031468818537021e-06,
+      "loss": 0.0735,
+      "step": 7617
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6012474903974218e-06,
+      "loss": 0.0692,
+      "step": 7618
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5993491268671423e-06,
+      "loss": 0.0711,
+      "step": 7619
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5974517914952048e-06,
+      "loss": 0.0693,
+      "step": 7620
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5955554845138277e-06,
+      "loss": 0.0656,
+      "step": 7621
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5936602061550955e-06,
+      "loss": 0.0694,
+      "step": 7622
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5917659566509746e-06,
+      "loss": 0.0705,
+      "step": 7623
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.589872736233302e-06,
+      "loss": 0.0699,
+      "step": 7624
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5879805451337905e-06,
+      "loss": 0.0643,
+      "step": 7625
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5860893835840253e-06,
+      "loss": 0.0698,
+      "step": 7626
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.584199251815467e-06,
+      "loss": 0.0679,
+      "step": 7627
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5823101500594495e-06,
+      "loss": 0.0716,
+      "step": 7628
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5804220785471814e-06,
+      "loss": 0.0699,
+      "step": 7629
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5785350375097418e-06,
+      "loss": 0.0681,
+      "step": 7630
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5766490271780865e-06,
+      "loss": 0.0702,
+      "step": 7631
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5747640477830462e-06,
+      "loss": 0.0732,
+      "step": 7632
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.572880099555324e-06,
+      "loss": 0.07,
+      "step": 7633
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5709971827254988e-06,
+      "loss": 0.0643,
+      "step": 7634
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5691152975240153e-06,
+      "loss": 0.0667,
+      "step": 7635
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.567234444181206e-06,
+      "loss": 0.0777,
+      "step": 7636
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.565354622927261e-06,
+      "loss": 0.0681,
+      "step": 7637
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5634758339922619e-06,
+      "loss": 0.0724,
+      "step": 7638
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5615980776061456e-06,
+      "loss": 0.0601,
+      "step": 7639
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5597213539987344e-06,
+      "loss": 0.0718,
+      "step": 7640
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5578456633997209e-06,
+      "loss": 0.0699,
+      "step": 7641
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5559710060386724e-06,
+      "loss": 0.0641,
+      "step": 7642
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5540973821450312e-06,
+      "loss": 0.0707,
+      "step": 7643
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5522247919481026e-06,
+      "loss": 0.067,
+      "step": 7644
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5503532356770822e-06,
+      "loss": 0.069,
+      "step": 7645
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5484827135610259e-06,
+      "loss": 0.073,
+      "step": 7646
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546613225828868e-06,
+      "loss": 0.072,
+      "step": 7647
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5447447727094168e-06,
+      "loss": 0.0725,
+      "step": 7648
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5428773544313514e-06,
+      "loss": 0.0637,
+      "step": 7649
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5410109712232258e-06,
+      "loss": 0.0648,
+      "step": 7650
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5391456233134693e-06,
+      "loss": 0.0789,
+      "step": 7651
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.537281310930383e-06,
+      "loss": 0.07,
+      "step": 7652
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5354180343021364e-06,
+      "loss": 0.0616,
+      "step": 7653
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.533555793656779e-06,
+      "loss": 0.0702,
+      "step": 7654
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5316945892222302e-06,
+      "loss": 0.0743,
+      "step": 7655
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.529834421226285e-06,
+      "loss": 0.0709,
+      "step": 7656
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5279752898966094e-06,
+      "loss": 0.0675,
+      "step": 7657
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5261171954607435e-06,
+      "loss": 0.0696,
+      "step": 7658
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5242601381460964e-06,
+      "loss": 0.0686,
+      "step": 7659
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.522404118179961e-06,
+      "loss": 0.0727,
+      "step": 7660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5205491357894886e-06,
+      "loss": 0.0758,
+      "step": 7661
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5186951912017156e-06,
+      "loss": 0.0671,
+      "step": 7662
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.516842284643546e-06,
+      "loss": 0.0637,
+      "step": 7663
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5149904163417573e-06,
+      "loss": 0.0675,
+      "step": 7664
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5131395865230026e-06,
+      "loss": 0.0687,
+      "step": 7665
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5112897954137984e-06,
+      "loss": 0.0694,
+      "step": 7666
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5094410432405504e-06,
+      "loss": 0.0776,
+      "step": 7667
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5075933302295198e-06,
+      "loss": 0.0727,
+      "step": 7668
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.505746656606858e-06,
+      "loss": 0.0723,
+      "step": 7669
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5039010225985683e-06,
+      "loss": 0.0618,
+      "step": 7670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5020564284305506e-06,
+      "loss": 0.0623,
+      "step": 7671
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.500212874328557e-06,
+      "loss": 0.071,
+      "step": 7672
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4983703605182243e-06,
+      "loss": 0.0749,
+      "step": 7673
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4965288872250582e-06,
+      "loss": 0.0704,
+      "step": 7674
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4946884546744323e-06,
+      "loss": 0.0629,
+      "step": 7675
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4928490630916059e-06,
+      "loss": 0.0648,
+      "step": 7676
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.491010712701697e-06,
+      "loss": 0.072,
+      "step": 7677
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4891734037297035e-06,
+      "loss": 0.0717,
+      "step": 7678
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.487337136400493e-06,
+      "loss": 0.0684,
+      "step": 7679
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4855019109388091e-06,
+      "loss": 0.07,
+      "step": 7680
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4836677275692646e-06,
+      "loss": 0.069,
+      "step": 7681
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4818345865163452e-06,
+      "loss": 0.0706,
+      "step": 7682
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4800024880044106e-06,
+      "loss": 0.0688,
+      "step": 7683
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4781714322576935e-06,
+      "loss": 0.0669,
+      "step": 7684
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4763414195002934e-06,
+      "loss": 0.074,
+      "step": 7685
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.474512449956188e-06,
+      "loss": 0.067,
+      "step": 7686
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.472684523849226e-06,
+      "loss": 0.0662,
+      "step": 7687
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4708576414031273e-06,
+      "loss": 0.0683,
+      "step": 7688
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4690318028414864e-06,
+      "loss": 0.0656,
+      "step": 7689
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.467207008387762e-06,
+      "loss": 0.068,
+      "step": 7690
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4653832582653015e-06,
+      "loss": 0.0723,
+      "step": 7691
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.463560552697304e-06,
+      "loss": 0.0686,
+      "step": 7692
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4617388919068598e-06,
+      "loss": 0.0651,
+      "step": 7693
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4599182761169162e-06,
+      "loss": 0.062,
+      "step": 7694
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4580987055503016e-06,
+      "loss": 0.0764,
+      "step": 7695
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4562801804297155e-06,
+      "loss": 0.0668,
+      "step": 7696
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4544627009777213e-06,
+      "loss": 0.0718,
+      "step": 7697
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4526462674167685e-06,
+      "loss": 0.0655,
+      "step": 7698
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.450830879969165e-06,
+      "loss": 0.0699,
+      "step": 7699
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4490165388571032e-06,
+      "loss": 0.0665,
+      "step": 7700
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4472032443026351e-06,
+      "loss": 0.07,
+      "step": 7701
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4453909965276925e-06,
+      "loss": 0.0711,
+      "step": 7702
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4435797957540765e-06,
+      "loss": 0.0727,
+      "step": 7703
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.441769642203461e-06,
+      "loss": 0.0714,
+      "step": 7704
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4399605360973912e-06,
+      "loss": 0.0628,
+      "step": 7705
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4381524776572842e-06,
+      "loss": 0.0719,
+      "step": 7706
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4363454671044309e-06,
+      "loss": 0.0682,
+      "step": 7707
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4345395046599885e-06,
+      "loss": 0.0684,
+      "step": 7708
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4327345905449896e-06,
+      "loss": 0.0699,
+      "step": 7709
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4309307249803394e-06,
+      "loss": 0.0652,
+      "step": 7710
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4291279081868138e-06,
+      "loss": 0.0722,
+      "step": 7711
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4273261403850602e-06,
+      "loss": 0.0645,
+      "step": 7712
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4255254217955983e-06,
+      "loss": 0.0623,
+      "step": 7713
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4237257526388137e-06,
+      "loss": 0.0671,
+      "step": 7714
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.421927133134976e-06,
+      "loss": 0.0745,
+      "step": 7715
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4201295635042135e-06,
+      "loss": 0.0672,
+      "step": 7716
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.418333043966532e-06,
+      "loss": 0.0687,
+      "step": 7717
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4165375747418097e-06,
+      "loss": 0.0736,
+      "step": 7718
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4147431560497937e-06,
+      "loss": 0.07,
+      "step": 7719
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4129497881101063e-06,
+      "loss": 0.0693,
+      "step": 7720
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4111574711422305e-06,
+      "loss": 0.0716,
+      "step": 7721
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4093662053655389e-06,
+      "loss": 0.0656,
+      "step": 7722
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4075759909992547e-06,
+      "loss": 0.0674,
+      "step": 7723
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4057868282624932e-06,
+      "loss": 0.0652,
+      "step": 7724
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.403998717374223e-06,
+      "loss": 0.0701,
+      "step": 7725
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.402211658553294e-06,
+      "loss": 0.0666,
+      "step": 7726
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4004256520184256e-06,
+      "loss": 0.065,
+      "step": 7727
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3986406979882073e-06,
+      "loss": 0.0701,
+      "step": 7728
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3968567966811008e-06,
+      "loss": 0.0692,
+      "step": 7729
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3950739483154342e-06,
+      "loss": 0.0656,
+      "step": 7730
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.393292153109418e-06,
+      "loss": 0.0728,
+      "step": 7731
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3915114112811213e-06,
+      "loss": 0.0711,
+      "step": 7732
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3897317230484907e-06,
+      "loss": 0.0602,
+      "step": 7733
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3879530886293425e-06,
+      "loss": 0.0734,
+      "step": 7734
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3861755082413663e-06,
+      "loss": 0.0684,
+      "step": 7735
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3843989821021197e-06,
+      "loss": 0.0625,
+      "step": 7736
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.382623510429034e-06,
+      "loss": 0.066,
+      "step": 7737
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3808490934394058e-06,
+      "loss": 0.0664,
+      "step": 7738
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3790757313504089e-06,
+      "loss": 0.0705,
+      "step": 7739
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.377303424379085e-06,
+      "loss": 0.0666,
+      "step": 7740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3755321727423488e-06,
+      "loss": 0.071,
+      "step": 7741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.373761976656982e-06,
+      "loss": 0.0671,
+      "step": 7742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3719928363396418e-06,
+      "loss": 0.0731,
+      "step": 7743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3702247520068546e-06,
+      "loss": 0.0688,
+      "step": 7744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3684577238750108e-06,
+      "loss": 0.065,
+      "step": 7745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3666917521603862e-06,
+      "loss": 0.0737,
+      "step": 7746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3649268370791123e-06,
+      "loss": 0.0719,
+      "step": 7747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3631629788471988e-06,
+      "loss": 0.077,
+      "step": 7748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3614001776805263e-06,
+      "loss": 0.0685,
+      "step": 7749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3596384337948433e-06,
+      "loss": 0.0745,
+      "step": 7750
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3578777474057737e-06,
+      "loss": 0.0708,
+      "step": 7751
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3561181187288008e-06,
+      "loss": 0.0714,
+      "step": 7752
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3543595479792959e-06,
+      "loss": 0.0695,
+      "step": 7753
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.352602035372481e-06,
+      "loss": 0.067,
+      "step": 7754
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3508455811234677e-06,
+      "loss": 0.0703,
+      "step": 7755
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3490901854472239e-06,
+      "loss": 0.069,
+      "step": 7756
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.347335848558593e-06,
+      "loss": 0.0659,
+      "step": 7757
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3455825706722913e-06,
+      "loss": 0.0734,
+      "step": 7758
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3438303520029006e-06,
+      "loss": 0.0644,
+      "step": 7759
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3420791927648792e-06,
+      "loss": 0.0654,
+      "step": 7760
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3403290931725443e-06,
+      "loss": 0.0694,
+      "step": 7761
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3385800534401017e-06,
+      "loss": 0.0656,
+      "step": 7762
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3368320737816088e-06,
+      "loss": 0.0687,
+      "step": 7763
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3350851544110044e-06,
+      "loss": 0.0723,
+      "step": 7764
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3333392955420942e-06,
+      "loss": 0.0677,
+      "step": 7765
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3315944973885543e-06,
+      "loss": 0.0701,
+      "step": 7766
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.329850760163931e-06,
+      "loss": 0.0667,
+      "step": 7767
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.328108084081643e-06,
+      "loss": 0.0655,
+      "step": 7768
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3263664693549739e-06,
+      "loss": 0.0777,
+      "step": 7769
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3246259161970808e-06,
+      "loss": 0.0779,
+      "step": 7770
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3228864248209915e-06,
+      "loss": 0.0596,
+      "step": 7771
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3211479954396034e-06,
+      "loss": 0.0707,
+      "step": 7772
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3194106282656827e-06,
+      "loss": 0.0696,
+      "step": 7773
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.317674323511866e-06,
+      "loss": 0.0669,
+      "step": 7774
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3159390813906636e-06,
+      "loss": 0.0677,
+      "step": 7775
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3142049021144444e-06,
+      "loss": 0.0624,
+      "step": 7776
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.312471785895465e-06,
+      "loss": 0.0686,
+      "step": 7777
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3107397329458348e-06,
+      "loss": 0.0642,
+      "step": 7778
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3090087434775433e-06,
+      "loss": 0.0653,
+      "step": 7779
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3072788177024454e-06,
+      "loss": 0.0664,
+      "step": 7780
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3055499558322682e-06,
+      "loss": 0.0681,
+      "step": 7781
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3038221580786104e-06,
+      "loss": 0.0656,
+      "step": 7782
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3020954246529304e-06,
+      "loss": 0.065,
+      "step": 7783
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3003697557665728e-06,
+      "loss": 0.0686,
+      "step": 7784
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.298645151630733e-06,
+      "loss": 0.0725,
+      "step": 7785
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2969216124564954e-06,
+      "loss": 0.0637,
+      "step": 7786
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.295199138454799e-06,
+      "loss": 0.0641,
+      "step": 7787
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2934777298364575e-06,
+      "loss": 0.0688,
+      "step": 7788
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.291757386812158e-06,
+      "loss": 0.07,
+      "step": 7789
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2900381095924508e-06,
+      "loss": 0.0688,
+      "step": 7790
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2883198983877621e-06,
+      "loss": 0.068,
+      "step": 7791
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2866027534083802e-06,
+      "loss": 0.0652,
+      "step": 7792
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2848866748644695e-06,
+      "loss": 0.0669,
+      "step": 7793
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2831716629660607e-06,
+      "loss": 0.0658,
+      "step": 7794
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2814577179230558e-06,
+      "loss": 0.0697,
+      "step": 7795
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2797448399452228e-06,
+      "loss": 0.0708,
+      "step": 7796
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.278033029242204e-06,
+      "loss": 0.0736,
+      "step": 7797
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.276322286023506e-06,
+      "loss": 0.0704,
+      "step": 7798
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2746126104985113e-06,
+      "loss": 0.0729,
+      "step": 7799
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2729040028764627e-06,
+      "loss": 0.0715,
+      "step": 7800
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2711964633664785e-06,
+      "loss": 0.0659,
+      "step": 7801
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2694899921775472e-06,
+      "loss": 0.0672,
+      "step": 7802
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2677845895185214e-06,
+      "loss": 0.0752,
+      "step": 7803
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2660802555981311e-06,
+      "loss": 0.0703,
+      "step": 7804
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2643769906249615e-06,
+      "loss": 0.067,
+      "step": 7805
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2626747948074846e-06,
+      "loss": 0.0735,
+      "step": 7806
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2609736683540253e-06,
+      "loss": 0.0634,
+      "step": 7807
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2592736114727933e-06,
+      "loss": 0.0614,
+      "step": 7808
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2575746243718501e-06,
+      "loss": 0.0778,
+      "step": 7809
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2558767072591438e-06,
+      "loss": 0.0641,
+      "step": 7810
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.254179860342476e-06,
+      "loss": 0.0706,
+      "step": 7811
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2524840838295282e-06,
+      "loss": 0.0771,
+      "step": 7812
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.250789377927848e-06,
+      "loss": 0.0725,
+      "step": 7813
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2490957428448447e-06,
+      "loss": 0.0634,
+      "step": 7814
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2474031787878105e-06,
+      "loss": 0.0718,
+      "step": 7815
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2457116859638919e-06,
+      "loss": 0.0745,
+      "step": 7816
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2440212645801197e-06,
+      "loss": 0.0702,
+      "step": 7817
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2423319148433777e-06,
+      "loss": 0.0716,
+      "step": 7818
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.240643636960428e-06,
+      "loss": 0.0757,
+      "step": 7819
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.238956431137901e-06,
+      "loss": 0.0616,
+      "step": 7820
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2372702975822925e-06,
+      "loss": 0.0708,
+      "step": 7821
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.235585236499972e-06,
+      "loss": 0.0724,
