diff --git "a/multimodal-vicuna-7b-v1.5-point-naivemc/trainer_state.json" "b/multimodal-vicuna-7b-v1.5-point-naivemc/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/multimodal-vicuna-7b-v1.5-point-naivemc/trainer_state.json"
@@ -0,0 +1,39553 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 6588,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0101010101010103e-07,
+      "loss": 2.9355,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0202020202020205e-07,
+      "loss": 2.8047,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 2.6426,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 2.8281,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.05050505050505e-07,
+      "loss": 2.998,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 2.8906,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.070707070707071e-07,
+      "loss": 2.7031,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.080808080808082e-07,
+      "loss": 2.832,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 2.5938,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 2.7285,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.111111111111111e-06,
+      "loss": 2.7559,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 2.4668,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3131313131313134e-06,
+      "loss": 2.6523,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 2.7363,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 2.6328,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6161616161616164e-06,
+      "loss": 2.8164,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7171717171717173e-06,
+      "loss": 2.6924,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 2.6172,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9191919191919192e-06,
+      "loss": 2.748,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 2.6348,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 2.7832,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 2.7988,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3232323232323234e-06,
+      "loss": 2.8262,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 2.4062,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5252525252525258e-06,
+      "loss": 2.5986,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6262626262626267e-06,
+      "loss": 2.7148,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 2.5664,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8282828282828286e-06,
+      "loss": 2.4346,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9292929292929295e-06,
+      "loss": 2.3086,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 2.4297,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.131313131313132e-06,
+      "loss": 2.3516,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.232323232323233e-06,
+      "loss": 2.4482,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 2.3398,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 2.4844,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5353535353535356e-06,
+      "loss": 2.1777,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 2.124,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.737373737373738e-06,
+      "loss": 2.2354,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8383838383838385e-06,
+      "loss": 1.999,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 2.123,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 1.8857,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.141414141414142e-06,
+      "loss": 2.0537,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 2.0254,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.343434343434344e-06,
+      "loss": 1.748,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.9326,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.9355,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.646464646464647e-06,
+      "loss": 1.9648,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.747474747474748e-06,
+      "loss": 1.8154,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 1.957,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.94949494949495e-06,
+      "loss": 1.9648,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0505050505050515e-06,
+      "loss": 1.7939,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 1.7744,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.252525252525253e-06,
+      "loss": 1.791,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.353535353535354e-06,
+      "loss": 1.9355,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.8105,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.8574,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.656565656565657e-06,
+      "loss": 1.7725,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 1.7656,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.858585858585859e-06,
+      "loss": 1.7461,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.95959595959596e-06,
+      "loss": 1.9062,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.7441,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.1616161616161615e-06,
+      "loss": 1.7598,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.262626262626264e-06,
+      "loss": 1.709,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 1.8682,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.464646464646466e-06,
+      "loss": 1.7529,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.565656565656566e-06,
+      "loss": 1.7959,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.7588,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.767676767676769e-06,
+      "loss": 1.8994,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.868686868686869e-06,
+      "loss": 1.5605,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 1.5063,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.070707070707071e-06,
+      "loss": 1.6963,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.171717171717172e-06,
+      "loss": 1.7998,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.6064,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.373737373737374e-06,
+      "loss": 1.7217,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.474747474747476e-06,
+      "loss": 1.5537,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 1.8232,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.676767676767677e-06,
+      "loss": 1.7275,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.5605,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 1.9131,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.97979797979798e-06,
+      "loss": 1.7129,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.08080808080808e-06,
+      "loss": 1.6504,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.5498,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.282828282828283e-06,
+      "loss": 1.8281,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.383838383838384e-06,
+      "loss": 1.6836,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.6221,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.585858585858587e-06,
+      "loss": 1.5938,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.686868686868687e-06,
+      "loss": 1.5342,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.6582,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.833,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.98989898989899e-06,
+      "loss": 1.4917,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.6367,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.191919191919193e-06,
+      "loss": 1.8115,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.292929292929294e-06,
+      "loss": 1.5693,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 1.5947,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.494949494949497e-06,
+      "loss": 1.7432,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.595959595959597e-06,
+      "loss": 1.3789,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 1.9316,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.797979797979798e-06,
+      "loss": 1.6875,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8989898989899e-06,
+      "loss": 1.624,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1e-05,
+      "loss": 1.6279,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0101010101010103e-05,
+      "loss": 1.6963,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0202020202020202e-05,
+      "loss": 1.5615,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 1.6807,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0404040404040405e-05,
+      "loss": 1.5386,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0505050505050507e-05,
+      "loss": 1.5537,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0606060606060606e-05,
+      "loss": 1.8135,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0707070707070708e-05,
+      "loss": 1.6553,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.080808080808081e-05,
+      "loss": 1.8242,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 1.5713,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1010101010101011e-05,
+      "loss": 1.666,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.6299,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1212121212121212e-05,
+      "loss": 1.6533,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1313131313131314e-05,
+      "loss": 1.5791,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1414141414141415e-05,
+      "loss": 1.5586,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 1.4854,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1616161616161616e-05,
+      "loss": 1.6758,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1717171717171718e-05,
+      "loss": 1.5161,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.181818181818182e-05,
+      "loss": 1.6914,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191919191919192e-05,
+      "loss": 1.6709,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2020202020202021e-05,
+      "loss": 1.5771,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 1.5518,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2222222222222224e-05,
+      "loss": 1.4106,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2323232323232323e-05,
+      "loss": 1.4922,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2424242424242425e-05,
+      "loss": 1.7002,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2525252525252527e-05,
+      "loss": 1.4717,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2626262626262626e-05,
+      "loss": 1.5186,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 1.6328,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2828282828282829e-05,
+      "loss": 1.5479,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2929292929292931e-05,
+      "loss": 1.6758,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3030303030303032e-05,
+      "loss": 1.5469,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3131313131313132e-05,
+      "loss": 1.4395,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3232323232323234e-05,
+      "loss": 1.6006,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.5938,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3434343434343436e-05,
+      "loss": 1.6143,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3535353535353538e-05,
+      "loss": 1.4648,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3636363636363637e-05,
+      "loss": 1.5781,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3737373737373739e-05,
+      "loss": 1.4492,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.383838383838384e-05,
+      "loss": 1.459,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 1.6641,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.404040404040404e-05,
+      "loss": 1.5049,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4141414141414143e-05,
+      "loss": 1.6421,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4242424242424245e-05,
+      "loss": 1.6514,
+      "step": 141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4343434343434344e-05,
+      "loss": 1.4512,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4444444444444446e-05,
+      "loss": 1.6396,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 1.6924,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4646464646464649e-05,
+      "loss": 1.4604,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4747474747474747e-05,
+      "loss": 1.5684,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.484848484848485e-05,
+      "loss": 1.4736,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4949494949494952e-05,
+      "loss": 1.6206,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.505050505050505e-05,
+      "loss": 1.415,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 1.7124,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5252525252525255e-05,
+      "loss": 1.6865,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5353535353535354e-05,
+      "loss": 1.6016,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5454545454545454e-05,
+      "loss": 1.7061,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.6465,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.565656565656566e-05,
+      "loss": 1.5625,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 1.6016,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.585858585858586e-05,
+      "loss": 1.5723,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.595959595959596e-05,
+      "loss": 1.5791,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.606060606060606e-05,
+      "loss": 1.6797,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.616161616161616e-05,
+      "loss": 1.585,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6262626262626265e-05,
+      "loss": 1.54,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 1.3945,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6464646464646466e-05,
+      "loss": 1.6904,
+      "step": 163
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6565656565656567e-05,
+      "loss": 1.3955,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 1.626,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6767676767676768e-05,
+      "loss": 1.6455,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.686868686868687e-05,
+      "loss": 1.5693,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 1.4932,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.707070707070707e-05,
+      "loss": 1.4568,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7171717171717173e-05,
+      "loss": 1.6758,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7272727272727274e-05,
+      "loss": 1.5762,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7373737373737375e-05,
+      "loss": 1.4167,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7474747474747475e-05,
+      "loss": 1.457,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 1.3354,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.767676767676768e-05,
+      "loss": 1.4902,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 1.6572,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.787878787878788e-05,
+      "loss": 1.2915,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.797979797979798e-05,
+      "loss": 1.6084,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8080808080808085e-05,
+      "loss": 1.6836,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 1.4316,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8282828282828286e-05,
+      "loss": 1.3477,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8383838383838387e-05,
+      "loss": 1.6719,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8484848484848487e-05,
+      "loss": 1.5098,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8585858585858588e-05,
+      "loss": 1.4854,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8686868686868688e-05,
+      "loss": 1.3848,
+      "step": 185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 1.5879,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.888888888888889e-05,
+      "loss": 1.5869,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8989898989898993e-05,
+      "loss": 1.6528,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9090909090909094e-05,
+      "loss": 1.4468,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9191919191919194e-05,
+      "loss": 1.5889,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9292929292929295e-05,
+      "loss": 1.5859,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 1.6543,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9494949494949496e-05,
+      "loss": 1.5049,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9595959595959596e-05,
+      "loss": 1.5059,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.96969696969697e-05,
+      "loss": 1.373,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97979797979798e-05,
+      "loss": 1.5615,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98989898989899e-05,
+      "loss": 1.46,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 1.4492,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998791440535e-05,
+      "loss": 1.6924,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995165762428e-05,
+      "loss": 1.5752,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999989122966553e-05,
+      "loss": 1.5166,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999980663054373e-05,
+      "loss": 1.4097,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999969786027934e-05,
+      "loss": 1.625,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999956491889865e-05,
+      "loss": 1.4883,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999940780643376e-05,
+      "loss": 1.5078,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999922652292268e-05,
+      "loss": 1.4502,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999990210684092e-05,
+      "loss": 1.4854,
+      "step": 207
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999879144294306e-05,
+      "loss": 1.5771,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999853764657966e-05,
+      "loss": 1.3164,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982596793804e-05,
+      "loss": 1.6201,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999795754141247e-05,
+      "loss": 1.4883,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999763123274888e-05,
+      "loss": 1.3706,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999728075346848e-05,
+      "loss": 1.6006,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999690610365603e-05,
+      "loss": 1.5361,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999650728340207e-05,
+      "loss": 1.5352,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999608429280302e-05,
+      "loss": 1.457,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999956371319611e-05,
+      "loss": 1.3496,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951658009844e-05,
+      "loss": 1.4971,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999467029998682e-05,
+      "loss": 1.522,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999415062908816e-05,
+      "loss": 1.4678,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99993606788414e-05,
+      "loss": 1.3545,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999303877809586e-05,
+      "loss": 1.3926,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99992446598271e-05,
+      "loss": 1.7178,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999918302490825e-05,
+      "loss": 1.4971,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999118973067942e-05,
+      "loss": 1.3467,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999052504321653e-05,
+      "loss": 1.3462,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999898361868545e-05,
+      "loss": 1.4644,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999891231617599e-05,
+      "loss": 1.415,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998838596810497e-05,
+      "loss": 1.5059,
+      "step": 229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998762460606795e-05,
+      "loss": 1.4805,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998683907583293e-05,
+      "loss": 1.4727,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998602937758965e-05,
+      "loss": 1.5615,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998519551153392e-05,
+      "loss": 1.5684,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999843374778673e-05,
+      "loss": 1.4863,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999834552767971e-05,
+      "loss": 1.416,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998254890853667e-05,
+      "loss": 1.6396,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99981618373305e-05,
+      "loss": 1.5557,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998066367132707e-05,
+      "loss": 1.6113,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999796848028336e-05,
+      "loss": 1.4316,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997868176806127e-05,
+      "loss": 1.582,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999776545672524e-05,
+      "loss": 1.415,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999766032006554e-05,
+      "loss": 1.4434,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997552766852434e-05,
+      "loss": 1.418,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997442797111913e-05,
+      "loss": 1.5098,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999733041087057e-05,
+      "loss": 1.5127,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997215608155565e-05,
+      "loss": 1.4175,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997098388994646e-05,
+      "loss": 1.5039,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996978753416147e-05,
+      "loss": 1.5078,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996856701448985e-05,
+      "loss": 1.5439,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996732233122663e-05,
+      "loss": 1.5801,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996605348467262e-05,
+      "loss": 1.6436,
+      "step": 251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996476047513454e-05,
+      "loss": 1.4346,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996344330292496e-05,
+      "loss": 1.5723,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996210196836224e-05,
+      "loss": 1.54,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996073647177058e-05,
+      "loss": 1.5986,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995934681348003e-05,
+      "loss": 1.627,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999579329938265e-05,
+      "loss": 1.5283,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995649501315172e-05,
+      "loss": 1.5869,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995503287180327e-05,
+      "loss": 1.5537,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995354657013458e-05,
+      "loss": 1.4395,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999520361085049e-05,
+      "loss": 1.4932,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999505014872793e-05,
+      "loss": 1.7051,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994894270682877e-05,
+      "loss": 1.5527,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994735976753007e-05,
+      "loss": 1.4727,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994575266976577e-05,
+      "loss": 1.5347,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994412141392437e-05,
+      "loss": 1.5352,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994246600040012e-05,
+      "loss": 1.3721,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999407864295932e-05,
+      "loss": 1.3726,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 1.4893,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993735481776105e-05,
+      "loss": 1.6387,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999356027775653e-05,
+      "loss": 1.3057,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993382658174573e-05,
+      "loss": 1.4961,
+      "step": 272
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993202623073173e-05,
+      "loss": 1.4746,
+      "step": 273
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999302017249585e-05,
+      "loss": 1.4639,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992835306486693e-05,
+      "loss": 1.5488,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99926480250904e-05,
+      "loss": 1.5781,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992458328352228e-05,
+      "loss": 1.395,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992266216318037e-05,
+      "loss": 1.4971,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992071689034257e-05,
+      "loss": 1.6162,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991874746547914e-05,
+      "loss": 1.5693,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991675388906608e-05,
+      "loss": 1.3491,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999147361615852e-05,
+      "loss": 1.5181,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999126942835243e-05,
+      "loss": 1.353,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999106282553769e-05,
+      "loss": 1.3374,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990853807764236e-05,