+      "step": 7822
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2339012480971712e-06,
+      "loss": 0.0693,
+      "step": 7823
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2322183325799942e-06,
+      "loss": 0.0636,
+      "step": 7824
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2305364901544149e-06,
+      "loss": 0.0639,
+      "step": 7825
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2288557210262741e-06,
+      "loss": 0.0754,
+      "step": 7826
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2271760254012821e-06,
+      "loss": 0.0726,
+      "step": 7827
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2254974034850176e-06,
+      "loss": 0.0677,
+      "step": 7828
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.223819855482926e-06,
+      "loss": 0.0723,
+      "step": 7829
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.222143381600327e-06,
+      "loss": 0.0682,
+      "step": 7830
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.220467982042397e-06,
+      "loss": 0.0681,
+      "step": 7831
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2187936570141978e-06,
+      "loss": 0.0691,
+      "step": 7832
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2171204067206443e-06,
+      "loss": 0.07,
+      "step": 7833
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2154482313665283e-06,
+      "loss": 0.0672,
+      "step": 7834
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2137771311565106e-06,
+      "loss": 0.0686,
+      "step": 7835
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2121071062951096e-06,
+      "loss": 0.0666,
+      "step": 7836
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2104381569867297e-06,
+      "loss": 0.0791,
+      "step": 7837
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2087702834356264e-06,
+      "loss": 0.0678,
+      "step": 7838
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2071034858459374e-06,
+      "loss": 0.0625,
+      "step": 7839
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2054377644216576e-06,
+      "loss": 0.0684,
+      "step": 7840
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2037731193666603e-06,
+      "loss": 0.0631,
+      "step": 7841
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.202109550884677e-06,
+      "loss": 0.0661,
+      "step": 7842
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2004470591793149e-06,
+      "loss": 0.0707,
+      "step": 7843
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1987856444540491e-06,
+      "loss": 0.0658,
+      "step": 7844
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1971253069122124e-06,
+      "loss": 0.0697,
+      "step": 7845
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1954660467570256e-06,
+      "loss": 0.068,
+      "step": 7846
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1938078641915574e-06,
+      "loss": 0.0786,
+      "step": 7847
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1921507594187553e-06,
+      "loss": 0.0702,
+      "step": 7848
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1904947326414352e-06,
+      "loss": 0.0682,
+      "step": 7849
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1888397840622756e-06,
+      "loss": 0.072,
+      "step": 7850
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1871859138838294e-06,
+      "loss": 0.0754,
+      "step": 7851
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.185533122308511e-06,
+      "loss": 0.0692,
+      "step": 7852
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1838814095386086e-06,
+      "loss": 0.069,
+      "step": 7853
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1822307757762773e-06,
+      "loss": 0.0737,
+      "step": 7854
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1805812212235347e-06,
+      "loss": 0.0747,
+      "step": 7855
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1789327460822709e-06,
+      "loss": 0.0763,
+      "step": 7856
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1772853505542447e-06,
+      "loss": 0.0705,
+      "step": 7857
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1756390348410819e-06,
+      "loss": 0.073,
+      "step": 7858
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1739937991442763e-06,
+      "loss": 0.0695,
+      "step": 7859
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1723496436651838e-06,
+      "loss": 0.0682,
+      "step": 7860
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.17070656860504e-06,
+      "loss": 0.0756,
+      "step": 7861
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1690645741649342e-06,
+      "loss": 0.0696,
+      "step": 7862
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1674236605458379e-06,
+      "loss": 0.0708,
+      "step": 7863
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1657838279485789e-06,
+      "loss": 0.0706,
+      "step": 7864
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.164145076573856e-06,
+      "loss": 0.0657,
+      "step": 7865
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1625074066222385e-06,
+      "loss": 0.0668,
+      "step": 7866
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1608708182941618e-06,
+      "loss": 0.0675,
+      "step": 7867
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.159235311789929e-06,
+      "loss": 0.0617,
+      "step": 7868
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1576008873097043e-06,
+      "loss": 0.0689,
+      "step": 7869
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1559675450535345e-06,
+      "loss": 0.0701,
+      "step": 7870
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1543352852213186e-06,
+      "loss": 0.0684,
+      "step": 7871
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1527041080128299e-06,
+      "loss": 0.0702,
+      "step": 7872
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151074013627711e-06,
+      "loss": 0.0675,
+      "step": 7873
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149445002265468e-06,
+      "loss": 0.0811,
+      "step": 7874
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1478170741254769e-06,
+      "loss": 0.0723,
+      "step": 7875
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1461902294069804e-06,
+      "loss": 0.0653,
+      "step": 7876
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1445644683090907e-06,
+      "loss": 0.0725,
+      "step": 7877
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1429397910307794e-06,
+      "loss": 0.0696,
+      "step": 7878
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1413161977708952e-06,
+      "loss": 0.0713,
+      "step": 7879
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1396936887281506e-06,
+      "loss": 0.069,
+      "step": 7880
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1380722641011232e-06,
+      "loss": 0.0676,
+      "step": 7881
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.136451924088261e-06,
+      "loss": 0.0656,
+      "step": 7882
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1348326688878776e-06,
+      "loss": 0.0662,
+      "step": 7883
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.133214498698154e-06,
+      "loss": 0.0632,
+      "step": 7884
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1315974137171404e-06,
+      "loss": 0.0698,
+      "step": 7885
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1299814141427489e-06,
+      "loss": 0.0663,
+      "step": 7886
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1283665001727628e-06,
+      "loss": 0.0698,
+      "step": 7887
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1267526720048338e-06,
+      "loss": 0.0662,
+      "step": 7888
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1251399298364784e-06,
+      "loss": 0.0644,
+      "step": 7889
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.123528273865082e-06,
+      "loss": 0.0722,
+      "step": 7890
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1219177042878905e-06,
+      "loss": 0.0738,
+      "step": 7891
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1203082213020288e-06,
+      "loss": 0.0627,
+      "step": 7892
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1186998251044744e-06,
+      "loss": 0.0666,
+      "step": 7893
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1170925158920887e-06,
+      "loss": 0.0705,
+      "step": 7894
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.115486293861583e-06,
+      "loss": 0.0609,
+      "step": 7895
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1138811592095466e-06,
+      "loss": 0.0635,
+      "step": 7896
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1122771121324316e-06,
+      "loss": 0.0675,
+      "step": 7897
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1106741528265585e-06,
+      "loss": 0.0696,
+      "step": 7898
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1090722814881161e-06,
+      "loss": 0.0683,
+      "step": 7899
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1074714983131518e-06,
+      "loss": 0.0724,
+      "step": 7900
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1058718034975924e-06,
+      "loss": 0.0676,
+      "step": 7901
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1042731972372222e-06,
+      "loss": 0.0725,
+      "step": 7902
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1026756797276939e-06,
+      "loss": 0.0703,
+      "step": 7903
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1010792511645308e-06,
+      "loss": 0.0694,
+      "step": 7904
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0994839117431177e-06,
+      "loss": 0.0649,
+      "step": 7905
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0978896616587108e-06,
+      "loss": 0.0715,
+      "step": 7906
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0962965011064309e-06,
+      "loss": 0.0703,
+      "step": 7907
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.094704430281266e-06,
+      "loss": 0.0701,
+      "step": 7908
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0931134493780672e-06,
+      "loss": 0.0708,
+      "step": 7909
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0915235585915563e-06,
+      "loss": 0.0735,
+      "step": 7910
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.0667,
+      "step": 7911
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0883470481468162e-06,
+      "loss": 0.0675,
+      "step": 7912
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0867604288773615e-06,
+      "loss": 0.0726,
+      "step": 7913
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.085174900502144e-06,
+      "loss": 0.0661,
+      "step": 7914
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0835904632152138e-06,
+      "loss": 0.0718,
+      "step": 7915
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0820071172104962e-06,
+      "loss": 0.0645,
+      "step": 7916
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0804248626817737e-06,
+      "loss": 0.0647,
+      "step": 7917
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0788436998226991e-06,
+      "loss": 0.0688,
+      "step": 7918
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0772636288267923e-06,
+      "loss": 0.0613,
+      "step": 7919
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0756846498874386e-06,
+      "loss": 0.0721,
+      "step": 7920
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0741067631978918e-06,
+      "loss": 0.0724,
+      "step": 7921
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0725299689512635e-06,
+      "loss": 0.0692,
+      "step": 7922
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.070954267340547e-06,
+      "loss": 0.0669,
+      "step": 7923
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0693796585585837e-06,
+      "loss": 0.063,
+      "step": 7924
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.067806142798099e-06,
+      "loss": 0.0716,
+      "step": 7925
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0662337202516704e-06,
+      "loss": 0.0714,
+      "step": 7926
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0646623911117493e-06,
+      "loss": 0.0605,
+      "step": 7927
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0630921555706498e-06,
+      "loss": 0.0662,
+      "step": 7928
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0615230138205546e-06,
+      "loss": 0.0751,
+      "step": 7929
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0599549660535146e-06,
+      "loss": 0.0658,
+      "step": 7930
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0583880124614354e-06,
+      "loss": 0.0737,
+      "step": 7931
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0568221532361055e-06,
+      "loss": 0.0645,
+      "step": 7932
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0552573885691664e-06,
+      "loss": 0.0729,
+      "step": 7933
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0536937186521311e-06,
+      "loss": 0.0661,
+      "step": 7934
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.052131143676377e-06,
+      "loss": 0.0715,
+      "step": 7935
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0505696638331487e-06,
+      "loss": 0.0674,
+      "step": 7936
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0490092793135575e-06,
+      "loss": 0.0688,
+      "step": 7937
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.047449990308579e-06,
+      "loss": 0.0755,
+      "step": 7938
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.045891797009053e-06,
+      "loss": 0.0681,
+      "step": 7939
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.044334699605688e-06,
+      "loss": 0.0688,
+      "step": 7940
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0427786982890587e-06,
+      "loss": 0.068,
+      "step": 7941
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0412237932496038e-06,
+      "loss": 0.0714,
+      "step": 7942
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0396699846776293e-06,
+      "loss": 0.0684,
+      "step": 7943
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.038117272763306e-06,
+      "loss": 0.0733,
+      "step": 7944
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.036565657696672e-06,
+      "loss": 0.0727,
+      "step": 7945
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0350151396676257e-06,
+      "loss": 0.0673,
+      "step": 7946
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0334657188659414e-06,
+      "loss": 0.0681,
+      "step": 7947
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0319173954812489e-06,
+      "loss": 0.0695,
+      "step": 7948
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0303701697030499e-06,
+      "loss": 0.0698,
+      "step": 7949
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0288240417207084e-06,
+      "loss": 0.065,
+      "step": 7950
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0272790117234577e-06,
+      "loss": 0.0614,
+      "step": 7951
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.025735079900394e-06,
+      "loss": 0.0659,
+      "step": 7952
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0241922464404763e-06,
+      "loss": 0.0692,
+      "step": 7953
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.022650511532538e-06,
+      "loss": 0.0665,
+      "step": 7954
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.021109875365266e-06,
+      "loss": 0.0639,
+      "step": 7955
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.019570338127227e-06,
+      "loss": 0.0692,
+      "step": 7956
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0180319000068395e-06,
+      "loss": 0.0691,
+      "step": 7957
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0164945611923948e-06,
+      "loss": 0.0637,
+      "step": 7958
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0149583218720483e-06,
+      "loss": 0.0751,
+      "step": 7959
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0134231822338225e-06,
+      "loss": 0.0707,
+      "step": 7960
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.011889142465603e-06,
+      "loss": 0.0723,
+      "step": 7961
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0103562027551373e-06,
+      "loss": 0.0652,
+      "step": 7962
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0088243632900507e-06,
+      "loss": 0.0676,
+      "step": 7963
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.007293624257818e-06,
+      "loss": 0.0747,
+      "step": 7964
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.005763985845789e-06,
+      "loss": 0.0662,
+      "step": 7965
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0042354482411788e-06,
+      "loss": 0.072,
+      "step": 7966
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0027080116310627e-06,
+      "loss": 0.0735,
+      "step": 7967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0011816762023852e-06,
+      "loss": 0.0661,
+      "step": 7968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.996564421419575e-07,
+      "loss": 0.0734,
+      "step": 7969
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.981323096364504e-07,
+      "loss": 0.0678,
+      "step": 7970
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.966092788724025e-07,
+      "loss": 0.0656,
+      "step": 7971
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95087350036219e-07,
+      "loss": 0.0752,
+      "step": 7972
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.9356652331417e-07,
+      "loss": 0.0648,
+      "step": 7973
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.920467988923898e-07,
+      "loss": 0.0665,
+      "step": 7974
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.905281769568775e-07,
+      "loss": 0.0698,
+      "step": 7975
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.890106576935e-07,
+      "loss": 0.0686,
+      "step": 7976
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.874942412879807e-07,
+      "loss": 0.0681,
+      "step": 7977
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.859789279259225e-07,
+      "loss": 0.0647,
+      "step": 7978
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.84464717792778e-07,
+      "loss": 0.0704,
+      "step": 7979
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.829516110738779e-07,
+      "loss": 0.0738,
+      "step": 7980
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.814396079544065e-07,
+      "loss": 0.0674,
+      "step": 7981
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.799287086194209e-07,
+      "loss": 0.0659,
+      "step": 7982
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.784189132538424e-07,
+      "loss": 0.0723,
+      "step": 7983
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.769102220424498e-07,
+      "loss": 0.0644,
+      "step": 7984
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.754026351698987e-07,
+      "loss": 0.0726,
+      "step": 7985
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.73896152820697e-07,
+      "loss": 0.0689,
+      "step": 7986
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.723907751792306e-07,
+      "loss": 0.0747,
+      "step": 7987
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.708865024297365e-07,
+      "loss": 0.0699,
+      "step": 7988
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.693833347563264e-07,
+      "loss": 0.0765,
+      "step": 7989
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.67881272342973e-07,
+      "loss": 0.0711,
+      "step": 7990
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.66380315373515e-07,
+      "loss": 0.0709,
+      "step": 7991
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.648804640316533e-07,
+      "loss": 0.0705,
+      "step": 7992
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.633817185009587e-07,
+      "loss": 0.0664,
+      "step": 7993
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.618840789648577e-07,
+      "loss": 0.0688,
+      "step": 7994
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.603875456066503e-07,
+      "loss": 0.0609,
+      "step": 7995
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.588921186094968e-07,
+      "loss": 0.0628,
+      "step": 7996
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.57397798156423e-07,
+      "loss": 0.0671,
+      "step": 7997
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.559045844303193e-07,
+      "loss": 0.0716,
+      "step": 7998
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54412477613942e-07,
+      "loss": 0.0675,
+      "step": 7999
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.529214778899109e-07,
+      "loss": 0.0689,
+      "step": 8000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.514315854407042e-07,
+      "loss": 0.0715,
+      "step": 8001
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.499428004486788e-07,
+      "loss": 0.0682,
+      "step": 8002
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.484551230960404e-07,
+      "loss": 0.0656,
+      "step": 8003
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.469685535648698e-07,
+      "loss": 0.0734,
+      "step": 8004
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.454830920371105e-07,
+      "loss": 0.0711,
+      "step": 8005
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.439987386945615e-07,
+      "loss": 0.0686,
+      "step": 8006
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.42515493718903e-07,
+      "loss": 0.0642,
+      "step": 8007
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.410333572916597e-07,
+      "loss": 0.0673,
+      "step": 8008
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.395523295942399e-07,
+      "loss": 0.0665,
+      "step": 8009
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.380724108078997e-07,
+      "loss": 0.0687,
+      "step": 8010
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.365936011137743e-07,
+      "loss": 0.0686,
+      "step": 8011
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.351159006928489e-07,
+      "loss": 0.0716,
+      "step": 8012
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.336393097259833e-07,
+      "loss": 0.0717,
+      "step": 8013
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.321638283938961e-07,
+      "loss": 0.0728,
+      "step": 8014
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.306894568771729e-07,
+      "loss": 0.068,
+      "step": 8015
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.29216195356265e-07,
+      "loss": 0.0736,
+      "step": 8016
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.277440440114804e-07,
+      "loss": 0.0669,
+      "step": 8017
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.262730030229983e-07,
+      "loss": 0.067,
+      "step": 8018
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.248030725708612e-07,
+      "loss": 0.0707,
+      "step": 8019
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.23334252834972e-07,
+      "loss": 0.0707,
+      "step": 8020
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.218665439951024e-07,
+      "loss": 0.0661,
+      "step": 8021
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.203999462308844e-07,
+      "loss": 0.068,
+      "step": 8022
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.189344597218153e-07,
+      "loss": 0.0704,
+      "step": 8023
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.174700846472584e-07,
+      "loss": 0.0683,
+      "step": 8024
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.160068211864359e-07,
+      "loss": 0.073,
+      "step": 8025
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.14544669518439e-07,
+      "loss": 0.069,
+      "step": 8026
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.130836298222201e-07,
+      "loss": 0.0698,
+      "step": 8027
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.116237022765972e-07,
+      "loss": 0.0666,
+      "step": 8028
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.101648870602498e-07,
+      "loss": 0.0725,
+      "step": 8029
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.087071843517248e-07,
+      "loss": 0.0647,
+      "step": 8030
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.072505943294319e-07,
+      "loss": 0.0679,
+      "step": 8031
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.057951171716373e-07,
+      "loss": 0.0703,
+      "step": 8032
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.043407530564851e-07,
+      "loss": 0.0666,
+      "step": 8033
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.02887502161971e-07,
+      "loss": 0.067,
+      "step": 8034
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.01435364665959e-07,
+      "loss": 0.0644,
+      "step": 8035
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.999843407461773e-07,
+      "loss": 0.0653,
+      "step": 8036
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.985344305802179e-07,
+      "loss": 0.0632,
+      "step": 8037
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.97085634345537e-07,
+      "loss": 0.0685,
+      "step": 8038
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.956379522194469e-07,
+      "loss": 0.0663,
+      "step": 8039
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.941913843791394e-07,
+      "loss": 0.0625,
+      "step": 8040
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.927459310016507e-07,
+      "loss": 0.0726,
+      "step": 8041
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.91301592263899e-07,
+      "loss": 0.0705,
+      "step": 8042
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.89858368342652e-07,
+      "loss": 0.0667,
+      "step": 8043
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.884162594145462e-07,
+      "loss": 0.0633,
+      "step": 8044
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.869752656560848e-07,
+      "loss": 0.0688,
+      "step": 8045
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.8553538724363e-07,
+      "loss": 0.0645,
+      "step": 8046
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.8409662435341e-07,
+      "loss": 0.0623,
+      "step": 8047
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.826589771615124e-07,
+      "loss": 0.0731,
+      "step": 8048
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.812224458438945e-07,
+      "loss": 0.0735,
+      "step": 8049
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.797870305763712e-07,
+      "loss": 0.0656,
+      "step": 8050
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.783527315346252e-07,
+      "loss": 0.0704,
+      "step": 8051
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.769195488942018e-07,
+      "loss": 0.0646,
+      "step": 8052
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.75487482830506e-07,
+      "loss": 0.0703,
+      "step": 8053
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.7405653351881e-07,
+      "loss": 0.0649,
+      "step": 8054
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.726267011342504e-07,
+      "loss": 0.067,
+      "step": 8055
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.711979858518216e-07,
+      "loss": 0.0661,
+      "step": 8056
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.697703878463848e-07,
+      "loss": 0.069,
+      "step": 8057
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.683439072926658e-07,
+      "loss": 0.0741,
+      "step": 8058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.669185443652517e-07,
+      "loss": 0.0683,
+      "step": 8059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.654942992385929e-07,
+      "loss": 0.0679,
+      "step": 8060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.64071172087001e-07,
+      "loss": 0.0715,
+      "step": 8061
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.626491630846578e-07,
+      "loss": 0.0753,
+      "step": 8062
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.612282724055965e-07,
+      "loss": 0.0713,
+      "step": 8063
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.598085002237289e-07,
+      "loss": 0.0779,
+      "step": 8064
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.583898467128149e-07,
+      "loss": 0.0717,
+      "step": 8065
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.569723120464856e-07,
+      "loss": 0.0681,
+      "step": 8066
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.555558963982347e-07,
+      "loss": 0.0701,
+      "step": 8067
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.541405999414166e-07,
+      "loss": 0.0718,
+      "step": 8068
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.527264228492526e-07,
+      "loss": 0.0653,
+      "step": 8069
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.51313365294818e-07,
+      "loss": 0.064,
+      "step": 8070
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.499014274510653e-07,
+      "loss": 0.0723,
+      "step": 8071
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.484906094907952e-07,
+      "loss": 0.0664,
+      "step": 8072
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.470809115866818e-07,
+      "loss": 0.0707,
+      "step": 8073
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.456723339112571e-07,
+      "loss": 0.0673,
+      "step": 8074
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.442648766369188e-07,
+      "loss": 0.0708,
+      "step": 8075
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.428585399359246e-07,
+      "loss": 0.0664,
+      "step": 8076
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.41453323980398e-07,
+      "loss": 0.0604,
+      "step": 8077
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.400492289423235e-07,
+      "loss": 0.0685,
+      "step": 8078
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.386462549935471e-07,
+      "loss": 0.0702,
+      "step": 8079
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.372444023057802e-07,
+      "loss": 0.0747,
+      "step": 8080
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.358436710505968e-07,
+      "loss": 0.0618,
+      "step": 8081
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.344440613994309e-07,
+      "loss": 0.0649,
+      "step": 8082
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.330455735235832e-07,
+      "loss": 0.065,
+      "step": 8083
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.316482075942156e-07,
+      "loss": 0.0771,
+      "step": 8084
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.30251963782348e-07,
+      "loss": 0.0724,
+      "step": 8085
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.288568422588727e-07,
+      "loss": 0.0668,
+      "step": 8086
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.274628431945353e-07,
+      "loss": 0.0677,
+      "step": 8087
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.260699667599481e-07,
+      "loss": 0.0691,
+      "step": 8088
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.24678213125587e-07,
+      "loss": 0.0649,
+      "step": 8089
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.232875824617892e-07,
+      "loss": 0.0629,
+      "step": 8090
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.218980749387551e-07,
+      "loss": 0.0685,
+      "step": 8091
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.20509690726542e-07,
+      "loss": 0.0657,
+      "step": 8092
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.191224299950818e-07,
+      "loss": 0.066,
+      "step": 8093
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.177362929141553e-07,
+      "loss": 0.0694,
+      "step": 8094
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.163512796534178e-07,
+      "loss": 0.0603,
+      "step": 8095
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.149673903823785e-07,
+      "loss": 0.0702,
+      "step": 8096
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.135846252704116e-07,
+      "loss": 0.0714,
+      "step": 8097
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.122029844867562e-07,
+      "loss": 0.0661,
+      "step": 8098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.108224682005106e-07,
+      "loss": 0.0622,
+      "step": 8099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.094430765806382e-07,
+      "loss": 0.0679,
+      "step": 8100
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.080648097959576e-07,
+      "loss": 0.0712,
+      "step": 8101
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.066876680151637e-07,
+      "loss": 0.0681,
+      "step": 8102
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.053116514067994e-07,
+      "loss": 0.0706,
+      "step": 8103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.039367601392778e-07,
+      "loss": 0.0642,
+      "step": 8104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.025629943808711e-07,
+      "loss": 0.0658,
+      "step": 8105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.011903542997157e-07,
+      "loss": 0.0673,
+      "step": 8106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.998188400638107e-07,
+      "loss": 0.0729,
+      "step": 8107
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.984484518410141e-07,
+      "loss": 0.0678,
+      "step": 8108
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.970791897990516e-07,
+      "loss": 0.0701,
+      "step": 8109
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.957110541055024e-07,
+      "loss": 0.0645,
+      "step": 8110
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.943440449278161e-07,
+      "loss": 0.0646,
+      "step": 8111
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.92978162433301e-07,
+      "loss": 0.0632,
+      "step": 8112
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.916134067891279e-07,
+      "loss": 0.0684,
+      "step": 8113
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.902497781623297e-07,
+      "loss": 0.0678,
+      "step": 8114
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.888872767198041e-07,
+      "loss": 0.075,
+      "step": 8115
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.875259026283011e-07,
+      "loss": 0.0637,
+      "step": 8116
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.861656560544473e-07,
+      "loss": 0.067,
+      "step": 8117
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.848065371647162e-07,
+      "loss": 0.0722,
+      "step": 8118
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.834485461254592e-07,
+      "loss": 0.0663,
+      "step": 8119
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.820916831028758e-07,
+      "loss": 0.0743,
+      "step": 8120
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.807359482630339e-07,
+      "loss": 0.0693,
+      "step": 8121
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.793813417718644e-07,
+      "loss": 0.0748,
+      "step": 8122
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.780278637951521e-07,
+      "loss": 0.0686,
+      "step": 8123
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.766755144985572e-07,
+      "loss": 0.0757,
+      "step": 8124
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.753242940475869e-07,
+      "loss": 0.0649,
+      "step": 8125
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.739742026076247e-07,
+      "loss": 0.0711,
+      "step": 8126
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.726252403439038e-07,
+      "loss": 0.0676,
+      "step": 8127
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.712774074215257e-07,
+      "loss": 0.0721,
+      "step": 8128
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.699307040054527e-07,
+      "loss": 0.0667,
+      "step": 8129
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.685851302605062e-07,
+      "loss": 0.0718,
+      "step": 8130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.672406863513748e-07,
+      "loss": 0.065,
+      "step": 8131
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.658973724426011e-07,
+      "loss": 0.0703,
+      "step": 8132
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.64555188698598e-07,
+      "loss": 0.0611,
+      "step": 8133
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.632141352836319e-07,
+      "loss": 0.0739,
+      "step": 8134
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.618742123618383e-07,
+      "loss": 0.069,
+      "step": 8135
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.60535420097207e-07,
+      "loss": 0.0689,
+      "step": 8136
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.59197758653597e-07,
+      "loss": 0.0648,
+      "step": 8137
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.578612281947229e-07,
+      "loss": 0.0694,
+      "step": 8138
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.565258288841648e-07,
+      "loss": 0.0698,
+      "step": 8139
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.551915608853588e-07,
+      "loss": 0.0725,
+      "step": 8140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.538584243616121e-07,
+      "loss": 0.0686,
+      "step": 8141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.525264194760828e-07,
+      "loss": 0.0688,
+      "step": 8142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.511955463917975e-07,
+      "loss": 0.0712,
+      "step": 8143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.498658052716424e-07,
+      "loss": 0.0689,
+      "step": 8144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.485371962783639e-07,
+      "loss": 0.0678,
+      "step": 8145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.472097195745731e-07,
+      "loss": 0.0678,
+      "step": 8146
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.458833753227357e-07,
+      "loss": 0.0665,
+      "step": 8147
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.445581636851895e-07,
+      "loss": 0.0699,
+      "step": 8148
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.432340848241216e-07,
+      "loss": 0.073,
+      "step": 8149
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.41911138901592e-07,
+      "loss": 0.07,
+      "step": 8150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.405893260795128e-07,
+      "loss": 0.0693,
+      "step": 8151
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.392686465196608e-07,
+      "loss": 0.0714,
+      "step": 8152
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.379491003836781e-07,
+      "loss": 0.0682,
+      "step": 8153
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.366306878330587e-07,
+      "loss": 0.0674,
+      "step": 8154
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.353134090291702e-07,
+      "loss": 0.0654,
+      "step": 8155
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.339972641332272e-07,
+      "loss": 0.0706,
+      "step": 8156
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.326822533063216e-07,
+      "loss": 0.0648,
+      "step": 8157
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.313683767093915e-07,
+      "loss": 0.0664,
+      "step": 8158
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.300556345032439e-07,
+      "loss": 0.0652,
+      "step": 8159
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.287440268485479e-07,
+      "loss": 0.0688,
+      "step": 8160
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.274335539058297e-07,
+      "loss": 0.0675,
+      "step": 8161
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.261242158354787e-07,
+      "loss": 0.0728,
+      "step": 8162
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.248160127977477e-07,
+      "loss": 0.0723,
+      "step": 8163
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.235089449527443e-07,
+      "loss": 0.0721,
+      "step": 8164
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.222030124604428e-07,
+      "loss": 0.0721,
+      "step": 8165
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.208982154806765e-07,
+      "loss": 0.0672,
+      "step": 8166
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.195945541731397e-07,
+      "loss": 0.0667,
+      "step": 8167
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.182920286973882e-07,
+      "loss": 0.0749,
+      "step": 8168
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.16990639212839e-07,
+      "loss": 0.0743,
+      "step": 8169
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.156903858787711e-07,
+      "loss": 0.0686,
+      "step": 8170
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.143912688543165e-07,
+      "loss": 0.0634,
+      "step": 8171
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.13093288298482e-07,
+      "loss": 0.0651,
+      "step": 8172
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.117964443701242e-07,
+      "loss": 0.063,
+      "step": 8173
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.105007372279649e-07,
+      "loss": 0.0677,
+      "step": 8174
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.092061670305849e-07,
+      "loss": 0.0684,
+      "step": 8175
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.079127339364289e-07,
+      "loss": 0.0663,
+      "step": 8176
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.066204381038011e-07,
+      "loss": 0.0664,
+      "step": 8177
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.053292796908629e-07,
+      "loss": 0.0642,
+      "step": 8178
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.040392588556433e-07,
+      "loss": 0.064,
+      "step": 8179
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.027503757560239e-07,
+      "loss": 0.0732,
+      "step": 8180
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.014626305497574e-07,
+      "loss": 0.0698,
+      "step": 8181
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.001760233944455e-07,
+      "loss": 0.0647,
+      "step": 8182
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.9889055444756e-07,
+      "loss": 0.0609,
+      "step": 8183
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.976062238664294e-07,
+      "loss": 0.0664,
+      "step": 8184
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963230318082415e-07,
+      "loss": 0.0695,
+      "step": 8185
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.950409784300505e-07,
+      "loss": 0.0699,
+      "step": 8186
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.93760063888761e-07,
+      "loss": 0.063,
+      "step": 8187
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924802883411518e-07,
+      "loss": 0.064,
+      "step": 8188
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.912016519438503e-07,
+      "loss": 0.0675,
+      "step": 8189
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.899241548533497e-07,
+      "loss": 0.0664,
+      "step": 8190
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.886477972260042e-07,
+      "loss": 0.0685,
+      "step": 8191
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.873725792180285e-07,
+      "loss": 0.065,
+      "step": 8192
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.860985009854948e-07,
+      "loss": 0.0693,
+      "step": 8193
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.848255626843403e-07,
+      "loss": 0.0651,
+      "step": 8194
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.835537644703582e-07,
+      "loss": 0.0696,
+      "step": 8195
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.822831064992053e-07,
+      "loss": 0.0642,
+      "step": 8196
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.810135889263969e-07,
+      "loss": 0.0671,
+      "step": 8197
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.797452119073111e-07,
+      "loss": 0.0626,
+      "step": 8198
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.784779755971837e-07,
+      "loss": 0.0692,
+      "step": 8199
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.77211880151114e-07,
+      "loss": 0.0764,
+      "step": 8200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.759469257240591e-07,
+      "loss": 0.0742,
+      "step": 8201
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.74683112470833e-07,
+      "loss": 0.0638,
+      "step": 8202
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.73420440546122e-07,
+      "loss": 0.0639,
+      "step": 8203
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.72158910104459e-07,
+      "loss": 0.0675,
+      "step": 8204
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.708985213002451e-07,
+      "loss": 0.0678,
+      "step": 8205
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.696392742877389e-07,
+      "loss": 0.0705,
+      "step": 8206
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.683811692210617e-07,
+      "loss": 0.0654,
+      "step": 8207
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.671242062541928e-07,
+      "loss": 0.069,
+      "step": 8208
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.658683855409686e-07,
+      "loss": 0.0696,
+      "step": 8209
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.646137072350967e-07,
+      "loss": 0.0674,
+      "step": 8210
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.633601714901294e-07,
+      "loss": 0.0687,
+      "step": 8211
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.621077784594954e-07,
+      "loss": 0.0699,
+      "step": 8212
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.608565282964696e-07,
+      "loss": 0.0688,
+      "step": 8213
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.596064211541942e-07,
+      "loss": 0.0669,
+      "step": 8214
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.583574571856721e-07,
+      "loss": 0.0641,
+      "step": 8215
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.571096365437624e-07,
+      "loss": 0.0677,
+      "step": 8216
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.558629593811883e-07,
+      "loss": 0.0657,
+      "step": 8217
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.546174258505289e-07,
+      "loss": 0.0658,
+      "step": 8218
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.533730361042257e-07,
+      "loss": 0.0731,
+      "step": 8219
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.5212979029458e-07,
+      "loss": 0.0668,
+      "step": 8220
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.508876885737547e-07,
+      "loss": 0.0699,
+      "step": 8221
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.496467310937693e-07,
+      "loss": 0.0645,
+      "step": 8222
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.484069180065055e-07,
+      "loss": 0.0598,
+      "step": 8223
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.471682494637044e-07,
+      "loss": 0.0631,
+      "step": 8224
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.45930725616969e-07,
+      "loss": 0.078,
+      "step": 8225
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.446943466177569e-07,
+      "loss": 0.0659,
+      "step": 8226
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.434591126173905e-07,
+      "loss": 0.0657,
+      "step": 8227
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.422250237670502e-07,
+      "loss": 0.0675,
+      "step": 8228
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.409920802177772e-07,
+      "loss": 0.0756,
+      "step": 8229
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.397602821204718e-07,
+      "loss": 0.0705,
+      "step": 8230
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.385296296258914e-07,
+      "loss": 0.063,
+      "step": 8231
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.373001228846609e-07,
+      "loss": 0.0642,
+      "step": 8232
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.360717620472545e-07,
+      "loss": 0.0744,
+      "step": 8233