+      "loss": 1.5049,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999064237508259e-05,
+      "loss": 1.4624,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999042852754386e-05,
+      "loss": 1.3887,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990212265199738e-05,
+      "loss": 1.4395,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998999358810249e-05,
+      "loss": 1.5005,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998977249630498e-05,
+      "loss": 1.4658,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989548989860643e-05,
+      "loss": 1.4771,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998932306882351e-05,
+      "loss": 1.6055,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989094733248177e-05,
+      "loss": 1.4443,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988863983189848e-05,
+      "loss": 1.4106,
+      "step": 294
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998863081870429e-05,
+      "loss": 1.5166,
+      "step": 295
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988395239847865e-05,
+      "loss": 1.416,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988157246677514e-05,
+      "loss": 1.5391,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987916839250764e-05,
+      "loss": 1.5293,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987674017625726e-05,
+      "loss": 1.6797,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987428781861088e-05,
+      "loss": 1.4453,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987181132016128e-05,
+      "loss": 1.4351,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998693106815071e-05,
+      "loss": 1.4727,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986678590325273e-05,
+      "loss": 1.4517,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986423698600844e-05,
+      "loss": 1.4668,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986166393039038e-05,
+      "loss": 1.4829,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998590667370204e-05,
+      "loss": 1.4893,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985644540652637e-05,
+      "loss": 1.4526,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985379993954182e-05,
+      "loss": 1.3613,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985113033670625e-05,
+      "loss": 1.5527,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984843659866492e-05,
+      "loss": 1.2979,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998457187260689e-05,
+      "loss": 1.4814,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984297671957518e-05,
+      "loss": 1.6064,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998402105798465e-05,
+      "loss": 1.5039,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983742030755148e-05,
+      "loss": 1.5405,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983460590336458e-05,
+      "loss": 1.3032,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983176736796606e-05,
+      "loss": 1.4932,
+      "step": 316
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982890470204205e-05,
+      "loss": 1.5488,
+      "step": 317
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998260179062844e-05,
+      "loss": 1.5312,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99823106981391e-05,
+      "loss": 1.4346,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998201719280654e-05,
+      "loss": 1.4141,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981721274701705e-05,
+      "loss": 1.5527,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998142294389612e-05,
+      "loss": 1.4717,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981122200461898e-05,
+      "loss": 1.5967,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980819044471732e-05,
+      "loss": 1.4907,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980513475998894e-05,
+      "loss": 1.5283,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998020549511725e-05,
+      "loss": 1.5596,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979895101901236e-05,
+      "loss": 1.7227,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997958229642588e-05,
+      "loss": 1.48,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979267078766797e-05,
+      "loss": 1.5605,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978949449000168e-05,
+      "loss": 1.6992,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997862940720278e-05,
+      "loss": 1.5215,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997830695345198e-05,
+      "loss": 1.5029,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977982087825714e-05,
+      "loss": 1.5312,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977654810402504e-05,
+      "loss": 1.4248,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977325121261458e-05,
+      "loss": 1.4688,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997699302048227e-05,
+      "loss": 1.4619,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976658508145206e-05,
+      "loss": 1.457,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976321584331125e-05,
+      "loss": 1.4404,
+      "step": 338
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975982249121464e-05,
+      "loss": 1.623,
+      "step": 339
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975640502598243e-05,
+      "loss": 1.4385,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997529634484407e-05,
+      "loss": 1.4404,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974949775942134e-05,
+      "loss": 1.3628,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974600795976197e-05,
+      "loss": 1.2212,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974249405030613e-05,
+      "loss": 1.3984,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973895603190327e-05,
+      "loss": 1.2424,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997353939054085e-05,
+      "loss": 1.4609,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973180767168276e-05,
+      "loss": 1.5234,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99728197331593e-05,
+      "loss": 1.5518,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972456288601183e-05,
+      "loss": 1.416,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972090433581777e-05,
+      "loss": 1.5537,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997172216818951e-05,
+      "loss": 1.6348,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971351492513395e-05,
+      "loss": 1.4565,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970978406643034e-05,
+      "loss": 1.5381,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99706029106686e-05,
+      "loss": 1.416,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970225004680862e-05,
+      "loss": 1.4697,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969844688771157e-05,
+      "loss": 1.3999,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996946196303142e-05,
+      "loss": 1.2446,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969076827554153e-05,
+      "loss": 1.4326,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996868928243245e-05,
+      "loss": 1.5723,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996829932775999e-05,
+      "loss": 1.4014,
+      "step": 360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996790696363102e-05,
+      "loss": 1.4521,
+      "step": 361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967512190140385e-05,
+      "loss": 1.54,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996711500738351e-05,
+      "loss": 1.459,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966715415456394e-05,
+      "loss": 1.3984,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996631341445562e-05,
+      "loss": 1.7383,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965909004478364e-05,
+      "loss": 1.3584,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996550218562237e-05,
+      "loss": 1.5908,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965092957985977e-05,
+      "loss": 1.3789,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964681321668097e-05,
+      "loss": 1.3218,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964267276768227e-05,
+      "loss": 1.6123,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963850823386447e-05,
+      "loss": 1.6416,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996343196162342e-05,
+      "loss": 1.373,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963010691580388e-05,
+      "loss": 1.4219,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996258701335918e-05,
+      "loss": 1.5215,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962160927062196e-05,
+      "loss": 1.3281,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961732432792436e-05,
+      "loss": 1.5039,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996130153065347e-05,
+      "loss": 1.3906,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996086822074945e-05,
+      "loss": 1.3545,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960432503185113e-05,
+      "loss": 1.5645,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959994378065772e-05,
+      "loss": 1.4824,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959553845497333e-05,
+      "loss": 1.4717,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995911090558628e-05,
+      "loss": 1.5049,
+      "step": 382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995866555843967e-05,
+      "loss": 1.4883,
+      "step": 383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958217804165153e-05,
+      "loss": 1.4424,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957767642870956e-05,
+      "loss": 1.4619,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995731507466589e-05,
+      "loss": 1.541,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956860099659345e-05,
+      "loss": 1.4238,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995640271796129e-05,
+      "loss": 1.3438,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995594292968228e-05,
+      "loss": 1.5234,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995548073493346e-05,
+      "loss": 1.4609,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995501613382654e-05,
+      "loss": 1.585,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954549126473817e-05,
+      "loss": 1.5303,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954079712988183e-05,
+      "loss": 1.521,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953607893483093e-05,
+      "loss": 1.2231,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953133668072593e-05,
+      "loss": 1.541,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995265703687131e-05,
+      "loss": 1.5664,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952177999994445e-05,
+      "loss": 1.4463,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951696557557794e-05,
+      "loss": 1.4062,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951212709677728e-05,
+      "loss": 1.416,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950726456471195e-05,
+      "loss": 1.4434,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995023779805573e-05,
+      "loss": 1.5176,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949746734549447e-05,
+      "loss": 1.4883,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994925326607104e-05,
+      "loss": 1.3228,
+      "step": 403
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994875739273979e-05,
+      "loss": 1.5576,
+      "step": 404
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948259114675553e-05,
+      "loss": 1.4639,
+      "step": 405
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947758431998774e-05,
+      "loss": 1.5928,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947255344830465e-05,
+      "loss": 1.5107,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946749853292233e-05,
+      "loss": 1.5723,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994624195750626e-05,
+      "loss": 1.5234,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945731657595312e-05,
+      "loss": 1.5752,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 1.4131,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944703845892452e-05,
+      "loss": 1.4033,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944186334348976e-05,
+      "loss": 1.4326,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943666419177392e-05,
+      "loss": 1.4199,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943144100503372e-05,
+      "loss": 1.4609,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942619378453162e-05,
+      "loss": 1.5312,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99420922531536e-05,
+      "loss": 1.4668,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994156272473209e-05,
+      "loss": 1.5635,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941030793316635e-05,
+      "loss": 1.4194,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940496459035803e-05,
+      "loss": 1.3491,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993995972201875e-05,
+      "loss": 1.3584,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939420582395215e-05,
+      "loss": 1.4307,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938879040295508e-05,
+      "loss": 1.3862,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938335095850534e-05,
+      "loss": 1.501,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937788749191763e-05,
+      "loss": 1.3687,
+      "step": 425
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993724000045126e-05,
+      "loss": 1.5,
+      "step": 426
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993668884976166e-05,
+      "loss": 1.5,
+      "step": 427
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936135297256183e-05,
+      "loss": 1.4478,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993557934306863e-05,
+      "loss": 1.4639,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935020987333387e-05,
+      "loss": 1.4023,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934460230185406e-05,
+      "loss": 1.5098,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933897071760236e-05,
+      "loss": 1.2725,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933331512193992e-05,
+      "loss": 1.4092,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932763551623382e-05,
+      "loss": 1.3296,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993219319018569e-05,
+      "loss": 1.5889,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993162042801877e-05,
+      "loss": 1.6201,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993104526526108e-05,
+      "loss": 1.5156,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930467702051632e-05,
+      "loss": 1.4297,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992988773853003e-05,
+      "loss": 1.7041,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929305374836466e-05,
+      "loss": 1.186,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99287206111117e-05,
+      "loss": 1.458,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992813344749707e-05,
+      "loss": 1.4707,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927543884134514e-05,
+      "loss": 1.5889,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926951921166524e-05,
+      "loss": 1.4355,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926357558736192e-05,
+      "loss": 1.4314,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992576079698718e-05,
+      "loss": 1.4458,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925161636063732e-05,
+      "loss": 1.4609,
+      "step": 447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924560076110674e-05,
+      "loss": 1.4932,
+      "step": 448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923956117273406e-05,
+      "loss": 1.3896,
+      "step": 449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923349759697915e-05,
+      "loss": 1.502,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922741003530766e-05,
+      "loss": 1.2974,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922129848919103e-05,
+      "loss": 1.5811,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921516296010645e-05,
+      "loss": 1.3994,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99209003449537e-05,
+      "loss": 1.3672,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920281995897146e-05,
+      "loss": 1.4268,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919661248990446e-05,
+      "loss": 1.5615,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991903810438365e-05,
+      "loss": 1.5234,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991841256222737e-05,
+      "loss": 1.4961,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991778462267281e-05,
+      "loss": 1.5791,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991715428587175e-05,
+      "loss": 1.3389,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991652155197655e-05,
+      "loss": 1.4883,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915886421140152e-05,
+      "loss": 1.5273,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991524889351607e-05,
+      "loss": 1.29,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914608969258405e-05,
+      "loss": 1.4365,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913966648521838e-05,
+      "loss": 1.4346,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913321931461612e-05,
+      "loss": 1.4619,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991267481823358e-05,
+      "loss": 1.6763,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912025308994146e-05,
+      "loss": 1.5117,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911373403900312e-05,
+      "loss": 1.5498,
+      "step": 469
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991071910310964e-05,
+      "loss": 1.4424,
+      "step": 470
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910062406780292e-05,
+      "loss": 1.5928,
+      "step": 471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909403315070996e-05,
+      "loss": 1.5059,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990874182814106e-05,
+      "loss": 1.5879,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908077946150378e-05,
+      "loss": 1.5244,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907411669259416e-05,
+      "loss": 1.2866,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906742997629217e-05,
+      "loss": 1.3857,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906071931421413e-05,
+      "loss": 1.3335,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905398470798207e-05,
+      "loss": 1.4355,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904722615922382e-05,
+      "loss": 1.4766,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904044366957297e-05,
+      "loss": 1.3721,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903363724066903e-05,
+      "loss": 1.4243,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 1.4287,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901995257168808e-05,
+      "loss": 1.6094,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901307433491894e-05,
+      "loss": 1.4941,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990061721655121e-05,
+      "loss": 1.3604,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899924606513593e-05,
+      "loss": 1.3774,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899229603546456e-05,
+      "loss": 1.2671,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989853220781779e-05,
+      "loss": 1.2993,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897832419496163e-05,
+      "loss": 1.5479,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897130238750715e-05,
+      "loss": 1.3604,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896425665751183e-05,
+      "loss": 1.4307,
+      "step": 491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895718700667862e-05,
+      "loss": 1.4326,
+      "step": 492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895009343671638e-05,
+      "loss": 1.4033,
+      "step": 493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894297594933972e-05,
+      "loss": 1.374,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893583454626905e-05,
+      "loss": 1.6094,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989286692292304e-05,
+      "loss": 1.5107,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892147999995582e-05,
+      "loss": 1.4072,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891426686018308e-05,
+      "loss": 1.3813,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890702981165556e-05,
+      "loss": 1.3408,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889976885612257e-05,
+      "loss": 1.5654,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889248399533922e-05,
+      "loss": 1.4775,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988851752310663e-05,
+      "loss": 1.5234,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988778425650705e-05,
+      "loss": 1.48,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887048599912412e-05,
+      "loss": 1.2754,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886310553500537e-05,
+      "loss": 1.4292,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988557011744982e-05,
+      "loss": 1.4609,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884827291939236e-05,
+      "loss": 1.3604,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884082077148326e-05,
+      "loss": 1.5176,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883334473257223e-05,
+      "loss": 1.3398,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882584480446633e-05,
+      "loss": 1.3188,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881832098897836e-05,
+      "loss": 1.27,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881077328792697e-05,
+      "loss": 1.4111,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988032017031364e-05,
+      "loss": 1.6504,
+      "step": 513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879560623643694e-05,
+      "loss": 1.3765,
+      "step": 514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987879868896644e-05,
+      "loss": 1.4326,
+      "step": 515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987803436646605e-05,
+      "loss": 1.4473,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877267656327274e-05,
+      "loss": 1.4941,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876498558735427e-05,
+      "loss": 1.5049,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875727073876415e-05,
+      "loss": 1.46,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874953201936714e-05,
+      "loss": 1.4277,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874176943103375e-05,
+      "loss": 1.4556,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873398297564036e-05,
+      "loss": 1.5063,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.98726172655069e-05,
+      "loss": 1.2544,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871833847120746e-05,
+      "loss": 1.3701,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871048042594947e-05,
+      "loss": 1.4492,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870259852119434e-05,
+      "loss": 1.5371,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986946927588472e-05,
+      "loss": 1.3594,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868676314081907e-05,
+      "loss": 1.543,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867880966902653e-05,
+      "loss": 1.4932,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986708323453921e-05,
+      "loss": 1.3857,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866283117184393e-05,
+      "loss": 1.3926,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865480615031602e-05,
+      "loss": 1.4766,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864675728274813e-05,
+      "loss": 1.2905,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863868457108578e-05,
+      "loss": 1.3225,
+      "step": 534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863058801728023e-05,
+      "loss": 1.4673,
+      "step": 535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862246762328846e-05,
+      "loss": 1.4814,
+      "step": 536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861432339107333e-05,
+      "loss": 1.4756,
+      "step": 537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860615532260337e-05,
+      "loss": 1.5332,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859796341985295e-05,
+      "loss": 1.4277,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858974768480206e-05,
+      "loss": 1.4707,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985815081194366e-05,
+      "loss": 1.3418,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857324472574815e-05,
+      "loss": 1.4268,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985649575057341e-05,
+      "loss": 1.4365,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855664646139753e-05,
+      "loss": 1.4033,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985483115947474e-05,