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.348445472640185e-07,
+      "loss": 0.0662,
+      "step": 8234
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.336184786851451e-07,
+      "loss": 0.0676,
+      "step": 8235
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.323935564606975e-07,
+      "loss": 0.0687,
+      "step": 8236
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.311697807405925e-07,
+      "loss": 0.0633,
+      "step": 8237
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.299471516746092e-07,
+      "loss": 0.0636,
+      "step": 8238
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.287256694123878e-07,
+      "loss": 0.0719,
+      "step": 8239
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.275053341034187e-07,
+      "loss": 0.068,
+      "step": 8240
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.262861458970681e-07,
+      "loss": 0.062,
+      "step": 8241
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.250681049425477e-07,
+      "loss": 0.0623,
+      "step": 8242
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.238512113889339e-07,
+      "loss": 0.0655,
+      "step": 8243
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.226354653851641e-07,
+      "loss": 0.07,
+      "step": 8244
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.214208670800325e-07,
+      "loss": 0.068,
+      "step": 8245
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.202074166221961e-07,
+      "loss": 0.0629,
+      "step": 8246
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.18995114160168e-07,
+      "loss": 0.0668,
+      "step": 8247
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.177839598423252e-07,
+      "loss": 0.071,
+      "step": 8248
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.165739538168957e-07,
+      "loss": 0.0654,
+      "step": 8249
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.153650962319768e-07,
+      "loss": 0.0727,
+      "step": 8250
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.1415738723552e-07,
+      "loss": 0.0689,
+      "step": 8251
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.129508269753359e-07,
+      "loss": 0.0645,
+      "step": 8252
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.117454155990976e-07,
+      "loss": 0.0662,
+      "step": 8253
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.105411532543348e-07,
+      "loss": 0.0701,
+      "step": 8254
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.093380400884386e-07,
+      "loss": 0.0743,
+      "step": 8255
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.081360762486587e-07,
+      "loss": 0.0768,
+      "step": 8256
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.069352618821011e-07,
+      "loss": 0.0646,
+      "step": 8257
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.057355971357359e-07,
+      "loss": 0.0687,
+      "step": 8258
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.04537082156389e-07,
+      "loss": 0.0685,
+      "step": 8259
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.033397170907496e-07,
+      "loss": 0.0698,
+      "step": 8260
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.021435020853628e-07,
+      "loss": 0.0666,
+      "step": 8261
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.009484372866303e-07,
+      "loss": 0.07,
+      "step": 8262
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.997545228408219e-07,
+      "loss": 0.0684,
+      "step": 8263
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.985617588940551e-07,
+      "loss": 0.0747,
+      "step": 8264
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.973701455923198e-07,
+      "loss": 0.0658,
+      "step": 8265
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.961796830814514e-07,
+      "loss": 0.0668,
+      "step": 8266
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.949903715071559e-07,
+      "loss": 0.0726,
+      "step": 8267
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.9380221101499e-07,
+      "loss": 0.0604,
+      "step": 8268
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.92615201750375e-07,
+      "loss": 0.0672,
+      "step": 8269
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.914293438585894e-07,
+      "loss": 0.0681,
+      "step": 8270
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.902446374847681e-07,
+      "loss": 0.0691,
+      "step": 8271
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.890610827739118e-07,
+      "loss": 0.0679,
+      "step": 8272
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.878786798708735e-07,
+      "loss": 0.0657,
+      "step": 8273
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.866974289203675e-07,
+      "loss": 0.064,
+      "step": 8274
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.855173300669692e-07,
+      "loss": 0.0688,
+      "step": 8275
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.843383834551097e-07,
+      "loss": 0.0651,
+      "step": 8276
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.831605892290815e-07,
+      "loss": 0.0694,
+      "step": 8277
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.819839475330358e-07,
+      "loss": 0.0735,
+      "step": 8278
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.808084585109819e-07,
+      "loss": 0.0641,
+      "step": 8279
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.79634122306787e-07,
+      "loss": 0.0679,
+      "step": 8280
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.784609390641793e-07,
+      "loss": 0.0637,
+      "step": 8281
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.772889089267453e-07,
+      "loss": 0.0715,
+      "step": 8282
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761180320379289e-07,
+      "loss": 0.0685,
+      "step": 8283
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.749483085410356e-07,
+      "loss": 0.0669,
+      "step": 8284
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.737797385792298e-07,
+      "loss": 0.0672,
+      "step": 8285
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.726123222955283e-07,
+      "loss": 0.0739,
+      "step": 8286
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.714460598328164e-07,
+      "loss": 0.0666,
+      "step": 8287
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.702809513338292e-07,
+      "loss": 0.0635,
+      "step": 8288
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.691169969411691e-07,
+      "loss": 0.0695,
+      "step": 8289
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.67954196797289e-07,
+      "loss": 0.0699,
+      "step": 8290
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.667925510445061e-07,
+      "loss": 0.0633,
+      "step": 8291
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.656320598249965e-07,
+      "loss": 0.0687,
+      "step": 8292
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.644727232807878e-07,
+      "loss": 0.0708,
+      "step": 8293
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.633145415537777e-07,
+      "loss": 0.066,
+      "step": 8294
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.621575147857095e-07,
+      "loss": 0.0695,
+      "step": 8295
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.610016431182008e-07,
+      "loss": 0.0656,
+      "step": 8296
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.59846926692712e-07,
+      "loss": 0.0754,
+      "step": 8297
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.586933656505711e-07,
+      "loss": 0.0686,
+      "step": 8298
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.57540960132964e-07,
+      "loss": 0.078,
+      "step": 8299
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.563897102809334e-07,
+      "loss": 0.0658,
+      "step": 8300
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.552396162353813e-07,
+      "loss": 0.0643,
+      "step": 8301
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.54090678137067e-07,
+      "loss": 0.0723,
+      "step": 8302
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.529428961266114e-07,
+      "loss": 0.0714,
+      "step": 8303
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.517962703444901e-07,
+      "loss": 0.0634,
+      "step": 8304
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.506508009310396e-07,
+      "loss": 0.0621,
+      "step": 8305
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.495064880264544e-07,
+      "loss": 0.0662,
+      "step": 8306
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.48363331770787e-07,
+      "loss": 0.0689,
+      "step": 8307
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.472213323039488e-07,
+      "loss": 0.0625,
+      "step": 8308
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.460804897657102e-07,
+      "loss": 0.0639,
+      "step": 8309
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.449408042956971e-07,
+      "loss": 0.0687,
+      "step": 8310
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.438022760333994e-07,
+      "loss": 0.066,
+      "step": 8311
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.426649051181587e-07,
+      "loss": 0.0677,
+      "step": 8312
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.415286916891793e-07,
+      "loss": 0.076,
+      "step": 8313
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.403936358855233e-07,
+      "loss": 0.0668,
+      "step": 8314
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.392597378461107e-07,
+      "loss": 0.0615,
+      "step": 8315
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.381269977097203e-07,
+      "loss": 0.069,
+      "step": 8316
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.369954156149837e-07,
+      "loss": 0.0708,
+      "step": 8317
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.358649917004033e-07,
+      "loss": 0.0685,
+      "step": 8318
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.34735726104324e-07,
+      "loss": 0.0653,
+      "step": 8319
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.336076189649641e-07,
+      "loss": 0.0705,
+      "step": 8320
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.324806704203889e-07,
+      "loss": 0.0628,
+      "step": 8321
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.313548806085267e-07,
+      "loss": 0.0699,
+      "step": 8322
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.302302496671641e-07,
+      "loss": 0.0685,
+      "step": 8323
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.291067777339431e-07,
+      "loss": 0.0639,
+      "step": 8324
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.279844649463694e-07,
+      "loss": 0.0705,
+      "step": 8325
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.268633114417987e-07,
+      "loss": 0.072,
+      "step": 8326
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.257433173574533e-07,
+      "loss": 0.0637,
+      "step": 8327
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.24624482830407e-07,
+      "loss": 0.0649,
+      "step": 8328
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.235068079975958e-07,
+      "loss": 0.0699,
+      "step": 8329
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.223902929958113e-07,
+      "loss": 0.0698,
+      "step": 8330
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.212749379617044e-07,
+      "loss": 0.0682,
+      "step": 8331
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.201607430317834e-07,
+      "loss": 0.0662,
+      "step": 8332
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.190477083424173e-07,
+      "loss": 0.0696,
+      "step": 8333
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.17935834029829e-07,
+      "loss": 0.0672,
+      "step": 8334
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.168251202301e-07,
+      "loss": 0.0676,
+      "step": 8335
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.157155670791714e-07,
+      "loss": 0.072,
+      "step": 8336
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.146071747128433e-07,
+      "loss": 0.0641,
+      "step": 8337
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.134999432667708e-07,
+      "loss": 0.0632,
+      "step": 8338
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.123938728764688e-07,
+      "loss": 0.0672,
+      "step": 8339
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.11288963677311e-07,
+      "loss": 0.0672,
+      "step": 8340
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.101852158045239e-07,
+      "loss": 0.0683,
+      "step": 8341
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.090826293932005e-07,
+      "loss": 0.0676,
+      "step": 8342
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.079812045782828e-07,
+      "loss": 0.0641,
+      "step": 8343
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.068809414945752e-07,
+      "loss": 0.0653,
+      "step": 8344
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0578184027674e-07,
+      "loss": 0.0717,
+      "step": 8345
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.046839010592963e-07,
+      "loss": 0.0702,
+      "step": 8346
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.035871239766233e-07,
+      "loss": 0.0737,
+      "step": 8347
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.024915091629512e-07,
+      "loss": 0.0643,
+      "step": 8348
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.013970567523773e-07,
+      "loss": 0.0748,
+      "step": 8349
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.003037668788479e-07,
+      "loss": 0.0694,
+      "step": 8350
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.992116396761759e-07,
+      "loss": 0.0671,
+      "step": 8351
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981206752780232e-07,
+      "loss": 0.074,
+      "step": 8352
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.970308738179141e-07,
+      "loss": 0.0673,
+      "step": 8353
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.95942235429231e-07,
+      "loss": 0.0641,
+      "step": 8354
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.948547602452102e-07,
+      "loss": 0.0697,
+      "step": 8355
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.937684483989525e-07,
+      "loss": 0.0715,
+      "step": 8356
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.926833000234065e-07,
+      "loss": 0.0701,
+      "step": 8357
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.915993152513887e-07,
+      "loss": 0.0686,
+      "step": 8358
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.90516494215566e-07,
+      "loss": 0.0694,
+      "step": 8359
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.0664,
+      "step": 8360
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.883543438824701e-07,
+      "loss": 0.0713,
+      "step": 8361
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.87275014849824e-07,
+      "loss": 0.069,
+      "step": 8362
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.861968500826253e-07,
+      "loss": 0.0691,
+      "step": 8363
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.851198497128329e-07,
+      "loss": 0.0707,
+      "step": 8364
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.840440138722591e-07,
+      "loss": 0.0701,
+      "step": 8365
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.829693426925763e-07,
+      "loss": 0.0663,
+      "step": 8366
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.818958363053127e-07,
+      "loss": 0.0713,
+      "step": 8367
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.808234948418578e-07,
+      "loss": 0.0683,
+      "step": 8368
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.797523184334529e-07,
+      "loss": 0.0673,
+      "step": 8369
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.786823072112023e-07,
+      "loss": 0.067,
+      "step": 8370
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.776134613060646e-07,
+      "loss": 0.0725,
+      "step": 8371
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.765457808488538e-07,
+      "loss": 0.0627,
+      "step": 8372
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7547926597024675e-07,
+      "loss": 0.0676,
+      "step": 8373
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.744139168007722e-07,
+      "loss": 0.0705,
+      "step": 8374
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7334973347081926e-07,
+      "loss": 0.0708,
+      "step": 8375
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7228671611063373e-07,
+      "loss": 0.0672,
+      "step": 8376
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7122486485031813e-07,
+      "loss": 0.0654,
+      "step": 8377
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.701641798198353e-07,
+      "loss": 0.0645,
+      "step": 8378
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6910466114899797e-07,
+      "loss": 0.0666,
+      "step": 8379
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.680463089674869e-07,
+      "loss": 0.0726,
+      "step": 8380
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.669891234048285e-07,
+      "loss": 0.0727,
+      "step": 8381
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.659331045904181e-07,
+      "loss": 0.0728,
+      "step": 8382
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6487825265349564e-07,
+      "loss": 0.071,
+      "step": 8383
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6382456772316897e-07,
+      "loss": 0.0715,
+      "step": 8384
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.627720499283983e-07,
+      "loss": 0.0669,
+      "step": 8385
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.617206993980006e-07,
+      "loss": 0.0707,
+      "step": 8386
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6067051626065287e-07,
+      "loss": 0.0603,
+      "step": 8387
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5962150064488565e-07,
+      "loss": 0.0732,
+      "step": 8388
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5857365267908735e-07,
+      "loss": 0.0684,
+      "step": 8389
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5752697249150656e-07,
+      "loss": 0.0668,
+      "step": 8390
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.564814602102463e-07,
+      "loss": 0.0639,
+      "step": 8391
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5543711596326643e-07,
+      "loss": 0.0688,
+      "step": 8392
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5439393987838474e-07,
+      "loss": 0.0734,
+      "step": 8393
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.533519320832769e-07,
+      "loss": 0.07,
+      "step": 8394
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5231109270547526e-07,
+      "loss": 0.0669,
+      "step": 8395
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.512714218723646e-07,
+      "loss": 0.0657,
+      "step": 8396
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.502329197111943e-07,
+      "loss": 0.0642,
+      "step": 8397
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.491955863490649e-07,
+      "loss": 0.0667,
+      "step": 8398
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4815942191293706e-07,
+      "loss": 0.0717,
+      "step": 8399
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4712442652962705e-07,
+      "loss": 0.0703,
+      "step": 8400
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.460906003258081e-07,
+      "loss": 0.0717,
+      "step": 8401
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4505794342801225e-07,
+      "loss": 0.0697,
+      "step": 8402
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.440264559626228e-07,
+      "loss": 0.0678,
+      "step": 8403
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.429961380558878e-07,
+      "loss": 0.0678,
+      "step": 8404
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.419669898339063e-07,
+      "loss": 0.0619,
+      "step": 8405
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.409390114226353e-07,
+      "loss": 0.071,
+      "step": 8406
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.399122029478908e-07,
+      "loss": 0.0731,
+      "step": 8407
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3888656453534463e-07,
+      "loss": 0.0672,
+      "step": 8408
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.37862096310524e-07,
+      "loss": 0.0721,
+      "step": 8409
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3683879839881204e-07,
+      "loss": 0.0666,
+      "step": 8410
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.358166709254563e-07,
+      "loss": 0.0652,
+      "step": 8411
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.347957140155479e-07,
+      "loss": 0.074,
+      "step": 8412
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3377592779404787e-07,
+      "loss": 0.0671,
+      "step": 8413
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3275731238576645e-07,
+      "loss": 0.0668,
+      "step": 8414
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.317398679153706e-07,
+      "loss": 0.068,
+      "step": 8415
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3072359450738734e-07,
+      "loss": 0.0651,
+      "step": 8416
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.297084922861983e-07,
+      "loss": 0.0686,
+      "step": 8417
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.286945613760429e-07,
+      "loss": 0.0691,
+      "step": 8418
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2768180190101536e-07,
+      "loss": 0.0631,
+      "step": 8419
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.266702139850676e-07,
+      "loss": 0.0685,
+      "step": 8420
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2565979775200937e-07,
+      "loss": 0.0679,
+      "step": 8421
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.246505533255041e-07,
+      "loss": 0.0695,
+      "step": 8422
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.236424808290751e-07,
+      "loss": 0.0634,
+      "step": 8423
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.226355803860993e-07,
+      "loss": 0.0688,
+      "step": 8424
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2162985211981365e-07,
+      "loss": 0.0681,
+      "step": 8425
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.206252961533086e-07,
+      "loss": 0.0721,
+      "step": 8426
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1962191260952914e-07,
+      "loss": 0.0703,
+      "step": 8427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1861970161128584e-07,
+      "loss": 0.0682,
+      "step": 8428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.17618663281234e-07,
+      "loss": 0.0669,
+      "step": 8429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.166187977418934e-07,
+      "loss": 0.0665,
+      "step": 8430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1562010511563945e-07,
+      "loss": 0.0709,
+      "step": 8431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.146225855246977e-07,
+      "loss": 0.0617,
+      "step": 8432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.136262390911605e-07,
+      "loss": 0.0679,
+      "step": 8433
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1263106593696477e-07,
+      "loss": 0.0658,
+      "step": 8434
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1163706618391753e-07,
+      "loss": 0.0675,
+      "step": 8435
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.106442399536681e-07,
+      "loss": 0.068,
+      "step": 8436
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.096525873677337e-07,
+      "loss": 0.062,
+      "step": 8437
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0866210854747956e-07,
+      "loss": 0.0606,
+      "step": 8438
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.07672803614132e-07,
+      "loss": 0.0683,
+      "step": 8439