+      "loss": 1.6055,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985399529077982e-05,
+      "loss": 1.4473,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853157040257044e-05,
+      "loss": 1.4756,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852316408109022e-05,
+      "loss": 1.4312,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985147339453895e-05,
+      "loss": 1.3198,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985062799975059e-05,
+      "loss": 1.4814,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849780223948284e-05,
+      "loss": 1.5156,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848930067336952e-05,
+      "loss": 1.4492,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 1.5078,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984722261250975e-05,
+      "loss": 1.623,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846365314706592e-05,
+      "loss": 1.4971,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845505636919833e-05,
+      "loss": 1.6055,
+      "step": 556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844643579357265e-05,
+      "loss": 1.417,
+      "step": 557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843779142227258e-05,
+      "loss": 1.2866,
+      "step": 558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842912325738757e-05,
+      "loss": 1.5249,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842043130101278e-05,
+      "loss": 1.4639,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984117155552492e-05,
+      "loss": 1.3252,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840297602220356e-05,
+      "loss": 1.2388,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983942127039882e-05,
+      "loss": 1.3281,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838542560272147e-05,
+      "loss": 1.4307,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983766147205272e-05,
+      "loss": 1.5918,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836778005953513e-05,
+      "loss": 1.5762,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835892162188065e-05,
+      "loss": 1.4019,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835003940970504e-05,
+      "loss": 1.499,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983411334251552e-05,
+      "loss": 1.3818,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833220367038373e-05,
+      "loss": 1.4785,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832325014754916e-05,
+      "loss": 1.582,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831427285881567e-05,
+      "loss": 1.4287,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983052718063531e-05,
+      "loss": 1.4316,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829624699233717e-05,
+      "loss": 1.3926,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828719841894924e-05,
+      "loss": 1.5825,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982781260883765e-05,
+      "loss": 1.3154,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826903000281184e-05,
+      "loss": 1.3711,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825991016445387e-05,
+      "loss": 1.5469,
+      "step": 578
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825076657550696e-05,
+      "loss": 1.3608,
+      "step": 579
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824159923818124e-05,
+      "loss": 1.4697,
+      "step": 580
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823240815469257e-05,
+      "loss": 1.3843,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822319332726248e-05,
+      "loss": 1.3398,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982139547581184e-05,
+      "loss": 1.6162,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820469244949338e-05,
+      "loss": 1.457,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819540640362612e-05,
+      "loss": 1.3374,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818609662276134e-05,
+      "loss": 1.3965,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981767631091492e-05,
+      "loss": 1.3418,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981674058650458e-05,
+      "loss": 1.5068,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815802489271282e-05,
+      "loss": 1.3945,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981486201944178e-05,
+      "loss": 1.4121,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813919177243398e-05,
+      "loss": 1.3086,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812973962904026e-05,
+      "loss": 1.5205,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981202637665214e-05,
+      "loss": 1.625,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981107641871678e-05,
+      "loss": 1.3105,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810124089327564e-05,
+      "loss": 1.499,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980916938871468e-05,
+      "loss": 1.4814,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980821231710889e-05,
+      "loss": 1.5088,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807252874741527e-05,
+      "loss": 1.4888,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806291061844507e-05,
+      "loss": 1.2998,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805326878650305e-05,
+      "loss": 1.4277,
+      "step": 600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980436032539198e-05,
+      "loss": 1.6992,
+      "step": 601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803391402303153e-05,
+      "loss": 1.4512,
+      "step": 602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980242010961803e-05,
+      "loss": 1.4941,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980144644757138e-05,
+      "loss": 1.3379,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800470416398557e-05,
+      "loss": 1.4062,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979949201633547e-05,
+      "loss": 1.2983,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798511247618614e-05,
+      "loss": 1.3555,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797528110485055e-05,
+      "loss": 1.5547,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979654260517242e-05,
+      "loss": 1.3203,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979555473191893e-05,
+      "loss": 1.4395,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794564490963352e-05,
+      "loss": 1.3379,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793571882545048e-05,
+      "loss": 1.3799,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792576906903945e-05,
+      "loss": 1.5957,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791579564280533e-05,
+      "loss": 1.3545,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790579854915886e-05,
+      "loss": 1.2886,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789577779051643e-05,
+      "loss": 1.3062,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788573336930023e-05,
+      "loss": 1.4146,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787566528793806e-05,
+      "loss": 1.5078,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786557354886354e-05,
+      "loss": 1.3877,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978554581545159e-05,
+      "loss": 1.4238,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978453191073402e-05,
+      "loss": 1.5234,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783515640978717e-05,
+      "loss": 1.5332,
+      "step": 622
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782497006431326e-05,
+      "loss": 1.5547,
+      "step": 623
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 1.4951,
+      "step": 624
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780452643945706e-05,
+      "loss": 1.3354,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779426916501626e-05,
+      "loss": 1.4463,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977839882525375e-05,
+      "loss": 1.4727,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777368370450582e-05,
+      "loss": 1.4009,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977633555234119e-05,
+      "loss": 1.459,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775300371175225e-05,
+      "loss": 1.4619,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774262827202894e-05,
+      "loss": 1.5137,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773222920674987e-05,
+      "loss": 1.4219,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977218065184287e-05,
+      "loss": 1.3716,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771136020958458e-05,
+      "loss": 1.4707,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770089028274264e-05,
+      "loss": 1.6104,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976903967404335e-05,
+      "loss": 1.6396,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767987958519358e-05,
+      "loss": 1.4395,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766933881956504e-05,
+      "loss": 1.4414,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976587744460957e-05,
+      "loss": 1.3755,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764818646733904e-05,
+      "loss": 1.3057,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9763757488585437e-05,
+      "loss": 1.3833,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976269397042066e-05,
+      "loss": 1.4824,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976162809249664e-05,
+      "loss": 1.4375,
+      "step": 643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760559855071012e-05,
+      "loss": 1.4678,
+      "step": 644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975948925840198e-05,
+      "loss": 1.3042,
+      "step": 645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758416302748316e-05,
+      "loss": 1.4575,
+      "step": 646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757340988369377e-05,
+      "loss": 1.4277,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975626331552507e-05,
+      "loss": 1.5488,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755183284475884e-05,
+      "loss": 1.3721,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754100895482877e-05,
+      "loss": 1.4619,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975301614880768e-05,
+      "loss": 1.4375,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751929044712473e-05,
+      "loss": 1.3042,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975083958346004e-05,
+      "loss": 1.5156,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749747765313706e-05,
+      "loss": 1.3091,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974865359053738e-05,
+      "loss": 1.3042,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747557059395535e-05,
+      "loss": 1.4785,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974645817215322e-05,
+      "loss": 1.5059,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745356929076042e-05,
+      "loss": 1.3804,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744253330430194e-05,
+      "loss": 1.3452,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743147376482424e-05,
+      "loss": 1.4028,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974203906750005e-05,
+      "loss": 1.5986,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974092840375097e-05,
+      "loss": 1.4355,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973981538550364e-05,
+      "loss": 1.2871,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738700013027094e-05,
+      "loss": 1.3066,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737582286590927e-05,
+      "loss": 1.3716,
+      "step": 665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973646220646531e-05,
+      "loss": 1.5029,
+      "step": 666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735339772920977e-05,
+      "loss": 1.3574,
+      "step": 667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734214986229236e-05,
+      "loss": 1.4756,
+      "step": 668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733087846661957e-05,
+      "loss": 1.5645,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731958354491588e-05,
+      "loss": 1.3223,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973082650999114e-05,
+      "loss": 1.3716,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729692313434188e-05,
+      "loss": 1.4395,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728555765094886e-05,
+      "loss": 1.4189,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727416865247954e-05,
+      "loss": 1.3823,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972627561416867e-05,
+      "loss": 1.5596,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725132012132896e-05,
+      "loss": 1.6367,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723986059417045e-05,
+      "loss": 1.4404,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722837756298112e-05,
+      "loss": 1.5654,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721687103053656e-05,
+      "loss": 1.5498,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9720534099961806e-05,
+      "loss": 1.4492,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719378747301255e-05,
+      "loss": 1.5479,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971822104535126e-05,
+      "loss": 1.4199,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717060994391657e-05,
+      "loss": 1.5771,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715898594702845e-05,
+      "loss": 1.3965,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714733846565785e-05,
+      "loss": 1.5107,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713566750262016e-05,
+      "loss": 1.4121,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712397306073634e-05,
+      "loss": 1.373,
+      "step": 687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711225514283308e-05,
+      "loss": 1.4971,
+      "step": 688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9710051375174283e-05,
+      "loss": 1.4155,
+      "step": 689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708874889030348e-05,
+      "loss": 1.5425,
+      "step": 690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707696056135882e-05,
+      "loss": 1.4634,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970651487677582e-05,
+      "loss": 1.418,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705331351235673e-05,
+      "loss": 1.478,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970414547980151e-05,
+      "loss": 1.3813,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 1.5654,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970176670039825e-05,
+      "loss": 1.4492,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970057379300414e-05,
+      "loss": 1.3276,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9699378540865966e-05,
+      "loss": 1.583,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698180944272644e-05,
+      "loss": 1.3906,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969698100351364e-05,
+      "loss": 1.4443,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695778718879002e-05,
+      "loss": 1.3687,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969457409065933e-05,
+      "loss": 1.3291,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693367119145796e-05,
+      "loss": 1.4771,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9692157804630146e-05,
+      "loss": 1.2354,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690946147404682e-05,
+      "loss": 1.3477,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968973214776227e-05,
+      "loss": 1.4248,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688515805996363e-05,
+      "loss": 1.5078,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687297122400952e-05,
+      "loss": 1.4331,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968607609727061e-05,
+      "loss": 1.3057,
+      "step": 709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684852730900475e-05,
+      "loss": 1.3423,
+      "step": 710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683627023586252e-05,
+      "loss": 1.1985,
+      "step": 711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682398975624205e-05,
+      "loss": 1.375,
+      "step": 712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681168587311168e-05,
+      "loss": 1.4307,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679935858944545e-05,
+      "loss": 1.5625,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678700790822294e-05,
+      "loss": 1.4521,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967746338324295e-05,
+      "loss": 1.585,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676223636505608e-05,
+      "loss": 1.3926,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967498155090993e-05,
+      "loss": 1.6221,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673737126756145e-05,
+      "loss": 1.2925,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967249036434504e-05,
+      "loss": 1.5303,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967124126397798e-05,
+      "loss": 1.4912,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669989825956877e-05,
+      "loss": 1.4453,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668736050584224e-05,
+      "loss": 1.4199,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667479938163078e-05,
+      "loss": 1.3164,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966622148899705e-05,
+      "loss": 1.5225,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966496070339032e-05,
+      "loss": 1.5371,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663697581647643e-05,
+      "loss": 1.2944,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9662432124074325e-05,
+      "loss": 1.3408,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661164330976246e-05,
+      "loss": 1.5059,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965989420265984e-05,
+      "loss": 1.54,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965862173943212e-05,
+      "loss": 1.1499,
+      "step": 731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965734694160065e-05,
+      "loss": 1.4639,
+      "step": 732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656069809473567e-05,
+      "loss": 1.3154,
+      "step": 733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654790343359563e-05,
+      "loss": 1.2681,
+      "step": 734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9653508543567906e-05,
+      "loss": 1.459,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652224410408426e-05,
+      "loss": 1.3545,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650937944191505e-05,
+      "loss": 1.5361,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96496491452281e-05,
+      "loss": 1.377,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648358013829732e-05,
+      "loss": 1.6494,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647064550308478e-05,
+      "loss": 1.4131,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645768754976986e-05,
+      "loss": 1.3813,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644470628148466e-05,
+      "loss": 1.4355,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964317017013669e-05,
+      "loss": 1.3711,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641867381255993e-05,
+      "loss": 1.4336,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9640562261821275e-05,
+      "loss": 1.3945,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639254812147998e-05,
+      "loss": 1.5322,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963794503255219e-05,
+      "loss": 1.6035,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963663292335044e-05,
+      "loss": 1.4912,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635318484859905e-05,
+      "loss": 1.6221,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634001717398293e-05,
+      "loss": 1.4497,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632682621283884e-05,
+      "loss": 1.5879,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631361196835515e-05,
+      "loss": 1.335,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96300374443726e-05,
+      "loss": 1.3564,
+      "step": 753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.96287113642151e-05,
+      "loss": 1.3237,
+      "step": 754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627382956683546e-05,
+      "loss": 1.5332,
+      "step": 755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962605222209903e-05,
+      "loss": 1.1167,
+      "step": 756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624719160783205e-05,
+      "loss": 1.3896,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962338377305829e-05,
+      "loss": 1.4111,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962204605924706e-05,
+      "loss": 1.3823,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962070601967286e-05,
+      "loss": 1.4961,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619363654659587e-05,
+      "loss": 1.3213,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618018964531718e-05,
+      "loss": 1.4219,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961667194961427e-05,
+      "loss": 1.2358,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961532261023284e-05,
+      "loss": 1.3833,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613970946713574e-05,
+      "loss": 1.459,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961261695938319e-05,
+      "loss": 1.543,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961126064856896e-05,
+      "loss": 1.585,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960990201459872e-05,
+      "loss": 1.4258,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608541057800867e-05,
+      "loss": 1.3389,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607177778504365e-05,
+      "loss": 1.5664,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960581217703873e-05,
+      "loss": 1.3691,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604444253734044e-05,
+      "loss": 1.4775,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603074008920955e-05,
+      "loss": 1.4746,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601701442930667e-05,
+      "loss": 1.5957,
+      "step": 774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600326556094942e-05,
+      "loss": 1.4053,
+      "step": 775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598949348746104e-05,
+      "loss": 1.5488,
+      "step": 776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597569821217052e-05,
+      "loss": 1.4541,
+      "step": 777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596187973841218e-05,
+      "loss": 1.5488,
+      "step": 778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594803806952628e-05,
+      "loss": 1.6523,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593417320885836e-05,
+      "loss": 1.4248,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592028515975983e-05,
+      "loss": 1.5879,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590637392558754e-05,
+      "loss": 1.4482,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95892439509704e-05,
+      "loss": 1.373,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958784819154774e-05,
+      "loss": 1.5771,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586450114628138e-05,
+      "loss": 1.5479,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9585049720549526e-05,
+      "loss": 1.4443,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95836470096504e-05,
+      "loss": 1.4551,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582241982269807e-05,
+      "loss": 1.396,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580834638747363e-05,
+      "loss": 1.5166,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957942497942324e-05,
+      "loss": 1.4419,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578013004638166e-05,
+      "loss": 1.5049,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576598714733434e-05,
+      "loss": 1.5322,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957518211005089e-05,
+      "loss": 1.4175,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573763190932954e-05,
+      "loss": 1.3364,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957234195772259e-05,
+      "loss": 1.4717,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570918410763324e-05,
+      "loss": 1.5518,
+      "step": 796
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956949255039925e-05,
+      "loss": 1.5215,
+      "step": 797
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568064376975013e-05,
+      "loss": 1.292,
+      "step": 798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566633890835816e-05,
+      "loss": 1.374,
+      "step": 799
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956520109232743e-05,
+      "loss": 1.4375,
+      "step": 800
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956376598179618e-05,
+      "loss": 1.4756,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562328559588943e-05,
+      "loss": 1.4219,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9560888826053162e-05,
+      "loss": 1.4912,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559446781536847e-05,
+      "loss": 1.3848,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9558002426388546e-05,
+      "loss": 1.4697,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556555760957387e-05,
+      "loss": 1.4204,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555106785593035e-05,
+      "loss": 1.4932,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955365550064573e-05,
+      "loss": 1.5527,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955220190646627e-05,
+      "loss": 1.4316,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550746003405996e-05,
+      "loss": 1.3857,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549287791816825e-05,
+      "loss": 1.375,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547827272051218e-05,
+      "loss": 1.5107,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546364444462207e-05,
+      "loss": 1.4395,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544899309403372e-05,
+      "loss": 1.4883,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954343186722885e-05,
+      "loss": 1.3359,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954196211829334e-05,
+      "loss": 1.5938,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95404900629521e-05,
+      "loss": 1.5234,
+      "step": 817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953901570156095e-05,
+      "loss": 1.0977,
+      "step": 818
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537539034476247e-05,
+      "loss": 1.5391,
+      "step": 819
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953606006205493e-05,
+      "loss": 1.3652,
+      "step": 820
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534578784654474e-05,
+      "loss": 1.6055,
+      "step": 821
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953309520263293e-05,
+      "loss": 1.4404,
+      "step": 822
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95316093163489e-05,
+      "loss": 1.3276,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530121126161528e-05,
+      "loss": 1.3711,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528630632430533e-05,
+      "loss": 1.3555,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527137835516192e-05,
+      "loss": 1.4131,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525642735779323e-05,
+      "loss": 1.4404,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524145333581315e-05,
+      "loss": 1.5449,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522645629284102e-05,
+      "loss": 1.1714,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952114362325019e-05,
+      "loss": 1.3169,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951963931584262e-05,
+      "loss": 1.4238,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518132707425007e-05,
+      "loss": 1.4463,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516623798361518e-05,
+      "loss": 1.4248,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951511258901687e-05,
+      "loss": 1.5283,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513599079756344e-05,
+      "loss": 1.4668,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512083270945768e-05,
+      "loss": 1.7012,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 1.2891,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509044756140592e-05,
+      "loss": 1.4912,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507522050880435e-05,
+      "loss": 1.4443,
+      "step": 839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505997047539122e-05,
+      "loss": 1.5332,
+      "step": 840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504469746485266e-05,
+      "loss": 1.502,
+      "step": 841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950294014808803e-05,
+      "loss": 1.3799,
+      "step": 842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501408252717136e-05,
+      "loss": 1.4229,
+      "step": 843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499874060742865e-05,
+      "loss": 1.5234,
+      "step": 844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.949833757253605e-05,
+      "loss": 1.4165,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496798788468077e-05,
+      "loss": 1.3452,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495257708910885e-05,
+      "loss": 1.3237,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493714334236973e-05,
+      "loss": 1.4062,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492168664819394e-05,
+      "loss": 1.3887,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490620701031756e-05,
+      "loss": 1.4072,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489070443248215e-05,
+      "loss": 1.3086,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487517891843497e-05,
+      "loss": 1.417,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485963047192862e-05,
+      "loss": 1.2793,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484405909672138e-05,
+      "loss": 1.5664,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482846479657707e-05,
+      "loss": 1.416,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481284757526497e-05,
+      "loss": 1.5254,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479720743655992e-05,
+      "loss": 1.5322,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947815443842424e-05,
+      "loss": 1.5596,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476585842209835e-05,
+      "loss": 1.3662,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9475014955391925e-05,
+      "loss": 1.4551,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473441778350204e-05,
+      "loss": 1.4121,
+      "step": 861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471866311464937e-05,
+      "loss": 1.3643,
+      "step": 862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470288555116933e-05,
+      "loss": 1.5107,
+      "step": 863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468708509687545e-05,
+      "loss": 1.5352,
+      "step": 864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467126175558698e-05,
+      "loss": 1.4019,
+      "step": 865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465541553112857e-05,
+      "loss": 1.5645,
+      "step": 866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946395464273305e-05,
+      "loss": 1.6758,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946236544480284e-05,
+      "loss": 1.4004,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460773959706366e-05,
+      "loss": 1.5518,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945918018782831e-05,
+      "loss": 1.335,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457584129553894e-05,
+      "loss": 1.2827,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455985785268914e-05,
+      "loss": 1.4473,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454385155359704e-05,
+      "loss": 1.3564,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945278224021316e-05,
+      "loss": 1.3501,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945117704021672e-05,
+      "loss": 1.3047,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449569555758384e-05,
+      "loss": 1.3818,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447959787226697e-05,
+      "loss": 1.375,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446347735010765e-05,
+      "loss": 1.5928,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444733399500234e-05,
+      "loss": 1.4414,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944311678108531e-05,
+      "loss": 1.3887,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441497880156752e-05,
+      "loss": 1.5977,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943987669710586e-05,
+      "loss": 1.4146,
+      "step": 882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438253232324503e-05,
+      "loss": 1.3335,
+      "step": 883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436627486205087e-05,
+      "loss": 1.3906,
+      "step": 884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434999459140572e-05,
+      "loss": 1.5996,
+      "step": 885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943336915152447e-05,
+      "loss": 1.3506,
+      "step": 886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9431736563750856e-05,
+      "loss": 1.4502,
+      "step": 887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430101696214335e-05,
+      "loss": 1.4512,
+      "step": 888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942846454931008e-05,
+      "loss": 1.4541,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9426825123433805e-05,
+      "loss": 1.4717,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425183418981783e-05,
+      "loss": 1.3608,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942353943635083e-05,
+      "loss": 1.5449,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421893175938317e-05,
+      "loss": 1.4033,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420244638142166e-05,
+      "loss": 1.4688,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418593823360848e-05,
+      "loss": 1.4805,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941694073199338e-05,
+      "loss": 1.4214,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415285364439343e-05,
+      "loss": 1.4434,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941362772109885e-05,
+      "loss": 1.3838,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941196780237258e-05,
+      "loss": 1.3789,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410305608661745e-05,
+      "loss": 1.5654,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940864114036813e-05,
+      "loss": 1.3281,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940697439789405e-05,
+      "loss": 1.4287,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9405305381642376e-05,
+      "loss": 1.3511,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940363409201653e-05,
+      "loss": 1.5693,
+      "step": 904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401960529420483e-05,
+      "loss": 1.4688,
+      "step": 905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940028469425876e-05,
+      "loss": 1.3535,
+      "step": 906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9398606586936418e-05,
+      "loss": 1.4839,
+      "step": 907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 1.5088,
+      "step": 908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939524355743293e-05,
+      "loss": 1.377,
+      "step": 909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939355863606466e-05,
+      "loss": 1.2729,
+      "step": 910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391871444161547e-05,
+      "loss": 1.4482,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390181982131407e-05,
+      "loss": 1.3052,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93884902503826e-05,
+      "loss": 1.4082,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386796249324037e-05,
+      "loss": 1.2139,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938509997936518e-05,
+      "loss": 1.666,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383401440916042e-05,
+      "loss": 1.5215,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938170063438717e-05,
+      "loss": 1.3208,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379997560189677e-05,
+      "loss": 1.3042,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378292218735212e-05,
+      "loss": 1.3682,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937658461043598e-05,
+      "loss": 1.395,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374874735704726e-05,
+      "loss": 1.5479,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373162594954752e-05,
+      "loss": 1.4346,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93714481885999e-05,
+      "loss": 1.5986,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369731517054556e-05,
+      "loss": 1.5801,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9368012580733674e-05,
+      "loss": 1.4307,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936629138005273e-05,
+      "loss": 1.5254,
+      "step": 926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936456791542776e-05,
+      "loss": 1.4517,
+      "step": 927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362842187275354e-05,
+      "loss": 1.2739,
+      "step": 928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361114196012635e-05,
+      "loss": 1.3926,
+      "step": 929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359383942057274e-05,
+      "loss": 1.3267,
+      "step": 930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357651425827507e-05,
+      "loss": 1.4844,
+      "step": 931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935591664774209e-05,
+      "loss": 1.2417,
+      "step": 932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354179608220347e-05,
+      "loss": 1.4111,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935244030768214e-05,
+      "loss": 1.4971,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935069874654788e-05,
+      "loss": 1.6367,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934895492523852e-05,
+      "loss": 1.2388,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934720884417556e-05,
+      "loss": 1.5244,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934546050378106e-05,
+      "loss": 1.4966,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343709904477604e-05,
+      "loss": 1.4883,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341957046688332e-05,
+      "loss": 1.4473,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340201930836938e-05,
+      "loss": 1.4424,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933844455734765e-05,
+      "loss": 1.459,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933668492664525e-05,
+      "loss": 1.3926,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933492303915505e-05,
+      "loss": 1.4053,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333158895302934e-05,
+      "loss": 1.5078,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331392495515308e-05,
+      "loss": 1.5166,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329623840219135e-05,
+      "loss": 1.5898,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327852929841918e-05,
+      "loss": 1.5137,
+      "step": 948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326079764811707e-05,
+      "loss": 1.3472,
+      "step": 949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324304345557098e-05,
+      "loss": 1.4814,
+      "step": 950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322526672507234e-05,
+      "loss": 1.4707,
+      "step": 951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9320746746091796e-05,
+      "loss": 1.2705,
+      "step": 952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931896456674101e-05,
+      "loss": 1.3418,
+      "step": 953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931718013488566e-05,
+      "loss": 1.4072,
+      "step": 954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315393450957054e-05,
+      "loss": 1.436,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313604515387065e-05,
+      "loss": 1.376,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311813328608088e-05,
+      "loss": 1.3545,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310019891053082e-05,
+      "loss": 1.4092,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930822420315554e-05,
+      "loss": 1.4092,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306426265349506e-05,
+      "loss": 1.3682,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304626078069556e-05,
+      "loss": 1.4956,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930282364175082e-05,
+      "loss": 1.4346,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301018956828966e-05,
+      "loss": 1.3418,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929921202374021e-05,
+      "loss": 1.4419,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297402842921305e-05,
+      "loss": 1.5127,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9295591414809562e-05,
+      "loss": 1.5303,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293777739842814e-05,
+      "loss": 1.4785,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291961818459454e-05,
+      "loss": 1.4658,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290143651098405e-05,
+      "loss": 1.4585,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928832323819915e-05,
+      "loss": 1.4443,
+      "step": 970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286500580201695e-05,
+      "loss": 1.4443,
+      "step": 971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92846756775466e-05,
+      "loss": 1.4458,
+      "step": 972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928284853067497e-05,
+      "loss": 1.5181,
+      "step": 973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281019140028447e-05,
+      "loss": 1.4268,
+      "step": 974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279187506049214e-05,
+      "loss": 1.3857,
+      "step": 975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927735362918e-05,
+      "loss": 1.4922,
+      "step": 976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927551750986407e-05,
+      "loss": 1.4136,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273679148545246e-05,
+      "loss": 1.3667,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 1.3174,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269995701676855e-05,
+      "loss": 1.4023,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926815061701762e-05,
+      "loss": 1.6152,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926630329213615e-05,
+      "loss": 1.3135,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926445372747897e-05,
+      "loss": 1.3296,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262601923493134e-05,
+      "loss": 1.4805,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260747880626252e-05,
+      "loss": 1.4727,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258891599326468e-05,
+      "loss": 1.5166,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925703308004246e-05,
+      "loss": 1.5195,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255172323223465e-05,
+      "loss": 1.5176,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925330932931924e-05,
+      "loss": 1.2529,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251444098780095e-05,
+      "loss": 1.3672,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249576632056884e-05,
+      "loss": 1.2705,
+      "step": 991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924770692960099e-05,
+      "loss": 1.48,
+      "step": 992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245834991864344e-05,
+      "loss": 1.4399,
+      "step": 993
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924396081929942e-05,
+      "loss": 1.3179,
+      "step": 994
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924208441235922e-05,
+      "loss": 1.4795,
+      "step": 995
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240205771497296e-05,
+      "loss": 1.5176,
+      "step": 996
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9238324897167744e-05,
+      "loss": 1.2881,
+      "step": 997
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236441789825185e-05,
+      "loss": 1.4062,
+      "step": 998
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9234556449924796e-05,
+      "loss": 1.3027,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232668877922283e-05,
+      "loss": 1.373,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230779074273895e-05,
+      "loss": 1.5664,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228887039436417e-05,
+      "loss": 1.4258,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226992773867182e-05,
+      "loss": 1.3018,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225096278024053e-05,
+      "loss": 1.5068,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223197552365433e-05,
+      "loss": 1.3442,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221296597350268e-05,
+      "loss": 1.4189,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9219393413438047e-05,
+      "loss": 1.4844,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9217488001088784e-05,
+      "loss": 1.5439,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921558036076305e-05,
+      "loss": 1.3438,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213670492921933e-05,
+      "loss": 1.4043,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211758398027077e-05,
+      "loss": 1.377,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920984407654066e-05,
+      "loss": 1.6328,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920792752892539e-05,
+      "loss": 1.3369,
+      "step": 1013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206008755644524e-05,
+      "loss": 1.4863,
+      "step": 1014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920408775716185e-05,
+      "loss": 1.3896,
+      "step": 1015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202164533941698e-05,
+      "loss": 1.376,
+      "step": 1016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200239086448934e-05,
+      "loss": 1.3828,
+      "step": 1017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919831141514896e-05,
+      "loss": 1.4111,
+      "step": 1018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919638152050772e-05,
+      "loss": 1.5605,
+      "step": 1019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194449402991687e-05,
+      "loss": 1.4688,
+      "step": 1020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9192515063067882e-05,
+      "loss": 1.4805,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190578501203854e-05,
+      "loss": 1.5234,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188639717867695e-05,
+      "loss": 1.4609,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186698713528037e-05,
+      "loss": 1.2451,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9184755488654036e-05,
+      "loss": 1.4844,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9182810043715394e-05,
+      "loss": 1.2876,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918086237918235e-05,
+      "loss": 1.5205,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917891249552568e-05,
+      "loss": 1.3086,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176960393216686e-05,
+      "loss": 1.3862,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175006072727226e-05,
+      "loss": 1.3125,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917304953452967e-05,
+      "loss": 1.3247,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171090779096947e-05,
+      "loss": 1.4575,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169129806902507e-05,
+      "loss": 1.2861,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916716661842034e-05,
+      "loss": 1.4951,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165201214124972e-05,
+      "loss": 1.3037,
+      "step": 1035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163233594491467e-05,
+      "loss": 1.5659,
+      "step": 1036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161263759995417e-05,
+      "loss": 1.3281,
+      "step": 1037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159291711112962e-05,
+      "loss": 1.3916,
+      "step": 1038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9157317448320763e-05,
+      "loss": 1.5439,
+      "step": 1039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155340972096022e-05,
+      "loss": 1.2607,
+      "step": 1040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153362282916487e-05,
+      "loss": 1.3457,
+      "step": 1041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151381381260418e-05,
+      "loss": 1.4565,
+      "step": 1042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914939826760663e-05,
+      "loss": 1.5117,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147412942434463e-05,
+      "loss": 1.2163,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145425406223794e-05,
+      "loss": 1.4004,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143435659455037e-05,
+      "loss": 1.4814,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914144370260913e-05,
+      "loss": 1.5195,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139449536167562e-05,
+      "loss": 1.2827,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137453160612337e-05,
+      "loss": 1.2422,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 1.3564,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913345378409166e-05,
+      "loss": 1.3887,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131450784092905e-05,
+      "loss": 1.3584,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129445576913886e-05,
+      "loss": 1.4893,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9127438163039297e-05,
+      "loss": 1.3232,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125428542954345e-05,