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.06684672688773e-07,
+      "loss": 0.0662,
+      "step": 8440
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.056977158923381e-07,
+      "loss": 0.0713,
+      "step": 8441
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0471193334562507e-07,
+      "loss": 0.0708,
+      "step": 8442
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.037273251692797e-07,
+      "loss": 0.0746,
+      "step": 8443
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.027438914838122e-07,
+      "loss": 0.0658,
+      "step": 8444
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.017616324095819e-07,
+      "loss": 0.066,
+      "step": 8445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.007805480668114e-07,
+      "loss": 0.0698,
+      "step": 8446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.998006385755726e-07,
+      "loss": 0.0653,
+      "step": 8447
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9882190405579945e-07,
+      "loss": 0.0619,
+      "step": 8448
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.978443446272784e-07,
+      "loss": 0.0646,
+      "step": 8449
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.968679604096548e-07,
+      "loss": 0.0695,
+      "step": 8450
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.958927515224265e-07,
+      "loss": 0.072,
+      "step": 8451
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.949187180849501e-07,
+      "loss": 0.0636,
+      "step": 8452
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.939458602164381e-07,
+      "loss": 0.06,
+      "step": 8453
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9297417803595863e-07,
+      "loss": 0.0676,
+      "step": 8454
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9200367166243757e-07,
+      "loss": 0.0644,
+      "step": 8455
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9103434121465334e-07,
+      "loss": 0.0688,
+      "step": 8456
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.900661868112454e-07,
+      "loss": 0.0702,
+      "step": 8457
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8909920857070237e-07,
+      "loss": 0.0704,
+      "step": 8458
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.881334066113784e-07,
+      "loss": 0.0675,
+      "step": 8459
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8716878105147347e-07,
+      "loss": 0.0554,
+      "step": 8460
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8620533200905083e-07,
+      "loss": 0.0692,
+      "step": 8461
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8524305960202735e-07,
+      "loss": 0.067,
+      "step": 8462
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.842819639481732e-07,
+      "loss": 0.0645,
+      "step": 8463
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8332204516512206e-07,
+      "loss": 0.0665,
+      "step": 8464
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.823633033703533e-07,
+      "loss": 0.0688,
+      "step": 8465
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.814057386812131e-07,
+      "loss": 0.0664,
+      "step": 8466
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.804493512148921e-07,
+      "loss": 0.0714,
+      "step": 8467
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7949414108844893e-07,
+      "loss": 0.0693,
+      "step": 8468
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.785401084187879e-07,
+      "loss": 0.0707,
+      "step": 8469
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.775872533226743e-07,
+      "loss": 0.0683,
+      "step": 8470
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.766355759167295e-07,
+      "loss": 0.0743,
+      "step": 8471
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.75685076317428e-07,
+      "loss": 0.0634,
+      "step": 8472
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7473575464110456e-07,
+      "loss": 0.0723,
+      "step": 8473
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.737876110039429e-07,
+      "loss": 0.0653,
+      "step": 8474
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.728406455219891e-07,
+      "loss": 0.0705,
+      "step": 8475
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.718948583111426e-07,
+      "loss": 0.0695,
+      "step": 8476
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.709502494871575e-07,
+      "loss": 0.0677,
+      "step": 8477
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7000681916564583e-07,
+      "loss": 0.0666,
+      "step": 8478
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.690645674620741e-07,
+      "loss": 0.0686,
+      "step": 8479
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.681234944917644e-07,
+      "loss": 0.0686,
+      "step": 8480
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6718360036989587e-07,
+      "loss": 0.0654,
+      "step": 8481
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.662448852115008e-07,
+      "loss": 0.0715,
+      "step": 8482
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.653073491314707e-07,
+      "loss": 0.071,
+      "step": 8483
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.643709922445493e-07,
+      "loss": 0.0734,
+      "step": 8484
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634358146653383e-07,
+      "loss": 0.0696,
+      "step": 8485
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.62501816508295e-07,
+      "loss": 0.0641,
+      "step": 8486
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6156899788773017e-07,
+      "loss": 0.0623,
+      "step": 8487
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606373589178136e-07,
+      "loss": 0.0646,
+      "step": 8488
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.597068997125663e-07,
+      "loss": 0.0721,
+      "step": 8489
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5877762038587283e-07,
+      "loss": 0.0653,
+      "step": 8490
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5784952105146207e-07,
+      "loss": 0.072,
+      "step": 8491
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.569226018229266e-07,
+      "loss": 0.0619,
+      "step": 8492
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5599686281371227e-07,
+      "loss": 0.0664,
+      "step": 8493
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5507230413712067e-07,
+      "loss": 0.0667,
+      "step": 8494
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.541489259063102e-07,
+      "loss": 0.0679,
+      "step": 8495
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.532267282342905e-07,
+      "loss": 0.0735,
+      "step": 8496
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.523057112339334e-07,
+      "loss": 0.067,
+      "step": 8497
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5138587501795884e-07,
+      "loss": 0.068,
+      "step": 8498
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.504672196989467e-07,
+      "loss": 0.0669,
+      "step": 8499
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4954974538933264e-07,
+      "loss": 0.0686,
+      "step": 8500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4863345220140675e-07,
+      "loss": 0.0647,
+      "step": 8501
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.477183402473139e-07,
+      "loss": 0.0619,
+      "step": 8502
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.468044096390544e-07,
+      "loss": 0.0698,
+      "step": 8503
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4589166048848546e-07,
+      "loss": 0.0707,
+      "step": 8504
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.449800929073177e-07,
+      "loss": 0.0626,
+      "step": 8505
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.440697070071197e-07,
+      "loss": 0.0617,
+      "step": 8506
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.431605028993112e-07,
+      "loss": 0.0697,
+      "step": 8507
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4225248069517326e-07,
+      "loss": 0.0697,
+      "step": 8508
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4134564050583684e-07,
+      "loss": 0.0699,
+      "step": 8509
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4043998244229105e-07,
+      "loss": 0.076,
+      "step": 8510
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3953550661537826e-07,
+      "loss": 0.069,
+      "step": 8511
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.386322131358022e-07,
+      "loss": 0.0647,
+      "step": 8512
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3773010211411105e-07,
+      "loss": 0.0728,
+      "step": 8513
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.368291736607188e-07,
+      "loss": 0.0715,
+      "step": 8514
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.359294278858882e-07,
+      "loss": 0.0646,
+      "step": 8515
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.350308648997402e-07,
+      "loss": 0.0717,
+      "step": 8516
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3413348481225126e-07,
+      "loss": 0.0661,
+      "step": 8517
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3323728773324904e-07,
+      "loss": 0.067,
+      "step": 8518
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3234227377242354e-07,
+      "loss": 0.0698,
+      "step": 8519
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3144844303931057e-07,
+      "loss": 0.0719,
+      "step": 8520
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3055579564331253e-07,
+      "loss": 0.0666,
+      "step": 8521
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.296643316936765e-07,
+      "loss": 0.0691,
+      "step": 8522
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2877405129950966e-07,
+      "loss": 0.0726,
+      "step": 8523
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.278849545697738e-07,
+      "loss": 0.0676,
+      "step": 8524
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.269970416132873e-07,
+      "loss": 0.0633,
+      "step": 8525
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.261103125387211e-07,
+      "loss": 0.0648,
+      "step": 8526
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.252247674546005e-07,
+      "loss": 0.0643,
+      "step": 8527
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.243404064693112e-07,
+      "loss": 0.073,
+      "step": 8528
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2345722969108764e-07,
+      "loss": 0.0662,
+      "step": 8529
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.225752372280222e-07,
+      "loss": 0.0715,
+      "step": 8530
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.216944291880642e-07,
+      "loss": 0.0663,
+      "step": 8531
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.208148056790139e-07,
+      "loss": 0.0745,
+      "step": 8532
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.199363668085298e-07,
+      "loss": 0.0712,
+      "step": 8533
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.190591126841247e-07,
+      "loss": 0.0706,
+      "step": 8534
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.181830434131661e-07,
+      "loss": 0.0689,
+      "step": 8535
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.173081591028748e-07,
+      "loss": 0.0625,
+      "step": 8536
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.164344598603286e-07,
+      "loss": 0.067,
+      "step": 8537
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.155619457924608e-07,
+      "loss": 0.0724,
+      "step": 8538
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.146906170060582e-07,
+      "loss": 0.0669,
+      "step": 8539
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1382047360776325e-07,
+      "loss": 0.0664,
+      "step": 8540
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.12951515704073e-07,
+      "loss": 0.0718,
+      "step": 8541
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1208374340133797e-07,
+      "loss": 0.0607,
+      "step": 8542
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.112171568057687e-07,
+      "loss": 0.0705,
+      "step": 8543
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.103517560234226e-07,
+      "loss": 0.067,
+      "step": 8544
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.094875411602194e-07,
+      "loss": 0.0753,
+      "step": 8545
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0862451232193004e-07,
+      "loss": 0.0619,
+      "step": 8546
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0776266961418e-07,
+      "loss": 0.0674,
+      "step": 8547
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.069020131424527e-07,
+      "loss": 0.0665,
+      "step": 8548
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0604254301207945e-07,
+      "loss": 0.0691,
+      "step": 8549
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.051842593282572e-07,
+      "loss": 0.0691,
+      "step": 8550
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0432716219602534e-07,
+      "loss": 0.068,
+      "step": 8551
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.034712517202909e-07,
+      "loss": 0.0668,
+      "step": 8552
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0261652800580355e-07,
+      "loss": 0.0674,
+      "step": 8553
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0176299115717514e-07,
+      "loss": 0.0691,
+      "step": 8554
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0091064127887094e-07,
+      "loss": 0.0724,
+      "step": 8555
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.000594784752098e-07,
+      "loss": 0.0672,
+      "step": 8556
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.992095028503661e-07,
+      "loss": 0.07,
+      "step": 8557
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9836071450836776e-07,
+      "loss": 0.0689,
+      "step": 8558
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9751311355309953e-07,
+      "loss": 0.07,
+      "step": 8559
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9666670008829833e-07,
+      "loss": 0.0645,
+      "step": 8560
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9582147421755693e-07,
+      "loss": 0.0693,
+      "step": 8561
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.949774360443236e-07,
+      "loss": 0.066,
+      "step": 8562
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.941345856719013e-07,
+      "loss": 0.0655,
+      "step": 8563
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9329292320344425e-07,
+      "loss": 0.0657,
+      "step": 8564
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9245244874196665e-07,
+      "loss": 0.0687,
+      "step": 8565
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9161316239033177e-07,
+      "loss": 0.0634,
+      "step": 8566
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.907750642512619e-07,
+      "loss": 0.0668,
+      "step": 8567
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8993815442733166e-07,
+      "loss": 0.0709,
+      "step": 8568
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.891024330209702e-07,
+      "loss": 0.0641,
+      "step": 8569
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.882679001344624e-07,
+      "loss": 0.0632,
+      "step": 8570
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.874345558699465e-07,
+      "loss": 0.0686,
+      "step": 8571
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.866024003294166e-07,
+      "loss": 0.0616,
+      "step": 8572
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.857714336147188e-07,
+      "loss": 0.0691,
+      "step": 8573
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8494165582755863e-07,
+      "loss": 0.0704,
+      "step": 8574
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8411306706948807e-07,
+      "loss": 0.068,
+      "step": 8575
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8328566744192265e-07,
+      "loss": 0.0766,
+      "step": 8576
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8245945704612586e-07,
+      "loss": 0.0666,
+      "step": 8577
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.81634435983219e-07,
+      "loss": 0.0662,
+      "step": 8578
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8081060435417693e-07,
+      "loss": 0.0684,
+      "step": 8579
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.799879622598256e-07,
+      "loss": 0.0636,
+      "step": 8580
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7916650980085225e-07,
+      "loss": 0.0673,
+      "step": 8581
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7834624707779203e-07,
+      "loss": 0.0696,
+      "step": 8582
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.775271741910401e-07,
+      "loss": 0.0669,
+      "step": 8583
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.767092912408398e-07,
+      "loss": 0.0729,
+      "step": 8584
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7589259832729307e-07,
+      "loss": 0.0691,
+      "step": 8585
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.750770955503568e-07,
+      "loss": 0.07,
+      "step": 8586
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7426278300983877e-07,
+      "loss": 0.0683,
+      "step": 8587
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.734496608054049e-07,
+      "loss": 0.0711,
+      "step": 8588
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7263772903657113e-07,
+      "loss": 0.0665,
+      "step": 8589
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.718269878027113e-07,
+      "loss": 0.0669,
+      "step": 8590
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.710174372030516e-07,
+      "loss": 0.0673,
+      "step": 8591
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.70209077336675e-07,
+      "loss": 0.0667,
+      "step": 8592
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6940190830251454e-07,
+      "loss": 0.0693,
+      "step": 8593
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6859593019936124e-07,
+      "loss": 0.064,
+      "step": 8594
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.677911431258584e-07,
+      "loss": 0.0646,
+      "step": 8595
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6698754718050614e-07,
+      "loss": 0.0676,
+      "step": 8596
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.661851424616535e-07,
+      "loss": 0.0644,
+      "step": 8597
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.653839290675109e-07,
+      "loss": 0.0653,
+      "step": 8598
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6458390709613537e-07,
+      "loss": 0.0673,
+      "step": 8599
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6378507664544306e-07,
+      "loss": 0.0643,
+      "step": 8600
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.629874378132058e-07,
+      "loss": 0.0694,
+      "step": 8601
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.62190990697041e-07,
+      "loss": 0.0659,
+      "step": 8602
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6139573539443295e-07,
+      "loss": 0.0705,
+      "step": 8603
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6060167200270714e-07,
+      "loss": 0.0686,
+      "step": 8604
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5980880061905465e-07,
+      "loss": 0.0683,
+      "step": 8605
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5901712134051014e-07,
+      "loss": 0.0644,
+      "step": 8606
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5822663426397165e-07,
+      "loss": 0.0682,
+      "step": 8607
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.574373394861851e-07,
+      "loss": 0.0754,
+      "step": 8608
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.566492371037521e-07,
+      "loss": 0.0662,
+      "step": 8609
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5586232721313e-07,
+      "loss": 0.0668,
+      "step": 8610
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.550766099106294e-07,
+      "loss": 0.0627,
+      "step": 8611
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5429208529241354e-07,
+      "loss": 0.0654,
+      "step": 8612
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.535087534544989e-07,
+      "loss": 0.0676,
+      "step": 8613
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.527266144927609e-07,
+      "loss": 0.0675,
+      "step": 8614
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.519456685029242e-07,
+      "loss": 0.066,
+      "step": 8615
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5116591558056903e-07,
+      "loss": 0.0717,
+      "step": 8616
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5038735582113004e-07,
+      "loss": 0.0752,
+      "step": 8617
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.496099893198944e-07,
+      "loss": 0.0699,
+      "step": 8618
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.48833816172005e-07,
+      "loss": 0.0776,
+      "step": 8619
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4805883647245896e-07,
+      "loss": 0.0702,
+      "step": 8620
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4728505031610505e-07,
+      "loss": 0.0691,
+      "step": 8621
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4651245779764634e-07,
+      "loss": 0.0698,
+      "step": 8622
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.457410590116427e-07,
+      "loss": 0.0705,
+      "step": 8623
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.449708540525042e-07,
+      "loss": 0.0698,
+      "step": 8624
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4420184301449765e-07,
+      "loss": 0.07,
+      "step": 8625
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4343402599174116e-07,
+      "loss": 0.0694,
+      "step": 8626
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4266740307821057e-07,
+      "loss": 0.0683,
+      "step": 8627
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.419019743677298e-07,
+      "loss": 0.0705,
+      "step": 8628
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4113773995398383e-07,
+      "loss": 0.0697,
+      "step": 8629
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4037469993050347e-07,
+      "loss": 0.0671,
+      "step": 8630
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.396128543906806e-07,
+      "loss": 0.0682,
+      "step": 8631
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.388522034277552e-07,
+      "loss": 0.0647,
+      "step": 8632
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.38092747134826e-07,
+      "loss": 0.0671,
+      "step": 8633
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3733448560484207e-07,
+      "loss": 0.0631,
+      "step": 8634
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3657741893060582e-07,
+      "loss": 0.0766,
+      "step": 8635
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3582154720477758e-07,
+      "loss": 0.0757,
+      "step": 8636
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3506687051986555e-07,
+      "loss": 0.0671,
+      "step": 8637
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3431338896823917e-07,
+      "loss": 0.0642,
+      "step": 8638
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3356110264211364e-07,
+      "loss": 0.0677,
+      "step": 8639
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3281001163356187e-07,
+      "loss": 0.0614,
+      "step": 8640
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.320601160345115e-07,
+      "loss": 0.0614,
+      "step": 8641
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3131141593674244e-07,
+      "loss": 0.0619,
+      "step": 8642
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3056391143188806e-07,
+      "loss": 0.0714,
+      "step": 8643
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2981760261143517e-07,
+      "loss": 0.0657,
+      "step": 8644
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2907248956672513e-07,
+      "loss": 0.0708,
+      "step": 8645
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2832857238895167e-07,
+      "loss": 0.0685,
+      "step": 8646
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.275858511691642e-07,
+      "loss": 0.0658,
+      "step": 8647
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2684432599826444e-07,
+      "loss": 0.0663,
+      "step": 8648
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2610399696700758e-07,
+      "loss": 0.0673,
+      "step": 8649
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2536486416600222e-07,
+      "loss": 0.0656,
+      "step": 8650
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2462692768571382e-07,
+      "loss": 0.0636,
+      "step": 8651
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2389018761645453e-07,
+      "loss": 0.0601,
+      "step": 8652
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.231546440483967e-07,
+      "loss": 0.0642,
+      "step": 8653
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2242029707156276e-07,
+      "loss": 0.0633,
+      "step": 8654
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2168714677583082e-07,
+      "loss": 0.0724,
+      "step": 8655
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2095519325093128e-07,
+      "loss": 0.0652,
+      "step": 8656
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2022443658644588e-07,
+      "loss": 0.0644,
+      "step": 8657
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1949487687181525e-07,
+      "loss": 0.0686,
+      "step": 8658
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.18766514196328e-07,
+      "loss": 0.0699,
+      "step": 8659
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1803934864913167e-07,
+      "loss": 0.068,
+      "step": 8660
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1731338031922066e-07,
+      "loss": 0.0592,
+      "step": 8661
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1658860929544834e-07,
+      "loss": 0.0661,
+      "step": 8662
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.158650356665193e-07,
+      "loss": 0.0636,
+      "step": 8663
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1514265952099267e-07,
+      "loss": 0.0727,
+      "step": 8664
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.144214809472811e-07,
+      "loss": 0.0662,
+      "step": 8665
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.137015000336451e-07,
+      "loss": 0.0625,
+      "step": 8666
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.129827168682097e-07,
+      "loss": 0.0651,
+      "step": 8667
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1226513153894345e-07,
+      "loss": 0.0686,
+      "step": 8668
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1154874413367165e-07,
+      "loss": 0.0665,
+      "step": 8669
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1083355474007417e-07,
+      "loss": 0.0664,
+      "step": 8670
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1011956344568429e-07,
+      "loss": 0.068,
+      "step": 8671
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0940677033788547e-07,
+      "loss": 0.0677,
+      "step": 8672
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0869517550391682e-07,
+      "loss": 0.0655,
+      "step": 8673
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0798477903087423e-07,
+      "loss": 0.076,
+      "step": 8674
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0727558100569812e-07,
+      "loss": 0.0692,
+      "step": 8675
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.065675815151913e-07,
+      "loss": 0.0656,
+      "step": 8676
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0586078064600335e-07,
+      "loss": 0.0718,
+      "step": 8677
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0515517848464284e-07,
+      "loss": 0.0749,
+      "step": 8678
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0445077511746625e-07,
+      "loss": 0.0638,
+      "step": 8679
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0374757063068573e-07,
+      "loss": 0.0616,
+      "step": 8680
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0304556511036798e-07,
+      "loss": 0.0687,
+      "step": 8681
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.023447586424332e-07,
+      "loss": 0.0698,
+      "step": 8682
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0164515131264939e-07,
+      "loss": 0.0677,
+      "step": 8683
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0094674320664364e-07,
+      "loss": 0.0679,
+      "step": 8684
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0024953440989426e-07,
+      "loss": 0.073,
+      "step": 8685
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9955352500773295e-07,
+      "loss": 0.072,
+      "step": 8686
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9885871508534605e-07,
+      "loss": 0.0716,
+      "step": 8687
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9816510472776773e-07,
+      "loss": 0.0699,
+      "step": 8688
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9747269401989343e-07,
+      "loss": 0.0687,
+      "step": 8689
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9678148304646317e-07,
+      "loss": 0.0725,
+      "step": 8690
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.960914718920792e-07,
+      "loss": 0.0663,
+      "step": 8691
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9540266064118962e-07,
+      "loss": 0.0703,
+      "step": 8692
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9471504937809693e-07,
+      "loss": 0.0697,
+      "step": 8693
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.940286381869616e-07,
+      "loss": 0.0633,
+      "step": 8694
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9334342715178979e-07,
+      "loss": 0.0762,
+      "step": 8695
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.926594163564488e-07,
+      "loss": 0.0697,
+      "step": 8696
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.919766058846506e-07,
+      "loss": 0.0693,
+      "step": 8697
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9129499581996945e-07,
+      "loss": 0.0706,
+      "step": 8698
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9061458624582307e-07,
+      "loss": 0.0687,
+      "step": 8699
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8993537724549038e-07,
+      "loss": 0.0707,
+      "step": 8700
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8925736890209824e-07,
+      "loss": 0.0642,
+      "step": 8701
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.88580561298628e-07,
+      "loss": 0.0649,
+      "step": 8702
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8790495451791568e-07,
+      "loss": 0.0704,
+      "step": 8703
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8723054864264955e-07,
+      "loss": 0.0616,
+      "step": 8704
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8655734375536915e-07,
+      "loss": 0.064,
+      "step": 8705
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8588533993846747e-07,
+      "loss": 0.0646,
+      "step": 8706
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8521453727419203e-07,
+      "loss": 0.0713,
+      "step": 8707
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8454493584464272e-07,
+      "loss": 0.0652,
+      "step": 8708
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.838765357317729e-07,
+      "loss": 0.0795,
+      "step": 8709
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8320933701738709e-07,
+      "loss": 0.0661,
+      "step": 8710
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8254333978314552e-07,
+      "loss": 0.0718,
+      "step": 8711
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.818785441105575e-07,
+      "loss": 0.0716,
+      "step": 8712
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8121495008099012e-07,
+      "loss": 0.0721,
+      "step": 8713
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.805525577756584e-07,
+      "loss": 0.0744,
+      "step": 8714
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7989136727563528e-07,
+      "loss": 0.0711,
+      "step": 8715
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.792313786618427e-07,
+      "loss": 0.0712,
+      "step": 8716
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7857259201505606e-07,
+      "loss": 0.0662,
+      "step": 8717
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.779150074159064e-07,
+      "loss": 0.0614,
+      "step": 8718
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7725862494487377e-07,
+      "loss": 0.067,
+      "step": 8719
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7660344468229508e-07,
+      "loss": 0.0683,
+      "step": 8720
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.75949466708355e-07,
+      "loss": 0.0638,
+      "step": 8721
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7529669110309733e-07,
+      "loss": 0.0648,
+      "step": 8722
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.746451179464137e-07,
+      "loss": 0.0647,
+      "step": 8723
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7399474731805145e-07,
+      "loss": 0.0649,
+      "step": 8724
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7334557929760797e-07,
+      "loss": 0.0675,
+      "step": 8725
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7269761396453754e-07,
+      "loss": 0.0675,
+      "step": 8726
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7205085139814336e-07,
+      "loss": 0.0722,
+      "step": 8727
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7140529167758103e-07,
+      "loss": 0.0732,
+      "step": 8728
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7076093488186506e-07,
+      "loss": 0.0691,
+      "step": 8729
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7011778108985578e-07,
+      "loss": 0.0663,
+      "step": 8730
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6947583038026906e-07,
+      "loss": 0.0727,
+      "step": 8731
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6883508283167428e-07,
+      "loss": 0.0686,
+      "step": 8732
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6819553852249092e-07,
+      "loss": 0.0694,
+      "step": 8733
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6755719753099643e-07,
+      "loss": 0.0693,
+      "step": 8734
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.669200599353149e-07,
+      "loss": 0.0678,
+      "step": 8735
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6628412581342513e-07,
+      "loss": 0.0678,
+      "step": 8736
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.656493952431626e-07,
+      "loss": 0.0697,
+      "step": 8737
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6501586830220852e-07,
+      "loss": 0.0612,
+      "step": 8738
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6438354506810305e-07,
+      "loss": 0.07,
+      "step": 8739
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6375242561823545e-07,
+      "loss": 0.0665,
+      "step": 8740
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6312251002984835e-07,
+      "loss": 0.0707,
+      "step": 8741
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6249379838003788e-07,
+      "loss": 0.0713,
+      "step": 8742
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6186629074575133e-07,
+      "loss": 0.072,
+      "step": 8743
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6123998720379175e-07,
+      "loss": 0.0674,
+      "step": 8744
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6061488783081002e-07,
+      "loss": 0.067,
+      "step": 8745
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5999099270331387e-07,
+      "loss": 0.0675,
+      "step": 8746
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5936830189766218e-07,
+      "loss": 0.0699,
+      "step": 8747
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.58746815490064e-07,
+      "loss": 0.0733,
+      "step": 8748
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5812653355658625e-07,
+      "loss": 0.066,
+      "step": 8749
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5750745617314377e-07,
+      "loss": 0.0692,
+      "step": 8750
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5688958341550597e-07,
+      "loss": 0.0708,
+      "step": 8751
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5627291535929345e-07,
+      "loss": 0.0775,
+      "step": 8752
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5565745207998252e-07,
+      "loss": 0.0685,
+      "step": 8753
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5504319365289843e-07,
+      "loss": 0.0738,
+      "step": 8754
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.544301401532211e-07,
+      "loss": 0.0676,
+      "step": 8755
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.538182916559816e-07,
+      "loss": 0.0712,
+      "step": 8756
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.532076482360656e-07,
+      "loss": 0.0684,
+      "step": 8757
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5259820996820884e-07,
+      "loss": 0.0682,
+      "step": 8758
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5198997692700058e-07,
+      "loss": 0.0666,
+      "step": 8759
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5138294918688345e-07,
+      "loss": 0.0697,
+      "step": 8760
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.507771268221514e-07,
+      "loss": 0.0713,
+      "step": 8761
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5017250990695065e-07,
+      "loss": 0.0678,
+      "step": 8762
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.495690985152809e-07,
+      "loss": 0.0692,
+      "step": 8763
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4896689272099417e-07,
+      "loss": 0.0662,
+      "step": 8764
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4836589259779488e-07,
+      "loss": 0.0632,
+      "step": 8765
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.477660982192397e-07,
+      "loss": 0.0639,
+      "step": 8766
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4716750965873438e-07,
+      "loss": 0.0635,
+      "step": 8767
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4657012698954587e-07,
+      "loss": 0.0736,
+      "step": 8768
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4597395028478456e-07,
+      "loss": 0.0717,
+      "step": 8769
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4537897961741653e-07,
+      "loss": 0.0667,
+      "step": 8770
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4478521506026133e-07,
+      "loss": 0.0679,
+      "step": 8771
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.441926566859908e-07,
+      "loss": 0.0674,
+      "step": 8772
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4360130456712695e-07,
+      "loss": 0.0651,
+      "step": 8773
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.430111587760452e-07,
+      "loss": 0.0644,
+      "step": 8774
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.424222193849767e-07,
+      "loss": 0.0683,
+      "step": 8775
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4183448646599707e-07,
+      "loss": 0.0697,
+      "step": 8776
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4124796009104435e-07,
+      "loss": 0.0682,
+      "step": 8777
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4066264033190002e-07,
+      "loss": 0.0719,
+      "step": 8778
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4007852726020233e-07,
+      "loss": 0.071,
+      "step": 8779
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3949562094744295e-07,
+      "loss": 0.0657,
+      "step": 8780
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3891392146496153e-07,
+      "loss": 0.0639,
+      "step": 8781
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.383334288839533e-07,
+      "loss": 0.0784,
+      "step": 8782
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3775414327546477e-07,
+      "loss": 0.0745,
+      "step": 8783
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.371760647103959e-07,
+      "loss": 0.0666,
+      "step": 8784
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3659919325949678e-07,
+      "loss": 0.0639,
+      "step": 8785
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.36023528993372e-07,
+      "loss": 0.0734,
+      "step": 8786
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3544907198247527e-07,
+      "loss": 0.0639,
+      "step": 8787
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3487582229711694e-07,
+      "loss": 0.068,
+      "step": 8788
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3430378000745535e-07,
+      "loss": 0.0727,
+      "step": 8789
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3373294518350454e-07,
+      "loss": 0.0686,
+      "step": 8790
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.331633178951286e-07,
+      "loss": 0.0729,
+      "step": 8791
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3259489821204284e-07,
+      "loss": 0.0664,
+      "step": 8792
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3202768620381833e-07,
+      "loss": 0.0726,
+      "step": 8793
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3146168193987617e-07,
+      "loss": 0.0692,
+      "step": 8794
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3089688548948876e-07,
+      "loss": 0.0699,
+      "step": 8795
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3033329692178188e-07,
+      "loss": 0.0666,
+      "step": 8796
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2977091630573479e-07,
+      "loss": 0.066,
+      "step": 8797
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.292097437101747e-07,
+      "loss": 0.0686,
+      "step": 8798
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2864977920378662e-07,
+      "loss": 0.0723,
+      "step": 8799
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2809102285510356e-07,
+      "loss": 0.0628,
+      "step": 8800
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2753347473251076e-07,
+      "loss": 0.0732,
+      "step": 8801
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2697713490424922e-07,
+      "loss": 0.0678,
+      "step": 8802
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.264220034384067e-07,
+      "loss": 0.068,
+      "step": 8803
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2586808040292886e-07,
+      "loss": 0.068,
+      "step": 8804
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2531536586560698e-07,
+      "loss": 0.0727,
+      "step": 8805
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2476385989409146e-07,
+      "loss": 0.0653,
+      "step": 8806
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2421356255587824e-07,
+      "loss": 0.0656,
+      "step": 8807
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2366447391832237e-07,
+      "loss": 0.07,
+      "step": 8808
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.0655,
+      "step": 8809
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2256992301383663e-07,
+      "loss": 0.0658,
+      "step": 8810
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2202446088087184e-07,
+      "loss": 0.0638,
+      "step": 8811
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2148020771648562e-07,
+      "loss": 0.0685,
+      "step": 8812
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2093716358729135e-07,
+      "loss": 0.0696,
+      "step": 8813
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2039532855975144e-07,
+      "loss": 0.072,
+      "step": 8814
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1985470270018175e-07,
+      "loss": 0.0698,
+      "step": 8815
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1931528607474818e-07,
+      "loss": 0.071,
+      "step": 8816
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.187770787494713e-07,
+      "loss": 0.0669,
+      "step": 8817
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1824008079022286e-07,
+      "loss": 0.0726,
+      "step": 8818
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1770429226272473e-07,
+      "loss": 0.0653,
+      "step": 8819
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1716971323255444e-07,
+      "loss": 0.0753,
+      "step": 8820
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1663634376513743e-07,
+      "loss": 0.0678,
+      "step": 8821
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1610418392575373e-07,
+      "loss": 0.0637,
+      "step": 8822
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1557323377953456e-07,
+      "loss": 0.0649,
+      "step": 8823
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1504349339146237e-07,
+      "loss": 0.0697,
+      "step": 8824
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1451496282637308e-07,
+      "loss": 0.0678,
+      "step": 8825
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1398764214895274e-07,
+      "loss": 0.065,
+      "step": 8826
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1346153142374195e-07,
+      "loss": 0.0751,
+      "step": 8827
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1293663071513138e-07,
+      "loss": 0.0664,
+      "step": 8828
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1241294008736082e-07,
+      "loss": 0.069,
+      "step": 8829
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1189045960452894e-07,
+      "loss": 0.066,
+      "step": 8830
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1136918933058016e-07,
+      "loss": 0.0669,
+      "step": 8831
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1084912932931236e-07,
+      "loss": 0.0684,
+      "step": 8832
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1033027966437682e-07,
+      "loss": 0.069,
+      "step": 8833
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0981264039927608e-07,
+      "loss": 0.0676,
+      "step": 8834
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0929621159736392e-07,
+      "loss": 0.0667,
+      "step": 8835
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0878099332184532e-07,
+      "loss": 0.0693,
+      "step": 8836
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0826698563577987e-07,
+      "loss": 0.0696,
+      "step": 8837
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0775418860207498e-07,
+      "loss": 0.0642,
+      "step": 8838
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0724260228349382e-07,
+      "loss": 0.061,
+      "step": 8839
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0673222674264849e-07,
+      "loss": 0.0649,
+      "step": 8840
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.062230620420035e-07,
+      "loss": 0.0731,
+      "step": 8841
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0571510824387787e-07,
+      "loss": 0.0627,
+      "step": 8842
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0520836541043855e-07,
+      "loss": 0.0713,
+      "step": 8843
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0470283360370592e-07,
+      "loss": 0.0666,
+      "step": 8844
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0419851288555272e-07,
+      "loss": 0.0669,
+      "step": 8845
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0369540331770289e-07,
+      "loss": 0.0665,
+      "step": 8846
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0319350496173164e-07,
+      "loss": 0.0701,
+      "step": 8847
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0269281787906759e-07,
+      "loss": 0.0699,
+      "step": 8848
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0219334213098842e-07,
+      "loss": 0.0683,
+      "step": 8849
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.016950777786263e-07,
+      "loss": 0.0732,
+      "step": 8850