+      "loss": 1.3965,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123416717144783e-05,
+      "loss": 1.4199,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9121402686096892e-05,
+      "loss": 1.4561,
+      "step": 1057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119386450297487e-05,
+      "loss": 1.3447,
+      "step": 1058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9117368010233917e-05,
+      "loss": 1.4971,
+      "step": 1059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911534736639406e-05,
+      "loss": 1.3516,
+      "step": 1060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113324519266333e-05,
+      "loss": 1.5005,
+      "step": 1061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911129946933968e-05,
+      "loss": 1.3604,
+      "step": 1062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9109272217103584e-05,
+      "loss": 1.2998,
+      "step": 1063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107242763048052e-05,
+      "loss": 1.4004,
+      "step": 1064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9105211107663632e-05,
+      "loss": 1.4863,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103177251441386e-05,
+      "loss": 1.5264,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101141194872938e-05,
+      "loss": 1.4346,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909910293845042e-05,
+      "loss": 1.4033,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90970624826665e-05,
+      "loss": 1.4668,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095019828014382e-05,
+      "loss": 1.3828,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90929749749878e-05,
+      "loss": 1.4404,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090927924081027e-05,
+      "loss": 1.4668,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088878675788844e-05,
+      "loss": 1.3662,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086827230606592e-05,
+      "loss": 1.3831,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084773589030127e-05,
+      "loss": 1.3267,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082717751555833e-05,
+      "loss": 1.3872,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080659718680635e-05,
+      "loss": 1.4268,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078599490901984e-05,
+      "loss": 1.4717,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076537068717863e-05,
+      "loss": 1.4473,
+      "step": 1079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907447245262678e-05,
+      "loss": 1.4395,
+      "step": 1080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9072405643127777e-05,
+      "loss": 1.3604,
+      "step": 1081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907033664072043e-05,
+      "loss": 1.3101,
+      "step": 1082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9068265445904838e-05,
+      "loss": 1.374,
+      "step": 1083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066192059181635e-05,
+      "loss": 1.499,
+      "step": 1084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064116481051986e-05,
+      "loss": 1.4507,
+      "step": 1085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906203871201758e-05,
+      "loss": 1.4033,
+      "step": 1086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9059958752580638e-05,
+      "loss": 1.3076,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057876603243912e-05,
+      "loss": 1.3091,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9055792264510683e-05,
+      "loss": 1.4736,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053705736884756e-05,
+      "loss": 1.5352,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051617020870477e-05,
+      "loss": 1.415,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904952611697271e-05,
+      "loss": 1.1538,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904743302569685e-05,
+      "loss": 1.5166,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904533774754882e-05,
+      "loss": 1.3784,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904324028303508e-05,
+      "loss": 1.3389,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041140632662606e-05,
+      "loss": 1.4531,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039038796938916e-05,
+      "loss": 1.3721,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903693477637204e-05,
+      "loss": 1.4932,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903482857147055e-05,
+      "loss": 1.2812,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903272018274354e-05,
+      "loss": 1.3657,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9030609610700633e-05,
+      "loss": 1.3594,
+      "step": 1101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028496855851975e-05,
+      "loss": 1.5566,
+      "step": 1102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026381918708248e-05,
+      "loss": 1.1846,
+      "step": 1103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9024264799780657e-05,
+      "loss": 1.2378,
+      "step": 1104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902214549958094e-05,
+      "loss": 1.3623,
+      "step": 1105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9020024018621345e-05,
+      "loss": 1.4834,
+      "step": 1106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901790035741467e-05,
+      "loss": 1.3022,
+      "step": 1107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015774516474224e-05,
+      "loss": 1.168,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901364649631385e-05,
+      "loss": 1.4375,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901151629744791e-05,
+      "loss": 1.4326,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900938392039131e-05,
+      "loss": 1.3936,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007249365659462e-05,
+      "loss": 1.5029,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005112633768315e-05,
+      "loss": 1.4092,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002973725234343e-05,
+      "loss": 1.4336,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900083264057455e-05,
+      "loss": 1.4619,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899868938030645e-05,
+      "loss": 1.2822,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996543944948104e-05,
+      "loss": 1.3013,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899439633501809e-05,
+      "loss": 1.3823,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992246551035505e-05,
+      "loss": 1.4053,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899009459351998e-05,
+      "loss": 1.5459,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987940462991673e-05,
+      "loss": 1.3818,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985784159971256e-05,
+      "loss": 1.4307,
+      "step": 1122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983625684979933e-05,
+      "loss": 1.4717,
+      "step": 1123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981465038539442e-05,
+      "loss": 1.3403,
+      "step": 1124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979302221172027e-05,
+      "loss": 1.3804,
+      "step": 1125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977137233400477e-05,
+      "loss": 1.334,
+      "step": 1126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897497007574808e-05,
+      "loss": 1.4707,
+      "step": 1127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897280074873868e-05,
+      "loss": 1.252,
+      "step": 1128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970629252896618e-05,
+      "loss": 1.2319,
+      "step": 1129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8968455588746777e-05,
+      "loss": 1.5166,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896627975681455e-05,
+      "loss": 1.457,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896410175762587e-05,
+      "loss": 1.2075,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896192159170718e-05,
+      "loss": 1.3057,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959739259585458e-05,
+      "loss": 1.5371,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957554761788192e-05,
+      "loss": 1.3936,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8955368098843406e-05,
+      "loss": 1.3525,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953179271279638e-05,
+      "loss": 1.3213,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950988279625958e-05,
+      "loss": 1.4297,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948795124411952e-05,
+      "loss": 1.5322,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946599806167734e-05,
+      "loss": 1.4639,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894440232542394e-05,
+      "loss": 1.4375,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894220268271172e-05,
+      "loss": 1.457,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8940000878562758e-05,
+      "loss": 1.3975,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8937796913509258e-05,
+      "loss": 1.3867,
+      "step": 1144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935590788083945e-05,
+      "loss": 1.5127,
+      "step": 1145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933382502820064e-05,
+      "loss": 1.4707,
+      "step": 1146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893117205825138e-05,
+      "loss": 1.2729,
+      "step": 1147
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928959454912196e-05,
+      "loss": 1.4326,
+      "step": 1148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926744693337308e-05,
+      "loss": 1.5293,
+      "step": 1149
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8924527774062064e-05,
+      "loss": 1.3306,
+      "step": 1150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892230869762231e-05,
+      "loss": 1.3691,
+      "step": 1151
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920087464554427e-05,
+      "loss": 1.4351,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917864075395313e-05,
+      "loss": 1.4351,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891563853068239e-05,
+      "loss": 1.2769,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913410830953597e-05,
+      "loss": 1.4932,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911180976747398e-05,
+      "loss": 1.4404,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890894896860277e-05,
+      "loss": 1.4961,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906714807059218e-05,
+      "loss": 1.209,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890447849265677e-05,
+      "loss": 1.3096,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902240025935962e-05,
+      "loss": 1.3008,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889999940743786e-05,
+      "loss": 1.2773,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897756637704055e-05,
+      "loss": 1.521,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8895511717276643e-05,
+      "loss": 1.3301,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893264646698253e-05,
+      "loss": 1.2798,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891015426512026e-05,
+      "loss": 1.3242,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888764057261625e-05,
+      "loss": 1.3428,
+      "step": 1166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886510539491236e-05,
+      "loss": 1.3643,
+      "step": 1167
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884254873745557e-05,
+      "loss": 1.5352,
+      "step": 1168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881997060569813e-05,
+      "loss": 1.4258,
+      "step": 1169
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887973710050974e-05,
+      "loss": 1.4502,
+      "step": 1170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8877474994111598e-05,
+      "loss": 1.4639,
+      "step": 1171
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875210741922168e-05,
+      "loss": 1.3472,
+      "step": 1172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872944344488747e-05,
+      "loss": 1.3848,
+      "step": 1173
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887067580235915e-05,
+      "loss": 1.3403,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8868405116081706e-05,
+      "loss": 1.2915,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866132286205273e-05,
+      "loss": 1.4229,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863857313279215e-05,
+      "loss": 1.3481,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861580197853423e-05,
+      "loss": 1.2466,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8859300940478304e-05,
+      "loss": 1.4082,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885701954170478e-05,
+      "loss": 1.4893,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854736002084297e-05,
+      "loss": 1.3872,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88524503221688e-05,
+      "loss": 1.3691,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850162502510784e-05,
+      "loss": 1.3501,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847872543663228e-05,
+      "loss": 1.5371,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8845580446179646e-05,
+      "loss": 1.3691,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884328621061407e-05,
+      "loss": 1.2837,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840989837521036e-05,
+      "loss": 1.6553,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883869132745561e-05,
+      "loss": 1.2246,
+      "step": 1188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8836390680973372e-05,
+      "loss": 1.415,
+      "step": 1189
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8834087898630407e-05,
+      "loss": 1.2153,
+      "step": 1190
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8831782980983333e-05,
+      "loss": 1.4658,
+      "step": 1191
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829475928589272e-05,
+      "loss": 1.5713,
+      "step": 1192
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827166742005865e-05,
+      "loss": 1.5371,
+      "step": 1193
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882485542179127e-05,
+      "loss": 1.3506,
+      "step": 1194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822541968504167e-05,
+      "loss": 1.4419,
+      "step": 1195
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820226382703737e-05,
+      "loss": 1.6279,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881790866494969e-05,
+      "loss": 1.3311,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881558881580224e-05,
+      "loss": 1.5166,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881326683582213e-05,
+      "loss": 1.3701,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8810942725570603e-05,
+      "loss": 1.2075,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880861648560943e-05,
+      "loss": 1.3711,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8806288116500885e-05,
+      "loss": 1.5498,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803957618807764e-05,
+      "loss": 1.5264,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801624993093383e-05,
+      "loss": 1.4941,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799290239921558e-05,
+      "loss": 1.4434,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8796953359856626e-05,
+      "loss": 1.3281,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794614353463447e-05,
+      "loss": 1.3799,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8792273221307377e-05,
+      "loss": 1.5488,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789929963954296e-05,
+      "loss": 1.5234,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787584581970608e-05,
+      "loss": 1.4551,
+      "step": 1210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785237075923208e-05,
+      "loss": 1.3721,
+      "step": 1211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878288744637952e-05,
+      "loss": 1.4019,
+      "step": 1212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878053569390748e-05,
+      "loss": 1.25,
+      "step": 1213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877818181907553e-05,
+      "loss": 1.3501,
+      "step": 1214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775825822452636e-05,
+      "loss": 1.2783,
+      "step": 1215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773467704608262e-05,
+      "loss": 1.3208,
+      "step": 1216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8771107466112398e-05,
+      "loss": 1.2925,
+      "step": 1217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876874510753554e-05,
+      "loss": 1.5376,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8766380629448705e-05,
+      "loss": 1.4941,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876401403242341e-05,
+      "loss": 1.4126,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761645317031686e-05,
+      "loss": 1.5186,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875927448384608e-05,
+      "loss": 1.4502,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875690153343966e-05,
+      "loss": 1.4346,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754526466385986e-05,
+      "loss": 1.5889,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875214928325914e-05,
+      "loss": 1.4844,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8749769984633727e-05,
+      "loss": 1.5088,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747388571084843e-05,
+      "loss": 1.2925,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8745005043188104e-05,
+      "loss": 1.4199,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874261940151964e-05,
+      "loss": 1.5215,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8740231646656084e-05,
+      "loss": 1.5391,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8737841779174587e-05,
+      "loss": 1.394,
+      "step": 1231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8735449799652815e-05,
+      "loss": 1.3428,
+      "step": 1232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733055708668928e-05,
+      "loss": 1.5244,
+      "step": 1233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730659506801613e-05,
+      "loss": 1.5225,
+      "step": 1234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728261194630057e-05,
+      "loss": 1.4619,
+      "step": 1235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872586077273396e-05,
+      "loss": 1.5264,
+      "step": 1236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723458241693536e-05,
+      "loss": 1.4697,
+      "step": 1237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721053602089502e-05,
+      "loss": 1.2563,
+      "step": 1238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871864685450309e-05,
+      "loss": 1.3691,
+      "step": 1239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716237999516042e-05,
+      "loss": 1.3799,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.87138270377106e-05,
+      "loss": 1.4482,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871141396966953e-05,
+      "loss": 1.3545,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870899879597609e-05,
+      "loss": 1.25,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706581517214065e-05,
+      "loss": 1.364,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704162133967733e-05,
+      "loss": 1.3638,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701740646821894e-05,
+      "loss": 1.584,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869931705636185e-05,
+      "loss": 1.4043,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696891363173405e-05,
+      "loss": 1.4512,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869446356784288e-05,
+      "loss": 1.3862,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869203367095711e-05,
+      "loss": 1.3428,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868960167310342e-05,
+      "loss": 1.3154,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687167574869656e-05,
+      "loss": 1.4668,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868473137684417e-05,
+      "loss": 1.3535,
+      "step": 1253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868229307961582e-05,
+      "loss": 1.4678,
+      "step": 1254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679852683773973e-05,
+      "loss": 1.5,
+      "step": 1255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677410189908496e-05,
+      "loss": 1.4067,
+      "step": 1256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674965598609772e-05,
+      "loss": 1.377,
+      "step": 1257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672518910468692e-05,
+      "loss": 1.29,
+      "step": 1258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670070126076643e-05,
+      "loss": 1.2754,
+      "step": 1259
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667619246025527e-05,
+      "loss": 1.2539,
+      "step": 1260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665166270907753e-05,
+      "loss": 1.2671,
+      "step": 1261
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8662711201316235e-05,
+      "loss": 1.3477,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 1.3311,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865779478108614e-05,
+      "loss": 1.415,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655333431635926e-05,
+      "loss": 1.3579,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652869990088675e-05,
+      "loss": 1.5186,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865040445703984e-05,
+      "loss": 1.5127,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647936833085364e-05,
+      "loss": 1.2798,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86454671188217e-05,
+      "loss": 1.4111,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642995314845813e-05,
+      "loss": 1.541,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864052142175516e-05,
+      "loss": 1.2561,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638045440147715e-05,
+      "loss": 1.3447,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635567370621954e-05,
+      "loss": 1.3213,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8633087213776853e-05,
+      "loss": 1.4883,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630604970211892e-05,
+      "loss": 1.3745,
+      "step": 1275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628120640527067e-05,
+      "loss": 1.4795,
+      "step": 1276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625634225322863e-05,
+      "loss": 1.4219,
+      "step": 1277
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862314572520028e-05,
+      "loss": 1.3799,
+      "step": 1278
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620655140760816e-05,
+      "loss": 1.5693,
+      "step": 1279
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618162472606475e-05,
+      "loss": 1.376,
+      "step": 1280
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615667721339768e-05,
+      "loss": 1.5186,
+      "step": 1281
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.86131708875637e-05,
+      "loss": 1.4561,
+      "step": 1282
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610671971881787e-05,
+      "loss": 1.4248,
+      "step": 1283
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608170974898053e-05,
+      "loss": 1.4458,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860566789721701e-05,
+      "loss": 1.4072,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603162739443687e-05,
+      "loss": 1.4668,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860065550218361e-05,
+      "loss": 1.0337,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598146186042808e-05,
+      "loss": 1.4844,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859563479162781e-05,
+      "loss": 1.457,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593121319545653e-05,
+      "loss": 1.4209,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859060577040387e-05,
+      "loss": 1.3521,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85880881448105e-05,
+      "loss": 1.2974,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585568443374087e-05,
+      "loss": 1.3955,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858304666670367e-05,
+      "loss": 1.373,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858052281540879e-05,
+      "loss": 1.2563,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577996890099494e-05,
+      "loss": 1.4136,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575468891386324e-05,
+      "loss": 1.3721,
+      "step": 1297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572938819880337e-05,
+      "loss": 1.5479,
+      "step": 1298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8570406676193077e-05,
+      "loss": 1.293,
+      "step": 1299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567872460936586e-05,
+      "loss": 1.5176,
+      "step": 1300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8565336174723424e-05,
+      "loss": 1.418,