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.011980248829636e-07,
+      "loss": 0.0656,
+      "step": 8851
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0070218350483496e-07,
+      "loss": 0.0695,
+      "step": 8852
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.002075537049263e-07,
+      "loss": 0.0622,
+      "step": 8853
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.971413554377584e-08,
+      "loss": 0.0673,
+      "step": 8854
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.922192908177308e-08,
+      "loss": 0.0683,
+      "step": 8855
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.87309343791587e-08,
+      "loss": 0.0661,
+      "step": 8856
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.824115149602687e-08,
+      "loss": 0.0644,
+      "step": 8857
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.775258049232072e-08,
+      "loss": 0.0747,
+      "step": 8858
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.726522142783912e-08,
+      "loss": 0.0675,
+      "step": 8859
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.677907436222656e-08,
+      "loss": 0.0707,
+      "step": 8860
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.629413935498655e-08,
+      "loss": 0.0679,
+      "step": 8861
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.581041646546718e-08,
+      "loss": 0.0732,
+      "step": 8862
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.532790575287332e-08,
+      "loss": 0.0713,
+      "step": 8863
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.484660727625993e-08,
+      "loss": 0.0731,
+      "step": 8864
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.436652109453104e-08,
+      "loss": 0.0663,
+      "step": 8865
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.388764726644739e-08,
+      "loss": 0.0683,
+      "step": 8866
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.340998585061656e-08,
+      "loss": 0.0703,
+      "step": 8867
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.293353690550067e-08,
+      "loss": 0.0639,
+      "step": 8868
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.245830048941084e-08,
+      "loss": 0.0671,
+      "step": 8869
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.198427666051279e-08,
+      "loss": 0.0607,
+      "step": 8870
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.151146547682343e-08,
+      "loss": 0.0657,
+      "step": 8871
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.103986699620759e-08,
+      "loss": 0.0708,
+      "step": 8872
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.056948127638687e-08,
+      "loss": 0.0663,
+      "step": 8873
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.010030837493078e-08,
+      "loss": 0.0686,
+      "step": 8874
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.963234834926227e-08,
+      "loss": 0.0696,
+      "step": 8875
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.91656012566533e-08,
+      "loss": 0.0612,
+      "step": 8876
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.870006715423152e-08,
+      "loss": 0.0649,
+      "step": 8877
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.823574609897134e-08,
+      "loss": 0.0732,
+      "step": 8878
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.777263814770398e-08,
+      "loss": 0.0683,
+      "step": 8879
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.731074335710742e-08,
+      "loss": 0.0714,
+      "step": 8880
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.685006178371314e-08,
+      "loss": 0.0629,
+      "step": 8881
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.639059348390488e-08,
+      "loss": 0.065,
+      "step": 8882
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.593233851391658e-08,
+      "loss": 0.0687,
+      "step": 8883
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.547529692983447e-08,
+      "loss": 0.0689,
+      "step": 8884
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.501946878759715e-08,
+      "loss": 0.0709,
+      "step": 8885
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.456485414299109e-08,
+      "loss": 0.0712,
+      "step": 8886
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.411145305165847e-08,
+      "loss": 0.0705,
+      "step": 8887
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.365926556909043e-08,
+      "loss": 0.0731,
+      "step": 8888
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.32082917506305e-08,
+      "loss": 0.0598,
+      "step": 8889
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.275853165147452e-08,
+      "loss": 0.0617,
+      "step": 8890
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.230998532666735e-08,
+      "loss": 0.0697,
+      "step": 8891
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.186265283110728e-08,
+      "loss": 0.0688,
+      "step": 8892
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.141653421954387e-08,
+      "loss": 0.0747,
+      "step": 8893
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.097162954657678e-08,
+      "loss": 0.0662,
+      "step": 8894
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.05279388666591e-08,
+      "loss": 0.0658,
+      "step": 8895
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.008546223409409e-08,
+      "loss": 0.0671,
+      "step": 8896
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.964419970303506e-08,
+      "loss": 0.0678,
+      "step": 8897
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.920415132748993e-08,
+      "loss": 0.0697,
+      "step": 8898
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.876531716131674e-08,
+      "loss": 0.0628,
+      "step": 8899
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.832769725822364e-08,
+      "loss": 0.0659,
+      "step": 8900
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.789129167177111e-08,
+      "loss": 0.0702,
+      "step": 8901
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.745610045536978e-08,
+      "loss": 0.0669,
+      "step": 8902
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.702212366228479e-08,
+      "loss": 0.068,
+      "step": 8903
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.658936134563034e-08,
+      "loss": 0.0732,
+      "step": 8904
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.615781355837182e-08,
+      "loss": 0.065,
+      "step": 8905
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.5727480353327e-08,
+      "loss": 0.061,
+      "step": 8906
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.529836178316374e-08,
+      "loss": 0.0647,
+      "step": 8907
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.487045790040337e-08,
+      "loss": 0.0693,
+      "step": 8908
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.444376875741511e-08,
+      "loss": 0.0686,
+      "step": 8909
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.401829440642272e-08,
+      "loss": 0.0713,
+      "step": 8910
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.359403489950012e-08,
+      "loss": 0.0704,
+      "step": 8911
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.317099028857355e-08,
+      "loss": 0.0748,
+      "step": 8912
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.274916062541826e-08,
+      "loss": 0.0718,
+      "step": 8913
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.232854596166184e-08,
+      "loss": 0.0656,
+      "step": 8914
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.190914634878421e-08,
+      "loss": 0.07,
+      "step": 8915
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.149096183811654e-08,
+      "loss": 0.067,
+      "step": 8916
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.107399248083902e-08,
+      "loss": 0.0675,
+      "step": 8917
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.065823832798524e-08,
+      "loss": 0.0649,
+      "step": 8918
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.024369943044006e-08,
+      "loss": 0.069,
+      "step": 8919
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.983037583893848e-08,
+      "loss": 0.0744,
+      "step": 8920
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.941826760406778e-08,
+      "loss": 0.0634,
+      "step": 8921
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.90073747762643e-08,
+      "loss": 0.0723,
+      "step": 8922
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.859769740582001e-08,
+      "loss": 0.0644,
+      "step": 8923
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.818923554287371e-08,
+      "loss": 0.0603,
+      "step": 8924
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.778198923741764e-08,
+      "loss": 0.0737,
+      "step": 8925
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.737595853929524e-08,
+      "loss": 0.0742,
+      "step": 8926
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.6971143498199e-08,
+      "loss": 0.0652,
+      "step": 8927
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.656754416367705e-08,
+      "loss": 0.0713,
+      "step": 8928
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.616516058512323e-08,
+      "loss": 0.0741,
+      "step": 8929
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.576399281178813e-08,
+      "loss": 0.0699,
+      "step": 8930
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.536404089276915e-08,
+      "loss": 0.0686,
+      "step": 8931
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.496530487701602e-08,
+      "loss": 0.0693,
+      "step": 8932
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.456778481333081e-08,
+      "loss": 0.0765,
+      "step": 8933
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.417148075036794e-08,
+      "loss": 0.0668,
+      "step": 8934
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.37763927366275e-08,
+      "loss": 0.0644,
+      "step": 8935
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.338252082046859e-08,
+      "loss": 0.0604,
+      "step": 8936
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.298986505009375e-08,
+      "loss": 0.0681,
+      "step": 8937
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.259842547356231e-08,
+      "loss": 0.0698,
+      "step": 8938
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.220820213878153e-08,
+      "loss": 0.0695,
+      "step": 8939
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.181919509351209e-08,
+      "loss": 0.0672,
+      "step": 8940
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.143140438536366e-08,
+      "loss": 0.0682,
+      "step": 8941
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.10448300617994e-08,
+      "loss": 0.0711,
+      "step": 8942
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.065947217013146e-08,
+      "loss": 0.0627,
+      "step": 8943
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.027533075752213e-08,
+      "loss": 0.0664,
+      "step": 8944
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.989240587099043e-08,
+      "loss": 0.0692,
+      "step": 8945
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9510697557398914e-08,
+      "loss": 0.0713,
+      "step": 8946
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.913020586346685e-08,
+      "loss": 0.0702,
+      "step": 8947
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.875093083576366e-08,
+      "loss": 0.0715,
+      "step": 8948
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.837287252070667e-08,
+      "loss": 0.0654,
+      "step": 8949
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.799603096456774e-08,
+      "loss": 0.0693,
+      "step": 8950
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.762040621346887e-08,
+      "loss": 0.0639,
+      "step": 8951
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7245998313383286e-08,
+      "loss": 0.0696,
+      "step": 8952
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.687280731013212e-08,
+      "loss": 0.0713,
+      "step": 8953
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.650083324939437e-08,
+      "loss": 0.0673,
+      "step": 8954
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.613007617669364e-08,
+      "loss": 0.0683,
+      "step": 8955
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.576053613740695e-08,
+      "loss": 0.0662,
+      "step": 8956
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.539221317676369e-08,
+      "loss": 0.0672,
+      "step": 8957
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5025107339842234e-08,
+      "loss": 0.0676,
+      "step": 8958
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.46592186715722e-08,
+      "loss": 0.0709,
+      "step": 8959
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.429454721673555e-08,
+      "loss": 0.0737,
+      "step": 8960
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.393109301996546e-08,
+      "loss": 0.061,
+      "step": 8961
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.356885612574303e-08,
+      "loss": 0.0603,
+      "step": 8962
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3207836578403896e-08,
+      "loss": 0.068,
+      "step": 8963
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.284803442213271e-08,
+      "loss": 0.0616,
+      "step": 8964
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.248944970096648e-08,
+      "loss": 0.0614,
+      "step": 8965
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.213208245879231e-08,
+      "loss": 0.0637,
+      "step": 8966
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1775932739349667e-08,
+      "loss": 0.0701,
+      "step": 8967
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.142100058622479e-08,
+      "loss": 0.0696,
+      "step": 8968
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.106728604286071e-08,
+      "loss": 0.0661,
+      "step": 8969
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0714789152547236e-08,
+      "loss": 0.066,
+      "step": 8970
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0363509958427646e-08,
+      "loss": 0.0669,
+      "step": 8971
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.001344850349421e-08,
+      "loss": 0.0722,
+      "step": 8972
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.966460483059044e-08,
+      "loss": 0.0688,
+      "step": 8973
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9316978982414386e-08,
+      "loss": 0.0713,
+      "step": 8974
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8970571001507594e-08,
+      "loss": 0.066,
+      "step": 8975
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.862538093027169e-08,
+      "loss": 0.0606,
+      "step": 8976
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.828140881095067e-08,
+      "loss": 0.067,
+      "step": 8977
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.79386546856464e-08,
+      "loss": 0.0735,
+      "step": 8978
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.759711859630645e-08,
+      "loss": 0.0703,
+      "step": 8979
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.725680058473292e-08,
+      "loss": 0.0656,
+      "step": 8980
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.691770069257584e-08,
+      "loss": 0.067,
+      "step": 8981
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6579818961339784e-08,
+      "loss": 0.0582,
+      "step": 8982
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6243155432378336e-08,
+      "loss": 0.0685,
+      "step": 8983
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.590771014689299e-08,
+      "loss": 0.0674,
+      "step": 8984
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.557348314594201e-08,
+      "loss": 0.0628,
+      "step": 8985
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.524047447043045e-08,
+      "loss": 0.0734,
+      "step": 8986
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.49086841611146e-08,
+      "loss": 0.0648,
+      "step": 8987
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.457811225860309e-08,
+      "loss": 0.07,
+      "step": 8988
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4248758803355776e-08,
+      "loss": 0.0702,
+      "step": 8989
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.392062383568041e-08,
+      "loss": 0.0706,
+      "step": 8990
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.35937073957382e-08,
+      "loss": 0.0711,
+      "step": 8991
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.326800952354049e-08,
+      "loss": 0.0702,
+      "step": 8992
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.294353025895093e-08,
+      "loss": 0.0665,
+      "step": 8993
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.262026964167998e-08,
+      "loss": 0.0682,
+      "step": 8994
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.229822771129488e-08,
+      "loss": 0.0712,
+      "step": 8995
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.197740450720744e-08,
+      "loss": 0.0667,
+      "step": 8996
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1657800068684026e-08,
+      "loss": 0.0643,
+      "step": 8997
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.133941443484335e-08,
+      "loss": 0.0648,
+      "step": 8998
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1022247644649795e-08,
+      "loss": 0.0615,
+      "step": 8999
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0706299736923415e-08,
+      "loss": 0.0712,
+      "step": 9000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0391570750332176e-08,
+      "loss": 0.0706,
+      "step": 9001
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0078060723396375e-08,
+      "loss": 0.0674,
+      "step": 9002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9765769694486426e-08,
+      "loss": 0.0639,
+      "step": 9003
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9454697701822864e-08,
+      "loss": 0.0629,
+      "step": 9004
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.914484478347968e-08,
+      "loss": 0.0692,
+      "step": 9005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8836210977378776e-08,
+      "loss": 0.0728,
+      "step": 9006
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.852879632129436e-08,
+      "loss": 0.0685,
+      "step": 9007
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.822260085284968e-08,
+      "loss": 0.0683,
+      "step": 9008
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.791762460952253e-08,
+      "loss": 0.0674,
+      "step": 9009
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.76138676286375e-08,
+      "loss": 0.0634,
+      "step": 9010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.731132994737041e-08,
+      "loss": 0.0649,
+      "step": 9011
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.701001160275164e-08,
+      "loss": 0.0676,
+      "step": 9012
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.670991263165724e-08,
+      "loss": 0.0581,
+      "step": 9013
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6411033070818944e-08,
+      "loss": 0.0649,
+      "step": 9014
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.611337295681416e-08,
+      "loss": 0.0629,
+      "step": 9015
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.581693232607375e-08,
+      "loss": 0.0663,
+      "step": 9016
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5521711214880906e-08,
+      "loss": 0.0744,
+      "step": 9017
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5227709659367834e-08,
+      "loss": 0.0715,
+      "step": 9018
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.493492769551576e-08,
+      "loss": 0.0694,
+      "step": 9019
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4643365359158244e-08,
+      "loss": 0.0686,
+      "step": 9020
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.435302268598228e-08,
+      "loss": 0.0689,
+      "step": 9021
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.406389971152058e-08,
+      "loss": 0.0674,
+      "step": 9022
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3775996471160366e-08,
+      "loss": 0.0681,
+      "step": 9023
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.34893130001368e-08,
+      "loss": 0.0609,
+      "step": 9024
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.32038493335396e-08,
+      "loss": 0.0718,
+      "step": 9025
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.291960550630302e-08,
+      "loss": 0.07,
+      "step": 9026
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.263658155322036e-08,
+      "loss": 0.0683,
+      "step": 9027
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.235477750892613e-08,
+      "loss": 0.0666,
+      "step": 9028
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.207419340791495e-08,
+      "loss": 0.0705,
+      "step": 9029
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.179482928452382e-08,
+      "loss": 0.0717,
+      "step": 9030
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1516685172946485e-08,
+      "loss": 0.0668,
+      "step": 9031
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.12397611072246e-08,
+      "loss": 0.0631,
+      "step": 9032
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.096405712125106e-08,
+      "loss": 0.0666,
+      "step": 9033
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.068957324876887e-08,
+      "loss": 0.0662,
+      "step": 9034
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.041630952337227e-08,
+      "loss": 0.0677,
+      "step": 9035
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0144265978506725e-08,
+      "loss": 0.0682,
+      "step": 9036
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.987344264746783e-08,
+      "loss": 0.0711,
+      "step": 9037
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9603839563400183e-08,
+      "loss": 0.0695,
+      "step": 9038
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.933545675930294e-08,
+      "loss": 0.0738,
+      "step": 9039
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9068294268020935e-08,
+      "loss": 0.0711,
+      "step": 9040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8802352122254686e-08,
+      "loss": 0.0727,
+      "step": 9041
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8537630354551483e-08,
+      "loss": 0.0603,
+      "step": 9042
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.827412899731097e-08,
+      "loss": 0.0691,
+      "step": 9043
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.80118480827829e-08,
+      "loss": 0.0707,
+      "step": 9044
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7750787643068267e-08,
+      "loss": 0.0681,
+      "step": 9045
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7490947710119287e-08,
+      "loss": 0.0683,
+      "step": 9046
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7232328315736078e-08,
+      "loss": 0.0678,
+      "step": 9047
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.697492949157221e-08,
+      "loss": 0.0699,
+      "step": 9048
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.671875126913137e-08,
+      "loss": 0.0649,
+      "step": 9049
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.646379367976626e-08,
+      "loss": 0.0659,
+      "step": 9050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6210056754680802e-08,
+      "loss": 0.0682,
+      "step": 9051
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5957540524931267e-08,
+      "loss": 0.0634,
+      "step": 9052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5706245021421828e-08,
+      "loss": 0.0684,
+      "step": 9053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5456170274910096e-08,
+      "loss": 0.0632,
+      "step": 9054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5207316316001596e-08,
+      "loss": 0.0635,
+      "step": 9055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4959683175154182e-08,