+      "step": 1301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562797818166637e-05,
+      "loss": 1.4023,
+      "step": 1302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560257391879778e-05,
+      "loss": 1.3916,
+      "step": 1303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557714896476898e-05,
+      "loss": 1.3315,
+      "step": 1304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555170332572544e-05,
+      "loss": 1.4482,
+      "step": 1305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8552623700781774e-05,
+      "loss": 1.5103,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8550075001720133e-05,
+      "loss": 1.4341,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8547524236003675e-05,
+      "loss": 1.3042,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854497140424895e-05,
+      "loss": 1.4668,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542416507073005e-05,
+      "loss": 1.5352,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85398595450934e-05,
+      "loss": 1.3892,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853730051892817e-05,
+      "loss": 1.3994,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534739429195867e-05,
+      "loss": 1.4678,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532176276515538e-05,
+      "loss": 1.4824,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8529611061506725e-05,
+      "loss": 1.5059,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852704378478947e-05,
+      "loss": 1.4189,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8524474446984322e-05,
+      "loss": 1.3921,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521903048712314e-05,
+      "loss": 1.4678,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519329590594983e-05,
+      "loss": 1.3623,
+      "step": 1319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8516754073254367e-05,
+      "loss": 1.4668,
+      "step": 1320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514176497312997e-05,
+      "loss": 1.377,
+      "step": 1321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511596863393907e-05,
+      "loss": 1.4912,
+      "step": 1322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850901517212062e-05,
+      "loss": 1.4336,
+      "step": 1323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850643142411717e-05,
+      "loss": 1.3672,
+      "step": 1324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850384562000807e-05,
+      "loss": 1.4922,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501257760418347e-05,
+      "loss": 1.5,
+      "step": 1326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498667845973515e-05,
+      "loss": 1.3521,
+      "step": 1327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496075877299585e-05,
+      "loss": 1.4639,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849348185502307e-05,
+      "loss": 1.3662,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849088577977097e-05,
+      "loss": 1.4053,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488287652170797e-05,
+      "loss": 1.186,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485687472850538e-05,
+      "loss": 1.3022,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84830852424387e-05,
+      "loss": 1.2622,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 1.3887,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8477874630856712e-05,
+      "loss": 1.0239,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475266250946036e-05,
+      "loss": 1.501,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472655822462707e-05,
+      "loss": 1.4395,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8470043346037698e-05,
+      "loss": 1.4009,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467428822302474e-05,
+      "loss": 1.4087,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464812251888997e-05,
+      "loss": 1.6309,
+      "step": 1340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462193635429723e-05,
+      "loss": 1.4321,
+      "step": 1341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459572973557605e-05,
+      "loss": 1.3906,
+      "step": 1342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456950266906085e-05,
+      "loss": 1.293,
+      "step": 1343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454325516109103e-05,
+      "loss": 1.4463,
+      "step": 1344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451698721801094e-05,
+      "loss": 1.3242,
+      "step": 1345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8449069884616983e-05,
+      "loss": 1.481,
+      "step": 1346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446439005192194e-05,
+      "loss": 1.3491,
+      "step": 1347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8443806084162638e-05,
+      "loss": 1.4561,
+      "step": 1348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844117112216473e-05,
+      "loss": 1.3613,
+      "step": 1349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438534119835365e-05,
+      "loss": 1.3613,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435895077811944e-05,
+      "loss": 1.4409,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433253996732346e-05,
+      "loss": 1.3906,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843061087723496e-05,
+      "loss": 1.4023,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8427965719958657e-05,
+      "loss": 1.2759,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84253185255428e-05,
+      "loss": 1.4824,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422669294627252e-05,
+      "loss": 1.417,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842001802785236e-05,
+      "loss": 1.4932,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8417364725858968e-05,
+      "loss": 1.5049,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841470938928841e-05,
+      "loss": 1.502,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412052018782517e-05,
+      "loss": 1.4092,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84093926149836e-05,
+      "loss": 1.3691,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8406731178534473e-05,
+      "loss": 1.292,
+      "step": 1362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840406771007843e-05,
+      "loss": 1.3604,
+      "step": 1363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840140221025928e-05,
+      "loss": 1.1597,
+      "step": 1364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8398734679721287e-05,
+      "loss": 1.46,
+      "step": 1365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839606511910923e-05,
+      "loss": 1.4326,
+      "step": 1366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839339352906838e-05,
+      "loss": 1.3638,
+      "step": 1367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390719910244487e-05,
+      "loss": 1.5244,
+      "step": 1368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83880442632838e-05,
+      "loss": 1.3945,
+      "step": 1369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385366588833048e-05,
+      "loss": 1.2549,
+      "step": 1370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8382686887539463e-05,
+      "loss": 1.4512,
+      "step": 1371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380005160050762e-05,
+      "loss": 1.4468,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377321407015142e-05,
+      "loss": 1.3687,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8374635629081305e-05,
+      "loss": 1.3872,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371947826898437e-05,
+      "loss": 1.3691,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369258001116202e-05,
+      "loss": 1.3516,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366566152384773e-05,
+      "loss": 1.374,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363872281354796e-05,
+      "loss": 1.5762,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361176388677416e-05,
+      "loss": 1.3086,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358478475004263e-05,
+      "loss": 1.3979,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355778540987446e-05,
+      "loss": 1.2461,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835307658727958e-05,
+      "loss": 1.3555,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350372614533753e-05,
+      "loss": 1.459,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347666623403558e-05,
+      "loss": 1.3325,
+      "step": 1384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834495861454305e-05,
+      "loss": 1.3228,
+      "step": 1385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83422485886068e-05,
+      "loss": 1.4541,
+      "step": 1386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8339536546249848e-05,
+      "loss": 1.3389,
+      "step": 1387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336822488127727e-05,
+      "loss": 1.4893,
+      "step": 1388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334106414896454e-05,
+      "loss": 1.5723,
+      "step": 1389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8331388327212543e-05,
+      "loss": 1.3242,
+      "step": 1390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328668225732985e-05,
+      "loss": 1.4727,
+      "step": 1391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325946111115262e-05,
+      "loss": 1.3975,
+      "step": 1392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8323221984017337e-05,
+      "loss": 1.4805,
+      "step": 1393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320495845097668e-05,
+      "loss": 1.3311,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317767695015195e-05,
+      "loss": 1.3369,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315037534429344e-05,
+      "loss": 1.4307,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312305364000023e-05,
+      "loss": 1.3921,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830957118438764e-05,
+      "loss": 1.4434,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830683499625307e-05,
+      "loss": 1.375,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304096800257683e-05,
+      "loss": 1.2954,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830135659706334e-05,
+      "loss": 1.3696,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829861438733237e-05,
+      "loss": 1.3345,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829587017172761e-05,
+      "loss": 1.5088,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8293123950912362e-05,
+      "loss": 1.3877,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 1.5469,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8287625496306063e-05,
+      "loss": 1.3979,
+      "step": 1406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828487326384406e-05,
+      "loss": 1.4727,
+      "step": 1407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828211902882965e-05,
+      "loss": 1.4082,
+      "step": 1408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279362791928568e-05,
+      "loss": 1.5117,
+      "step": 1409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276604553807032e-05,
+      "loss": 1.3599,
+      "step": 1410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827384431513174e-05,
+      "loss": 1.3521,
+      "step": 1411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271082076569867e-05,
+      "loss": 1.3691,
+      "step": 1412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826831783878909e-05,
+      "loss": 1.4355,
+      "step": 1413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826555160245755e-05,
+      "loss": 1.3125,
+      "step": 1414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8262783368243885e-05,
+      "loss": 1.375,
+      "step": 1415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260013136817204e-05,
+      "loss": 1.3311,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257240908847113e-05,
+      "loss": 1.4897,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254466685003683e-05,
+      "loss": 1.2251,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8251690465957482e-05,
+      "loss": 1.54,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248912252379555e-05,
+      "loss": 1.4756,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246132044941433e-05,
+      "loss": 1.5332,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8243349844315116e-05,
+      "loss": 1.1636,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240565651173103e-05,
+      "loss": 1.457,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8237779466188363e-05,
+      "loss": 1.5488,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234991290034353e-05,
+      "loss": 1.3071,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232201123385003e-05,
+      "loss": 1.3721,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229408966914734e-05,
+      "loss": 1.4395,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226614821298444e-05,
+      "loss": 1.5107,
+      "step": 1428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8223818687211504e-05,
+      "loss": 1.4443,
+      "step": 1429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221020565329784e-05,
+      "loss": 1.2739,
+      "step": 1430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218220456329618e-05,
+      "loss": 1.5713,
+      "step": 1431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215418360887817e-05,
+      "loss": 1.4238,
+      "step": 1432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212614279681698e-05,
+      "loss": 1.4414,
+      "step": 1433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209808213389024e-05,
+      "loss": 1.4067,
+      "step": 1434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820700016268807e-05,
+      "loss": 1.3623,
+      "step": 1435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204190128257563e-05,
+      "loss": 1.3984,
+      "step": 1436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8201378110776725e-05,
+      "loss": 1.375,
+      "step": 1437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819856411092526e-05,
+      "loss": 1.458,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8195748129383338e-05,
+      "loss": 1.4585,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192930166831618e-05,
+      "loss": 1.4531,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190110223951233e-05,
+      "loss": 1.2695,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81872883014238e-05,
+      "loss": 1.3765,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8184464399931414e-05,
+      "loss": 1.4126,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181638520156637e-05,
+      "loss": 1.2568,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8178810662782522e-05,
+      "loss": 1.2354,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175980828492598e-05,
+      "loss": 1.375,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817314901797087e-05,
+      "loss": 1.5078,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817031523190181e-05,
+      "loss": 1.377,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167479470970394e-05,
+      "loss": 1.5615,
+      "step": 1449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164641735862046e-05,
+      "loss": 1.4648,
+      "step": 1450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161802027262685e-05,
+      "loss": 1.3945,
+      "step": 1451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158960345858705e-05,
+      "loss": 1.3032,
+      "step": 1452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815611669233697e-05,
+      "loss": 1.4097,
+      "step": 1453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153271067384827e-05,
+      "loss": 1.3203,
+      "step": 1454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150423471690098e-05,
+      "loss": 1.2559,
+      "step": 1455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814757390594108e-05,
+      "loss": 1.4697,
+      "step": 1456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144722370826545e-05,
+      "loss": 1.2612,
+      "step": 1457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141868867035745e-05,
+      "loss": 1.582,
+      "step": 1458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139013395258403e-05,
+      "loss": 1.4258,
+      "step": 1459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136155956184724e-05,
+      "loss": 1.4062,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133296550505388e-05,
+      "loss": 1.3779,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813043517891154e-05,
+      "loss": 1.4648,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127571842094813e-05,
+      "loss": 1.4512,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124706540747303e-05,
+      "loss": 1.4795,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121839275561593e-05,
+      "loss": 1.4336,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811897004723074e-05,
+      "loss": 1.25,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116098856448256e-05,
+      "loss": 1.3125,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113225703908153e-05,
+      "loss": 1.4463,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110350590304903e-05,
+      "loss": 1.51,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107473516333458e-05,
+      "loss": 1.3066,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104594482689236e-05,
+      "loss": 1.2588,
+      "step": 1471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810171349006814e-05,
+      "loss": 1.457,
+      "step": 1472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098830539166536e-05,
+      "loss": 1.3784,
+      "step": 1473
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095945630681265e-05,
+      "loss": 1.4365,
+      "step": 1474
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809305876530965e-05,
+      "loss": 1.4038,
+      "step": 1475
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.3516,
+      "step": 1476
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808727916669901e-05,
+      "loss": 1.4756,
+      "step": 1477
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808438643485698e-05,
+      "loss": 1.417,
+      "step": 1478
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081491748922596e-05,
+      "loss": 1.4717,
+      "step": 1479
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8078595109595546e-05,
+      "loss": 1.3574,
+      "step": 1480
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075696517575975e-05,
+      "loss": 1.376,
+      "step": 1481
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072795973564505e-05,
+      "loss": 1.3828,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8069893478262238e-05,
+      "loss": 1.4639,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066989032370738e-05,
+      "loss": 1.3069,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8064082636592043e-05,
+      "loss": 1.5713,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061174291628668e-05,
+      "loss": 1.5146,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058263998183586e-05,
+      "loss": 1.5684,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055351756960262e-05,
+      "loss": 1.3672,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8052437568662614e-05,
+      "loss": 1.5127,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804952143399503e-05,
+      "loss": 1.3887,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804660335366238e-05,
+      "loss": 1.3662,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043683328370003e-05,
+      "loss": 1.3774,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040761358823694e-05,
+      "loss": 1.457,
+      "step": 1493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037837445729733e-05,
+      "loss": 1.5376,
+      "step": 1494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8034911589794868e-05,
+      "loss": 1.4492,
+      "step": 1495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031983791726303e-05,
+      "loss": 1.2949,
+      "step": 1496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8029054052231735e-05,
+      "loss": 1.4277,
+      "step": 1497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8026122372019307e-05,
+      "loss": 1.4238,
+      "step": 1498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802318875179764e-05,
+      "loss": 1.3462,
+      "step": 1499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8020253192275834e-05,
+      "loss": 1.3574,
+      "step": 1500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801731569416344e-05,
+      "loss": 1.4668,
+      "step": 1501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801437625817049e-05,
+      "loss": 1.4766,
+      "step": 1502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8011434885007482e-05,
+      "loss": 1.4062,
+      "step": 1503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800849157538538e-05,
+      "loss": 1.3662,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005546330015615e-05,
+      "loss": 1.2754,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8002599149610088e-05,
+      "loss": 1.5693,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799965003488117e-05,
+      "loss": 1.5781,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799669898654169e-05,
+      "loss": 1.459,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7993746005304963e-05,
+      "loss": 1.458,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799079109188475e-05,
+      "loss": 1.3682,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987834246995296e-05,
+      "loss": 1.3457,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79848754713513e-05,
+      "loss": 1.2969,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981914765667938e-05,
+      "loss": 1.2686,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978952130660843e-05,
+      "loss": 1.4717,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797598756704612e-05,
+      "loss": 1.541,
+      "step": 1515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7973021075540346e-05,
+      "loss": 1.4863,
+      "step": 1516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970052656860547e-05,
+      "loss": 1.4238,
+      "step": 1517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796708231172423e-05,
+      "loss": 1.415,
+      "step": 1518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796411004084936e-05,
+      "loss": 1.4922,
+      "step": 1519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961135844954375e-05,
+      "loss": 1.3154,
+      "step": 1520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7958159724758173e-05,
+      "loss": 1.4199,
+      "step": 1521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955181680980115e-05,
+      "loss": 1.3623,
+      "step": 1522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7952201714340032e-05,
+      "loss": 1.3828,
+      "step": 1523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7949219825558212e-05,
+      "loss": 1.3911,
+      "step": 1524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794623601535542e-05,
+      "loss": 1.4028,
+      "step": 1525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794325028445288e-05,
+      "loss": 1.2271,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940262633572267e-05,
+      "loss": 1.208,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793727306343574e-05,
+      "loss": 1.4346,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7934281574765912e-05,
+      "loss": 1.4268,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7931288168285863e-05,
+      "loss": 1.4463,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928292844719135e-05,
+      "loss": 1.3564,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7925295604789733e-05,
+      "loss": 1.3579,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792229644922212e-05,
+      "loss": 1.4805,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919295378741234e-05,
+      "loss": 1.4395,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791629239407247e-05,
+      "loss": 1.4243,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791328749594168e-05,
+      "loss": 1.4287,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910280685075188e-05,
+      "loss": 1.2905,
+      "step": 1537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7907271962199776e-05,
+      "loss": 1.2759,
+      "step": 1538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790426132804268e-05,
+      "loss": 1.4189,
+      "step": 1539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790124878333162e-05,
+      "loss": 1.2949,
+      "step": 1540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789823432879475e-05,
+      "loss": 1.3252,
+      "step": 1541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895217965160712e-05,
+      "loss": 1.3208,
+      "step": 1542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892199693158586e-05,
+      "loss": 1.3193,
+      "step": 1543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7889179513517933e-05,
+      "loss": 1.3887,
+      "step": 1544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886157426968764e-05,
+      "loss": 1.4243,
+      "step": 1545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883133434241546e-05,
+      "loss": 1.272,
+      "step": 1546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 1.3564,
+      "step": 1547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7877079733177185e-05,
+      "loss": 1.4688,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874050026303292e-05,
+      "loss": 1.4219,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7871018416177856e-05,
+      "loss": 1.313,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786798490353366e-05,
+      "loss": 1.332,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864949489103932e-05,
+      "loss": 1.3818,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861912173622373e-05,
+      "loss": 1.3457,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785887295782314e-05,
+      "loss": 1.502,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785583184244084e-05,
+      "loss": 1.5205,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785278882821055e-05,
+      "loss": 1.3887,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784974391586781e-05,
+      "loss": 1.5137,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846697106148602e-05,
+      "loss": 1.2168,
+      "step": 1558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843648399789378e-05,
+      "loss": 1.2812,
+      "step": 1559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7840597797527048e-05,
+      "loss": 1.5386,
+      "step": 1560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783754530009898e-05,
+      "loss": 1.5732,
+      "step": 1561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834490908243e-05,
+      "loss": 1.5103,
+      "step": 1562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831434622697386e-05,
+      "loss": 1.418,
+      "step": 1563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7828376444200884e-05,
+      "loss": 1.3838,
+      "step": 1564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825316373492687e-05,
+      "loss": 1.3682,
+      "step": 1565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7822254411312455e-05,
+      "loss": 1.4688,
+      "step": 1566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78191905584003e-05,
+      "loss": 1.5518,
+      "step": 1567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7816124815496785e-05,
+      "loss": 1.3135,
+      "step": 1568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7813057183342944e-05,
+      "loss": 1.4048,
+      "step": 1569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.780998766268026e-05,
+      "loss": 1.2817,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780691625425067e-05,
+      "loss": 1.3604,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780384295879657e-05,
+      "loss": 1.4214,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800767777060813e-05,
+      "loss": 1.2505,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7797690709786705e-05,
+      "loss": 1.4404,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794611757718013e-05,
+      "loss": 1.4814,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7791530921598958e-05,
+      "loss": 1.3857,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7788448202174207e-05,
+      "loss": 1.2119,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785363600188894e-05,
+      "loss": 1.2161,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7782277116388606e-05,
+      "loss": 1.4854,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779188751519382e-05,
+      "loss": 1.3818,
+      "step": 1580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7776098506327716e-05,
+      "loss": 1.502,
+      "step": 1581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773006381560557e-05,
+      "loss": 1.4604,
+      "step": 1582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769912377965305e-05,
+      "loss": 1.418,
+      "step": 1583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7766816496289822e-05,
+      "loss": 1.3706,
+      "step": 1584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763718737282417e-05,
+      "loss": 1.4268,
+      "step": 1585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760619101691858e-05,
+      "loss": 1.4443,
+      "step": 1586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757517590267365e-05,
+      "loss": 1.459,
+      "step": 1587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7754414203758602e-05,
+      "loss": 1.334,
+      "step": 1588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77513089429157e-05,
+      "loss": 1.3403,
+      "step": 1589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774820180848924e-05,
+      "loss": 1.3711,
+      "step": 1590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774509280123025e-05,
+      "loss": 1.2515,
+      "step": 1591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7741981921890214e-05,
+      "loss": 1.3994,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773886917122107e-05,
+      "loss": 1.4482,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735754549975204e-05,
+      "loss": 1.2852,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773263805890546e-05,
+      "loss": 1.397,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772951969876513e-05,
+      "loss": 1.3999,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7726399470307957e-05,
+      "loss": 1.3657,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723277374288142e-05,
+      "loss": 1.4146,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7720153411460327e-05,
+      "loss": 1.542,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717027582579612e-05,
+      "loss": 1.5264,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7713899888401552e-05,
+      "loss": 1.543,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710770329682145e-05,
+      "loss": 1.5215,
+      "step": 1602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770763890717784e-05,
+      "loss": 1.4121,
+      "step": 1603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770450562164554e-05,
+      "loss": 1.4419,
+      "step": 1604
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77013704738426e-05,
+      "loss": 1.332,
+      "step": 1605
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7698233464526824e-05,
+      "loss": 1.2964,
+      "step": 1606
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769509459445646e-05,
+      "loss": 1.373,
+      "step": 1607
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691953864390208e-05,
+      "loss": 1.417,
+      "step": 1608
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7688811275087227e-05,
+      "loss": 1.4268,
+      "step": 1609
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768566682730711e-05,
+      "loss": 1.4648,
+      "step": 1610
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7682520521809918e-05,
+      "loss": 1.4473,
+      "step": 1611
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679372359356143e-05,
+      "loss": 1.4404,
+      "step": 1612
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7676222340706733e-05,
+      "loss": 1.2407,
+      "step": 1613
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767307046662309e-05,
+      "loss": 1.2495,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7669916737867055e-05,
+      "loss": 1.375,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7666761155200924e-05,
+      "loss": 1.1221,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663603719387433e-05,
+      "loss": 1.46,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 1.4854,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76572832913716e-05,
+      "loss": 1.3535,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765412030069698e-05,
+      "loss": 1.3423,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7650955459930445e-05,
+      "loss": 1.2749,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764778876983698e-05,
+      "loss": 1.3848,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7644620231182015e-05,
+      "loss": 1.4365,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641449844731417e-05,
+      "loss": 1.2422,
+      "step": 1624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763827761125151e-05,
+      "loss": 1.3926,
+      "step": 1625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635103531509056e-05,
+      "loss": 1.4775,
+      "step": 1626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631927606271276e-05,
+      "loss": 1.3193,
+      "step": 1627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762874983630582e-05,
+      "loss": 1.312,
+      "step": 1628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76255702223808e-05,
+      "loss": 1.3213,
+      "step": 1629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762238876526476e-05,
+      "loss": 1.1946,
+      "step": 1630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7619205465726703e-05,
+      "loss": 1.4248,
+      "step": 1631
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616020324536066e-05,
+      "loss": 1.3198,
+      "step": 1632
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761283334246273e-05,
+      "loss": 1.375,
+      "step": 1633
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609644520277044e-05,
+      "loss": 1.4307,
+      "step": 1634
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606453858749764e-05,
+      "loss": 1.3262,
+      "step": 1635
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760326135865212e-05,
+      "loss": 1.5205,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760006702075578e-05,
+      "loss": 1.4814,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759687084583285e-05,
+      "loss": 1.5488,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7593672834655886e-05,
+      "loss": 1.5928,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759047298799788e-05,
+      "loss": 1.4268,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587271306632277e-05,
+      "loss": 1.2646,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758406779133296e-05,
+      "loss": 1.311,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758086244287426e-05,
+      "loss": 1.3047,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7577655262030946e-05,
+      "loss": 1.3379,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757444624957823e-05,
+      "loss": 1.4448,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571235406291765e-05,
+      "loss": 1.4932,
+      "step": 1646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756802273294766e-05,
+      "loss": 1.3418,
+      "step": 1647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756480823032245e-05,
+      "loss": 1.5557,
+      "step": 1648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561591899193118e-05,
+      "loss": 1.3086,
+      "step": 1649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7558373740337093e-05,
+      "loss": 1.4395,
+      "step": 1650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555153754532242e-05,
+      "loss": 1.4697,
+      "step": 1651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755193194255687e-05,
+      "loss": 1.3633,
+      "step": 1652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548708305189724e-05,
+      "loss": 1.4512,
+      "step": 1653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7545482843210007e-05,
+      "loss": 1.3809,
+      "step": 1654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542255557397343e-05,
+      "loss": 1.2471,
+      "step": 1655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753902644853181e-05,
+      "loss": 1.25,
+      "step": 1656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753579551739392e-05,
+      "loss": 1.2959,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7532562764764623e-05,
+      "loss": 1.4248,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752932819142532e-05,
+      "loss": 1.375,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752609179815785e-05,
+      "loss": 1.418,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7522853585744472e-05,
+      "loss": 1.4551,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519613554967913e-05,
+      "loss": 1.4043,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516371706611325e-05,
+      "loss": 1.542,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513128041458302e-05,
+      "loss": 1.3101,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750988256029287e-05,
+      "loss": 1.2651,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7506635263899513e-05,
+      "loss": 1.4902,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750338615306313e-05,
+      "loss": 1.4209,
+      "step": 1667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7500135228569067e-05,
+      "loss": 1.4453,
+      "step": 1668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496882491203123e-05,
+      "loss": 1.4321,
+      "step": 1669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7493627941751517e-05,
+      "loss": 1.377,
+      "step": 1670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490371581000914e-05,
+      "loss": 1.5117,
+      "step": 1671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748711340973841e-05,
+      "loss": 1.3721,
+      "step": 1672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7483853428751552e-05,
+      "loss": 1.2874,
+      "step": 1673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748059163882831e-05,
+      "loss": 1.4873,
+      "step": 1674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477328040757097e-05,
+      "loss": 1.1401,
+      "step": 1675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747406263532677e-05,
+      "loss": 1.5576,
+      "step": 1676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747079542332661e-05,
+      "loss": 1.3438,
+      "step": 1677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7467526405546344e-05,
+      "loss": 1.4307,
+      "step": 1678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464255582776132e-05,
+      "loss": 1.436,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7460982955806565e-05,
+      "loss": 1.4463,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745770852542869e-05,
+      "loss": 1.3672,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745443229243396e-05,
+      "loss": 1.3066,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7451154257614287e-05,
+      "loss": 1.2197,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447874421762013e-05,
+      "loss": 1.4707,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744459278566991e-05,
+      "loss": 1.3135,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441309350131186e-05,
+      "loss": 1.4229,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438024115939495e-05,
+      "loss": 1.2739,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434737083888905e-05,
+      "loss": 1.3691,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 1.4678,
+      "step": 1689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428157629389552e-05,
+      "loss": 1.3975,
+      "step": 1690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424865208531117e-05,
+      "loss": 1.3931,
+      "step": 1691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742157099299445e-05,
+      "loss": 1.585,
+      "step": 1692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418274983575808e-05,
+      "loss": 1.4199,
+      "step": 1693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741497718107188e-05,
+      "loss": 1.3125,
+      "step": 1694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741167758627977e-05,
+      "loss": 1.3594,
+      "step": 1695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740837619999704e-05,
+      "loss": 1.4668,
+      "step": 1696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740507302302167e-05,
+      "loss": 1.4561,
+      "step": 1697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401768056152083e-05,
+      "loss": 1.5117,
+      "step": 1698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398461300187124e-05,
+      "loss": 1.5,
+      "step": 1699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7395152755926074e-05,
+      "loss": 1.4062,
+      "step": 1700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391842424168652e-05,
+      "loss": 1.3003,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7388530305715e-05,
+      "loss": 1.0669,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385216401365696e-05,
+      "loss": 1.4814,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381900711921757e-05,
+      "loss": 1.3789,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378583238184618e-05,
+      "loss": 1.333,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375263980956152e-05,
+      "loss": 1.3652,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737194294103867e-05,
+      "loss": 1.4077,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368620119234896e-05,
+      "loss": 1.1646,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365295516348002e-05,
+      "loss": 1.4824,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361969133181585e-05,
+      "loss": 1.2134,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735864097053967e-05,
+      "loss": 1.4854,
+      "step": 1711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7355311029226712e-05,
+      "loss": 1.3423,
+      "step": 1712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351979310047603e-05,
+      "loss": 1.3682,
+      "step": 1713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734864581380765e-05,
+      "loss": 1.4971,
+      "step": 1714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7345310541312606e-05,
+      "loss": 1.4707,
+      "step": 1715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734197349336864e-05,
+      "loss": 1.416,
+      "step": 1716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338634670782365e-05,
+      "loss": 1.4629,
+      "step": 1717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7335294074360806e-05,
+      "loss": 1.3477,
+      "step": 1718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331951704911426e-05,
+      "loss": 1.4209,
+      "step": 1719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732860756324212e-05,
+      "loss": 1.5811,
+      "step": 1720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.73252616501612e-05,
+      "loss": 1.3706,
+      "step": 1721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732191396647742e-05,
+      "loss": 1.3867,
+      "step": 1722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731856451299995e-05,
+      "loss": 1.4922,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7315213290538398e-05,
+      "loss": 1.4189,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731186029990279e-05,
+      "loss": 1.3696,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7308505541903584e-05,
+      "loss": 1.4463,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305149017351662e-05,
+      "loss": 1.4229,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7301790727058344e-05,
+      "loss": 1.3027,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298430671835362e-05,
+      "loss": 1.3418,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295068852494886e-05,
+      "loss": 1.3672,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291705269849504e-05,
+      "loss": 1.2065,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7288339924712233e-05,
+      "loss": 1.3687,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284972817896523e-05,
+      "loss": 1.4385,
+      "step": 1733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281603950216233e-05,
+      "loss": 1.3867,
+      "step": 1734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278233322485674e-05,
+      "loss": 1.4502,
+      "step": 1735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7274860935519552e-05,
+      "loss": 1.4707,
+      "step": 1736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271486790133022e-05,
+      "loss": 1.5957,
+      "step": 1737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726811088714165e-05,
+      "loss": 1.4316,
+      "step": 1738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264733227361435e-05,
+      "loss": 1.3027,
+      "step": 1739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261353811608798e-05,
+      "loss": 1.5195,
+      "step": 1740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257972640700583e-05,
+      "loss": 1.375,
+      "step": 1741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725458971545406e-05,
+      "loss": 1.3926,
+      "step": 1742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251205036686913e-05,
+      "loss": 1.4185,
+      "step": 1743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7247818605217276e-05,
+      "loss": 1.5342,
+      "step": 1744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7244430421863674e-05,
+      "loss": 1.4204,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7241040487445082e-05,
+      "loss": 1.5254,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7237648802780884e-05,
+      "loss": 1.4248,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723425536869089e-05,
+      "loss": 1.5801,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7230860185995332e-05,
+      "loss": 1.2134,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722746325551487e-05,
+      "loss": 1.4863,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722406457807058e-05,
+      "loss": 1.3076,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722066415448396e-05,
+      "loss": 1.499,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217261985576938e-05,
+      "loss": 1.3926,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7213858072171855e-05,
+      "loss": 1.3125,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7210452415091477e-05,
+      "loss": 1.374,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207045015158996e-05,
+      "loss": 1.374,
+      "step": 1756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720363587319802e-05,
+      "loss": 1.4736,
+      "step": 1757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7200224990032577e-05,
+      "loss": 1.3262,
+      "step": 1758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7196812366487115e-05,
+      "loss": 1.4614,
+      "step": 1759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193398003386514e-05,
+      "loss": 1.4619,
+      "step": 1760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718998190155606e-05,
+      "loss": 1.4058,
+      "step": 1761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7186564061821466e-05,
+      "loss": 1.374,
+      "step": 1762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7183144485008865e-05,
+      "loss": 1.4209,
+      "step": 1763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179723171944812e-05,
+      "loss": 1.1685,
+      "step": 1764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717630012345628e-05,
+      "loss": 1.3848,
+      "step": 1765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717287534037065e-05,
+      "loss": 1.4141,
+      "step": 1766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169448823515745e-05,
+      "loss": 1.2524,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7166020573719792e-05,
+      "loss": 1.3906,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7162590591811437e-05,
+      "loss": 1.4038,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159158878619746e-05,
+      "loss": 1.3491,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7155725434974208e-05,
+      "loss": 1.375,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152290261704728e-05,
+      "loss": 1.4189,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7148853359641627e-05,
+      "loss": 1.1943,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7145414729615643e-05,
+      "loss": 1.3867,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7141974372457936e-05,
+      "loss": 1.499,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138532289000082e-05,
+      "loss": 1.3608,
+      "step": 1776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713508848007407e-05,
+      "loss": 1.2661,
+      "step": 1777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131642946512314e-05,
+      "loss": 1.5488,
+      "step": 1778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712819568914764e-05,
+      "loss": 1.4199,
+      "step": 1779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124746708813286e-05,
+      "loss": 1.415,
+      "step": 1780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.712129600634292e-05,
+      "loss": 1.4111,
+      "step": 1781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117843582570608e-05,
+      "loss": 1.5215,
+      "step": 1782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7114389438330846e-05,
+      "loss": 1.2104,
+      "step": 1783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110933574458547e-05,
+      "loss": 1.4087,