+      "loss": 0.0655,
+      "step": 9056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4713270882675833e-08,
+      "loss": 0.0706,
+      "step": 9057
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4468079468724647e-08,
+      "loss": 0.0657,
+      "step": 9058
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4224108963309955e-08,
+      "loss": 0.071,
+      "step": 9059
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3981359396292315e-08,
+      "loss": 0.0663,
+      "step": 9060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.373983079737907e-08,
+      "loss": 0.0766,
+      "step": 9061
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3499523196134356e-08,
+      "loss": 0.0671,
+      "step": 9062
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3260436621966864e-08,
+      "loss": 0.0631,
+      "step": 9063
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3022571104140966e-08,
+      "loss": 0.0723,
+      "step": 9064
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2785926671766712e-08,
+      "loss": 0.0671,
+      "step": 9065
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2550503353808704e-08,
+      "loss": 0.0664,
+      "step": 9066
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2316301179079457e-08,
+      "loss": 0.0719,
+      "step": 9067
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2083320176242707e-08,
+      "loss": 0.0715,
+      "step": 9068
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.185156037381564e-08,
+      "loss": 0.0718,
+      "step": 9069
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1621021800160012e-08,
+      "loss": 0.0695,
+      "step": 9070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.139170448349437e-08,
+      "loss": 0.0672,
+      "step": 9071
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1163608451882923e-08,
+      "loss": 0.0624,
+      "step": 9072
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.093673373324334e-08,
+      "loss": 0.0668,
+      "step": 9073
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0711080355343416e-08,
+      "loss": 0.0649,
+      "step": 9074
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0486648345798832e-08,
+      "loss": 0.0702,
+      "step": 9075
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0263437732079837e-08,
+      "loss": 0.0704,
+      "step": 9076
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0041448541505693e-08,
+      "loss": 0.0641,
+      "step": 9077
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9820680801243554e-08,
+      "loss": 0.072,
+      "step": 9078
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9601134538315138e-08,
+      "loss": 0.0724,
+      "step": 9079
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.938280977959006e-08,
+      "loss": 0.0671,
+      "step": 9080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9165706551789174e-08,
+      "loss": 0.072,
+      "step": 9081
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8949824881483446e-08,
+      "loss": 0.064,
+      "step": 9082
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8735164795095072e-08,
+      "loss": 0.0627,
+      "step": 9083
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8521726318897483e-08,
+      "loss": 0.0695,
+      "step": 9084
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8309509479010888e-08,
+      "loss": 0.0729,
+      "step": 9085
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8098514301411185e-08,
+      "loss": 0.0668,
+      "step": 9086
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7888740811921045e-08,
+      "loss": 0.0604,
+      "step": 9087
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.768018903621438e-08,
+      "loss": 0.0695,
+      "step": 9088
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.747285899981632e-08,
+      "loss": 0.06,
+      "step": 9089
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7266750728101023e-08,
+      "loss": 0.0709,
+      "step": 9090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.706186424629608e-08,
+      "loss": 0.0662,
+      "step": 9091
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6858199579476986e-08,
+      "loss": 0.0615,
+      "step": 9092
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6655756752569363e-08,
+      "loss": 0.0642,
+      "step": 9093
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6454535790351168e-08,
+      "loss": 0.0638,
+      "step": 9094
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6254536717449364e-08,
+      "loss": 0.0631,
+      "step": 9095
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6055759558342153e-08,
+      "loss": 0.0685,
+      "step": 9096
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.585820433735785e-08,
+      "loss": 0.0671,
+      "step": 9097
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5661871078674895e-08,
+      "loss": 0.0696,
+      "step": 9098
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.546675980632295e-08,
+      "loss": 0.0694,
+      "step": 9099
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5272870544182917e-08,
+      "loss": 0.0731,
+      "step": 9100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5080203315983588e-08,
+      "loss": 0.0645,
+      "step": 9101
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4888758145304994e-08,
+      "loss": 0.0751,
+      "step": 9102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.46985350555795e-08,
+      "loss": 0.067,
+      "step": 9103
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4509534070088482e-08,
+      "loss": 0.067,
+      "step": 9104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.432175521196344e-08,
+      "loss": 0.072,
+      "step": 9105
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4135198504185988e-08,
+      "loss": 0.0681,
+      "step": 9106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3949863969590083e-08,
+      "loss": 0.0693,
+      "step": 9107
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3765751630858693e-08,
+      "loss": 0.0698,
+      "step": 9108
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3582861510524903e-08,
+      "loss": 0.0668,
+      "step": 9109
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3401193630971921e-08,
+      "loss": 0.0653,
+      "step": 9110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3220748014436402e-08,
+      "loss": 0.0709,
+      "step": 9111
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3041524683000683e-08,
+      "loss": 0.0642,
+      "step": 9112
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2863523658602772e-08,
+      "loss": 0.0639,
+      "step": 9113
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2686744963024133e-08,
+      "loss": 0.0713,
+      "step": 9114
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2511188617905235e-08,
+      "loss": 0.071,
+      "step": 9115
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2336854644730002e-08,
+      "loss": 0.0677,
+      "step": 9116
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2163743064834699e-08,
+      "loss": 0.0704,
+      "step": 9117
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1991853899409044e-08,
+      "loss": 0.0694,
+      "step": 9118
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1821187169488434e-08,
+      "loss": 0.0682,
+      "step": 9119
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1651742895960606e-08,
+      "loss": 0.065,
+      "step": 9120
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1483521099565631e-08,
+      "loss": 0.0762,
+      "step": 9121
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1316521800891488e-08,
+      "loss": 0.077,
+      "step": 9122
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1150745020376274e-08,
+      "loss": 0.0742,
+      "step": 9123
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0986190778311534e-08,
+      "loss": 0.0691,
+      "step": 9124
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0822859094834492e-08,
+      "loss": 0.0743,
+      "step": 9125
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0660749989938046e-08,
+      "loss": 0.0702,
+      "step": 9126
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0499863483460771e-08,
+      "loss": 0.0652,
+      "step": 9127
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0340199595093581e-08,
+      "loss": 0.0681,
+      "step": 9128
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0181758344379733e-08,
+      "loss": 0.0686,
+      "step": 9129
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0024539750709273e-08,
+      "loss": 0.0662,
+      "step": 9130
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.868543833325695e-09,
+      "loss": 0.0657,
+      "step": 9131
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.713770611319285e-09,
+      "loss": 0.0652,
+      "step": 9132
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.560220103633778e-09,
+      "loss": 0.0731,
+      "step": 9133
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.407892329063028e-09,
+      "loss": 0.0713,
+      "step": 9134
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.256787306249904e-09,
+      "loss": 0.0714,
+      "step": 9135
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.106905053687386e-09,
+      "loss": 0.0678,
+      "step": 9136
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.958245589720804e-09,
+      "loss": 0.069,
+      "step": 9137
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.81080893254449e-09,
+      "loss": 0.0651,
+      "step": 9138
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.66459510020179e-09,
+      "loss": 0.0643,
+      "step": 9139
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.519604110590608e-09,
+      "loss": 0.0694,
+      "step": 9140
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.375835981453417e-09,
+      "loss": 0.0672,
+      "step": 9141
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.233290730388366e-09,
+      "loss": 0.0693,
+      "step": 9142
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.091968374840387e-09,
+      "loss": 0.0683,
+      "step": 9143
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.951868932106754e-09,
+      "loss": 0.0679,
+      "step": 9144
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.812992419333753e-09,
+      "loss": 0.0674,
+      "step": 9145
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.675338853517788e-09,
+      "loss": 0.0702,
+      "step": 9146
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.538908251507605e-09,
+      "loss": 0.0675,
+      "step": 9147
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.403700630000954e-09,
+      "loss": 0.0733,
+      "step": 9148
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.269716005545713e-09,
+      "loss": 0.0632,
+      "step": 9149
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.136954394538764e-09,
+      "loss": 0.0706,
+      "step": 9150
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.005415813231553e-09,
+      "loss": 0.0646,
+      "step": 9151
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.875100277721203e-09,
+      "loss": 0.069,
+      "step": 9152
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.746007803958288e-09,
+      "loss": 0.0668,
+      "step": 9153
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.618138407740171e-09,
+      "loss": 0.0688,
+      "step": 9154
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.491492104719887e-09,
+      "loss": 0.0704,
+      "step": 9155
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.36606891039615e-09,
+      "loss": 0.0652,
+      "step": 9156
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.241868840118903e-09,
+      "loss": 0.0695,
+      "step": 9157
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.11889190909043e-09,
+      "loss": 0.0666,
+      "step": 9158
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.997138132362024e-09,
+      "loss": 0.0714,
+      "step": 9159
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.876607524833988e-09,
+      "loss": 0.0724,
+      "step": 9160
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.757300101257857e-09,
+      "loss": 0.0701,
+      "step": 9161
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6392158762375025e-09,
+      "loss": 0.0673,
+      "step": 9162
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.522354864223589e-09,
+      "loss": 0.0698,
+      "step": 9163
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.406717079519119e-09,
+      "loss": 0.0667,
+      "step": 9164
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.292302536277216e-09,
+      "loss": 0.066,
+      "step": 9165
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1791112485022335e-09,
+      "loss": 0.0644,
+      "step": 9166
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.067143230045313e-09,
+      "loss": 0.0681,
+      "step": 9167
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.956398494613268e-09,
+      "loss": 0.0712,
+      "step": 9168
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8468770557574815e-09,
+      "loss": 0.0614,
+      "step": 9169
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.738578926882787e-09,
+      "loss": 0.0671,
+      "step": 9170
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.631504121245245e-09,
+      "loss": 0.0685,
+      "step": 9171
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.52565265194882e-09,
+      "loss": 0.069,
+      "step": 9172
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.421024531948703e-09,
+      "loss": 0.0685,
+      "step": 9173
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.317619774051318e-09,
+      "loss": 0.065,
+      "step": 9174
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2154383909109866e-09,
+      "loss": 0.0678,
+      "step": 9175
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.11448039503326e-09,
+      "loss": 0.0649,
+      "step": 9176
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0147457987771424e-09,
+      "loss": 0.0678,
+      "step": 9177
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.916234614346204e-09,
+      "loss": 0.0673,
+      "step": 9178
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.818946853799688e-09,
+      "loss": 0.0705,
+      "step": 9179
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.722882529042515e-09,
+      "loss": 0.0688,
+      "step": 9180
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6280416518330586e-09,
+      "loss": 0.0658,
+      "step": 9181
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5344242337798096e-09,
+      "loss": 0.0654,
+      "step": 9182
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.44203028633916e-09,
+      "loss": 0.0673,
+      "step": 9183
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.350859820819841e-09,
+      "loss": 0.0716,
+      "step": 9184
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.260912848380704e-09,
+      "loss": 0.0706,
+      "step": 9185
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.172189380029611e-09,
+      "loss": 0.0726,
+      "step": 9186
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0846894266256534e-09,
+      "loss": 0.0693,
+      "step": 9187
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.998412998878042e-09,
+      "loss": 0.0627,
+      "step": 9188
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9133601073461083e-09,
+      "loss": 0.0704,
+      "step": 9189
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.829530762439303e-09,
+      "loss": 0.0651,
+      "step": 9190
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7469249744183078e-09,
+      "loss": 0.0698,
+      "step": 9191
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6655427533928136e-09,
+      "loss": 0.0742,
+      "step": 9192
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.585384109322631e-09,
+      "loss": 0.0695,
+      "step": 9193
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5064490520199104e-09,
+      "loss": 0.0685,
+      "step": 9194
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4287375911447032e-09,
+      "loss": 0.0701,
+      "step": 9195
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3522497362071793e-09,
+      "loss": 0.0641,
+      "step": 9196
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.27698549656985e-09,
+      "loss": 0.0672,
+      "step": 9197
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2029448814431254e-09,
+      "loss": 0.0653,
+      "step": 9198
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.130127899890866e-09,
+      "loss": 0.0661,
+      "step": 9199
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0585345608237217e-09,
+      "loss": 0.0754,
+      "step": 9200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9881648730046834e-09,
+      "loss": 0.067,
+      "step": 9201
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9190188450457505e-09,
+      "loss": 0.0702,
+      "step": 9202
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8510964854101533e-09,
+      "loss": 0.0657,
+      "step": 9203
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.784397802411242e-09,
+      "loss": 0.0662,
+      "step": 9204
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7189228042102657e-09,
+      "loss": 0.0651,
+      "step": 9205
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6546714988241453e-09,
+      "loss": 0.0697,
+      "step": 9206
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5916438941132595e-09,
+      "loss": 0.0681,
+      "step": 9207
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5298399977936584e-09,
+      "loss": 0.0704,
+      "step": 9208
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4692598174292916e-09,
+      "loss": 0.0686,
+      "step": 9209
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4099033604331181e-09,
+      "loss": 0.0646,
+      "step": 9210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3517706340726578e-09,
+      "loss": 0.062,
+      "step": 9211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2948616454599993e-09,
+      "loss": 0.0737,
+      "step": 9212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2391764015606821e-09,
+      "loss": 0.0743,
+      "step": 9213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1847149091914755e-09,
+      "loss": 0.0645,
+      "step": 9214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1314771750170484e-09,
+      "loss": 0.066,
+      "step": 9215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0794632055532995e-09,
+      "loss": 0.0666,
+      "step": 9216
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0286730071662476e-09,
+      "loss": 0.069,
+      "step": 9217
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.791065860720316e-10,
+      "loss": 0.0714,
+      "step": 9218
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.307639483369102e-10,
+      "loss": 0.0623,
+      "step": 9219
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.836450998783719e-10,
+      "loss": 0.0688,
+      "step": 9220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.377500464618049e-10,
+      "loss": 0.0658,
+      "step": 9221
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.93078793706048e-10,
+      "loss": 0.0688,
+      "step": 9222
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.496313470778394e-10,
+      "loss": 0.0754,
+      "step": 9223
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.074077118929268e-10,
+      "loss": 0.0638,
+      "step": 9224
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.664078933227292e-10,
+      "loss": 0.0645,
+      "step": 9225
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.266318963821239e-10,
+      "loss": 0.0703,
+      "step": 9226
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.880797259394389e-10,
+      "loss": 0.0751,
+      "step": 9227
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.507513867153425e-10,
+      "loss": 0.0636,
+      "step": 9228
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.146468832761819e-10,
+      "loss": 0.066,
+      "step": 9229
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.797662200428654e-10,
+      "loss": 0.0761,
+      "step": 9230
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.461094012830902e-10,
+      "loss": 0.0719,
+      "step": 9231
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.136764311157837e-10,
+      "loss": 0.0633,
+      "step": 9232
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8246731351110344e-10,
+      "loss": 0.0614,
+      "step": 9233
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.5248205228821664e-10,
+      "loss": 0.0647,
+      "step": 9234
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.237206511175206e-10,
+      "loss": 0.066,
+      "step": 9235
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.9618311351953254e-10,
+      "loss": 0.0722,
+      "step": 9236
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6986944286377937e-10,
+      "loss": 0.0635,
+      "step": 9237
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.4477964237212823e-10,
+      "loss": 0.0676,
+      "step": 9238
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2091371511323567e-10,
+      "loss": 0.0699,
+      "step": 9239
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.982716640092086e-10,
+      "loss": 0.0716,
+      "step": 9240
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7685349183116374e-10,
+      "loss": 0.0577,
+      "step": 9241
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.5665920120033762e-10,
+      "loss": 0.0675,
+      "step": 9242
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3768879458919693e-10,
+      "loss": 0.0668,
+      "step": 9243
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.199422743181078e-10,
+      "loss": 0.0705,
+      "step": 9244
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0341964255977666e-10,
+      "loss": 0.0694,
+      "step": 9245
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.812090133702988e-11,
+      "loss": 0.067,
+      "step": 9246
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.404605252059327e-11,
+      "loss": 0.0635,
+      "step": 9247
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.119509783464316e-11,
+      "loss": 0.0676,
+      "step": 9248
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.956803885236561e-11,
+      "loss": 0.066,
+      "step": 9249
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.916487699484606e-11,
+      "loss": 0.0681,
+      "step": 9250
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.998561353662055e-11,
+      "loss": 0.0675,
+      "step": 9251
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.203024960123479e-11,
+      "loss": 0.0705,
+      "step": 9252
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5298786161244138e-11,
+      "loss": 0.0706,
+      "step": 9253
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.791224041544311e-12,
+      "loss": 0.0582,
+      "step": 9254
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.507563917150904e-12,
+      "loss": 0.0728,
+      "step": 9255
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.4478063098687386e-12,
+      "loss": 0.0679,
+      "step": 9256
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.119515960634204e-13,
+      "loss": 0.0656,
+      "step": 9257
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.0656,
+      "step": 9258
+    },
+    {
+      "epoch": 3.0,
+      "step": 9258,
+      "total_flos": 2.02957017742259e+18,
+      "train_loss": 0.12219694624321974,
+      "train_runtime": 29392.6601,
+      "train_samples_per_second": 40.316,
+      "train_steps_per_second": 0.315
+    }
+  ],
+  "max_steps": 9258,
+  "num_train_epochs": 3,
+  "total_flos": 2.02957017742259e+18,
+  "trial_name": null,
+  "trial_params": null
+}