+      "step": 1784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107475991789028e-05,
+      "loss": 1.4092,
+      "step": 1785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710401669115803e-05,
+      "loss": 1.4863,
+      "step": 1786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100555673401704e-05,
+      "loss": 1.354,
+      "step": 1787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097092939356622e-05,
+      "loss": 1.3325,
+      "step": 1788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7093628489859772e-05,
+      "loss": 1.3965,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090162325748547e-05,
+      "loss": 1.2969,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.708669444786076e-05,
+      "loss": 1.4775,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7083224857034643e-05,
+      "loss": 1.2036,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7079753554108832e-05,
+      "loss": 1.5059,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707628053992239e-05,
+      "loss": 1.3892,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072805815314776e-05,
+      "loss": 1.4053,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7069329381125875e-05,
+      "loss": 1.3472,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7065851238195986e-05,
+      "loss": 1.5684,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062371387365817e-05,
+      "loss": 1.4502,
+      "step": 1798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705888982947649e-05,
+      "loss": 1.5068,
+      "step": 1799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705540656536953e-05,
+      "loss": 1.3887,
+      "step": 1800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.70519215958869e-05,
+      "loss": 1.4795,
+      "step": 1801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704843492187094e-05,
+      "loss": 1.5322,
+      "step": 1802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7044946544164434e-05,
+      "loss": 1.292,
+      "step": 1803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704145646361056e-05,
+      "loss": 1.5049,
+      "step": 1804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703796468105291e-05,
+      "loss": 1.4473,
+      "step": 1805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034471197335493e-05,
+      "loss": 1.5127,
+      "step": 1806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030976013302725e-05,
+      "loss": 1.375,
+      "step": 1807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702747912979943e-05,
+      "loss": 1.4199,
+      "step": 1808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702398054767085e-05,
+      "loss": 1.2588,
+      "step": 1809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702048026776263e-05,
+      "loss": 1.3999,
+      "step": 1810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016978290920835e-05,
+      "loss": 1.3799,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701347461799193e-05,
+      "loss": 1.5039,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7009969249822796e-05,
+      "loss": 1.3745,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006462187260723e-05,
+      "loss": 1.4404,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700295343115341e-05,
+      "loss": 1.3325,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6999442982348958e-05,
+      "loss": 1.3813,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995930841695894e-05,
+      "loss": 1.4531,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992417010043144e-05,
+      "loss": 1.3467,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6988901488240032e-05,
+      "loss": 1.3555,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985384277136314e-05,
+      "loss": 1.4092,
+      "step": 1820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6981865377582133e-05,
+      "loss": 1.4863,
+      "step": 1821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6978344790428053e-05,
+      "loss": 1.3408,
+      "step": 1822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974822516525044e-05,
+      "loss": 1.3823,
+      "step": 1823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6971298556724472e-05,
+      "loss": 1.3379,
+      "step": 1824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696777291187813e-05,
+      "loss": 1.3281,
+      "step": 1825
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964245582838207e-05,
+      "loss": 1.3491,
+      "step": 1826
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6960716570457295e-05,
+      "loss": 1.3311,
+      "step": 1827
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6957185875588403e-05,
+      "loss": 1.3093,
+      "step": 1828
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953653499084936e-05,
+      "loss": 1.2378,
+      "step": 1829
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695011944180072e-05,
+      "loss": 1.3662,
+      "step": 1830
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 1.4102,
+      "step": 1831
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943046288307328e-05,
+      "loss": 1.3037,
+      "step": 1832
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6939507193807818e-05,
+      "loss": 1.3521,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6935966421946883e-05,
+      "loss": 1.459,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932423973580375e-05,
+      "loss": 1.3203,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692887984956454e-05,
+      "loss": 1.5459,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692533405075604e-05,
+      "loss": 1.5098,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692178657801193e-05,
+      "loss": 1.5508,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6918237432189682e-05,
+      "loss": 1.397,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691468661414717e-05,
+      "loss": 1.4033,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911134124742655e-05,
+      "loss": 1.5605,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6907579964834833e-05,
+      "loss": 1.4043,
+      "step": 1842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6904024135282778e-05,
+      "loss": 1.228,
+      "step": 1843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6900466636945973e-05,
+      "loss": 1.4082,
+      "step": 1844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896907470684316e-05,
+      "loss": 1.4883,
+      "step": 1845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893346637358092e-05,
+      "loss": 1.4062,
+      "step": 1846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6889784137828004e-05,
+      "loss": 1.4854,
+      "step": 1847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688621997295515e-05,
+      "loss": 1.2959,
+      "step": 1848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688265414360102e-05,
+      "loss": 1.2842,
+      "step": 1849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687908665062753e-05,
+      "loss": 1.3887,
+      "step": 1850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6875517494896978e-05,
+      "loss": 1.4331,
+      "step": 1851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871946677272077e-05,
+      "loss": 1.3701,
+      "step": 1852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868374198615933e-05,
+      "loss": 1.3291,
+      "step": 1853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864800059792057e-05,
+      "loss": 1.2559,
+      "step": 1854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6861224261664357e-05,
+      "loss": 1.3477,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6857646805097154e-05,
+      "loss": 1.3682,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6854067690955155e-05,
+      "loss": 1.4844,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685048692010348e-05,
+      "loss": 1.5205,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684690449340764e-05,
+      "loss": 1.1943,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684332041173355e-05,
+      "loss": 1.4131,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6839734675947525e-05,
+      "loss": 1.4277,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683614728691628e-05,
+      "loss": 1.373,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6832558245506937e-05,
+      "loss": 1.3984,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828967552587002e-05,
+      "loss": 1.5557,
+      "step": 1864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825375209024385e-05,
+      "loss": 1.5391,
+      "step": 1865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821781215687406e-05,
+      "loss": 1.2876,
+      "step": 1866
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818185573444775e-05,
+      "loss": 1.499,
+      "step": 1867
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814588283165595e-05,
+      "loss": 1.3184,
+      "step": 1868
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810989345719383e-05,
+      "loss": 1.3301,
+      "step": 1869
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807388761976038e-05,
+      "loss": 1.5127,
+      "step": 1870
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6803786532805867e-05,
+      "loss": 1.4463,
+      "step": 1871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800182659079568e-05,
+      "loss": 1.3374,
+      "step": 1872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796577141668247e-05,
+      "loss": 1.4023,
+      "step": 1873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6792969981443397e-05,
+      "loss": 1.3525,
+      "step": 1874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6789361179276907e-05,
+      "loss": 1.3418,
+      "step": 1875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785750736041075e-05,
+      "loss": 1.4443,
+      "step": 1876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782138652608583e-05,
+      "loss": 1.3232,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6778524929852513e-05,
+      "loss": 1.4678,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774909568646352e-05,
+      "loss": 1.4146,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677129256986397e-05,
+      "loss": 1.3467,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676767393437964e-05,
+      "loss": 1.5029,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764053663068032e-05,
+      "loss": 1.4932,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6760431756804204e-05,
+      "loss": 1.4219,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756808216463617e-05,
+      "loss": 1.4648,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753183042922118e-05,
+      "loss": 1.5176,
+      "step": 1885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749556237055963e-05,
+      "loss": 1.5732,
+      "step": 1886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674592779974179e-05,
+      "loss": 1.5771,
+      "step": 1887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674229773185664e-05,
+      "loss": 1.3848,
+      "step": 1888
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6738666034277932e-05,
+      "loss": 1.3721,
+      "step": 1889
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6735032707883502e-05,
+      "loss": 1.6348,
+      "step": 1890
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673139775355156e-05,
+      "loss": 1.3408,
+      "step": 1891
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727761172160727e-05,
+      "loss": 1.3447,
+      "step": 1892
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6724122964590002e-05,
+      "loss": 1.6309,
+      "step": 1893
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6720483131718784e-05,
+      "loss": 1.2632,
+      "step": 1894
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671684167442686e-05,
+      "loss": 1.4395,
+      "step": 1895
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671319859359442e-05,
+      "loss": 1.3838,
+      "step": 1896
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670955389010204e-05,
+      "loss": 1.2568,
+      "step": 1897
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670590756483068e-05,
+      "loss": 1.374,
+      "step": 1898
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670225961866171e-05,
+      "loss": 1.3037,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698610052476873e-05,
+      "loss": 1.4639,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694958867158318e-05,
+      "loss": 1.3452,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 1.2734,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687651642650587e-05,
+      "loss": 1.3105,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668399560522765e-05,
+      "loss": 1.4307,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6680337952203483e-05,
+      "loss": 1.6045,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667667868446218e-05,
+      "loss": 1.4712,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673017802888227e-05,
+      "loss": 1.3799,
+      "step": 1907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666935530836651e-05,
+      "loss": 1.334,
+      "step": 1908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665691201782293e-05,
+      "loss": 1.4785,
+      "step": 1909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662025484021232e-05,
+      "loss": 1.4307,
+      "step": 1910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658358155969382e-05,
+      "loss": 1.377,
+      "step": 1911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665468921851317e-05,
+      "loss": 1.4121,
+      "step": 1912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665101867253943e-05,
+      "loss": 1.3945,
+      "step": 1913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647346518935374e-05,
+      "loss": 1.4922,
+      "step": 1914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6643672758588604e-05,
+      "loss": 1.3867,
+      "step": 1915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639997392387113e-05,
+      "loss": 1.5352,
+      "step": 1916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663632042121928e-05,
+      "loss": 1.4307,
+      "step": 1917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663264184597387e-05,
+      "loss": 1.334,
+      "step": 1918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6628961667540045e-05,
+      "loss": 1.3115,
+      "step": 1919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6625279886807342e-05,
+      "loss": 1.3213,
+      "step": 1920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621596504665695e-05,
+      "loss": 1.4062,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617911522005418e-05,
+      "loss": 1.2666,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6614224939717217e-05,
+      "loss": 1.3818,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610536758692183e-05,
+      "loss": 1.4229,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6606846979821795e-05,
+      "loss": 1.4414,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6603155603997908e-05,
+      "loss": 1.5762,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6599462632112782e-05,
+      "loss": 1.2832,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659576806505905e-05,
+      "loss": 1.3105,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6592071903729725e-05,
+      "loss": 1.4502,
+      "step": 1929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588374149018223e-05,
+      "loss": 1.396,
+      "step": 1930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658467480181833e-05,
+      "loss": 1.2354,
+      "step": 1931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658097386302422e-05,
+      "loss": 1.4277,
+      "step": 1932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577271333530463e-05,
+      "loss": 1.4814,
+      "step": 1933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6573567214231996e-05,
+      "loss": 1.4375,
+      "step": 1934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656986150602415e-05,
+      "loss": 1.3052,
+      "step": 1935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656615420980264e-05,
+      "loss": 1.5605,
+      "step": 1936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6562445326463566e-05,
+      "loss": 1.2529,
+      "step": 1937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558734856903406e-05,
+      "loss": 1.3638,
+      "step": 1938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655502280201902e-05,
+      "loss": 1.4355,
+      "step": 1939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6551309162707667e-05,
+      "loss": 1.5479,
+      "step": 1940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6547593939866974e-05,
+      "loss": 1.3896,
+      "step": 1941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654387713439495e-05,
+      "loss": 1.3271,
+      "step": 1942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6540158747189992e-05,
+      "loss": 1.4473,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653643877915088e-05,
+      "loss": 1.4512,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532717231176777e-05,
+      "loss": 1.1421,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652899410416722e-05,
+      "loss": 1.3413,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525269399022142e-05,
+      "loss": 1.4395,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652154311664184e-05,
+      "loss": 1.2622,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517815257927e-05,
+      "loss": 1.5469,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514085823778695e-05,
+      "loss": 1.4072,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510354815098373e-05,
+      "loss": 1.3682,
+      "step": 1951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650662223278786e-05,
+      "loss": 1.5625,
+      "step": 1952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650288807774937e-05,
+      "loss": 1.2065,
+      "step": 1953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649915235088549e-05,
+      "loss": 1.3643,
+      "step": 1954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649541505309919e-05,
+      "loss": 1.3799,
+      "step": 1955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6491676185293818e-05,
+      "loss": 1.3579,
+      "step": 1956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6487935748373102e-05,
+      "loss": 1.4033,
+      "step": 1957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484193743241153e-05,
+      "loss": 1.4512,
+      "step": 1958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648045017080246e-05,
+      "loss": 1.4912,
+      "step": 1959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6476705031961885e-05,
+      "loss": 1.3867,
+      "step": 1960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647295832762467e-05,
+      "loss": 1.5088,
+      "step": 1961
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469210058696448e-05,
+      "loss": 1.4834,
+      "step": 1962
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646546022608321e-05,
+      "loss": 1.3184,
+      "step": 1963
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646170883069134e-05,
+      "loss": 1.5137,
+      "step": 1964
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6457955873427593e-05,
+      "loss": 1.3447,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6454201355199108e-05,
+      "loss": 1.2739,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645044527691339e-05,
+      "loss": 1.3447,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644668763947833e-05,
+      "loss": 1.1619,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442928443802196e-05,
+      "loss": 1.3809,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643916769079363e-05,
+      "loss": 1.3252,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6435405381361645e-05,
+      "loss": 1.4385,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6431641516415645e-05,
+      "loss": 1.2671,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 1.2383,
+      "step": 1973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424109123621042e-05,
+      "loss": 1.4854,
+      "step": 1974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642034059759311e-05,
+      "loss": 1.396,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416570519692497e-05,
+      "loss": 1.4072,
+      "step": 1976
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412798890830477e-05,
+      "loss": 1.4854,
+      "step": 1977
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640902571191869e-05,
+      "loss": 1.3462,
+      "step": 1978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640525098386917e-05,
+      "loss": 1.1577,
+      "step": 1979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64014747075943e-05,
+      "loss": 1.4497,
+      "step": 1980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639769688400686e-05,
+      "loss": 1.4658,
+      "step": 1981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639391751401999e-05,
+      "loss": 1.3936,
+      "step": 1982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639013659854722e-05,
+      "loss": 1.3486,
+      "step": 1983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6386354138502423e-05,
+      "loss": 1.354,
+      "step": 1984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382570134799885e-05,
+      "loss": 1.3315,
+      "step": 1985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378784588354228e-05,
+      "loss": 1.4209,
+      "step": 1986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374997500080473e-05,
+      "loss": 1.5557,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6371208870894003e-05,
+      "loss": 1.5044,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367418701710572e-05,
+      "loss": 1.4775,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6363626993446314e-05,
+      "loss": 1.4282,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359833747017725e-05,
+      "loss": 1.3306,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356038963341675e-05,
+      "loss": 1.3643,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352242643335415e-05,
+      "loss": 1.4795,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634844478791656e-05,
+      "loss": 1.3496,
+      "step": 1994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6344645398003095e-05,
+      "loss": 1.4155,
+      "step": 1995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634084447451338e-05,
+      "loss": 1.4644,
+      "step": 1996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6337042018366135e-05,
+      "loss": 1.3066,
+      "step": 1997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6333238030480473e-05,
+      "loss": 1.4082,
+      "step": 1998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632943251177585e-05,
+      "loss": 1.4609,
+      "step": 1999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6325625463172113e-05,
+      "loss": 1.4404,
+      "step": 2000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6321816885589464e-05,
+      "loss": 1.3999,
+      "step": 2001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631800677994849e-05,
+      "loss": 1.4458,
+      "step": 2002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631419514717013e-05,
+      "loss": 1.2476,
+      "step": 2003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310381988175714e-05,
+      "loss": 1.3193,
+      "step": 2004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6306567303886914e-05,
+      "loss": 1.3901,
+      "step": 2005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302751095225794e-05,
+      "loss": 1.458,
+      "step": 2006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629893336311477e-05,
+      "loss": 1.3745,
+      "step": 2007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295114108476634e-05,
+      "loss": 1.5405,
+      "step": 2008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291293332234554e-05,
+      "loss": 1.3994,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287471035312048e-05,
+      "loss": 1.3413,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6283647218633013e-05,
+      "loss": 1.3633,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6279821883121708e-05,
+      "loss": 1.2896,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627599502970277e-05,
+      "loss": 1.4951,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6272166659301186e-05,
+      "loss": 1.3267,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6268336772842327e-05,
+      "loss": 1.5273,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626450537125192e-05,
+      "loss": 1.3262,
+      "step": 2016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626067245545606e-05,
+      "loss": 1.2676,
+      "step": 2017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256838026381197e-05,
+      "loss": 1.2822,
+      "step": 2018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6253002084954173e-05,
+      "loss": 1.2964,
+      "step": 2019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6249164632102177e-05,
+      "loss": 1.3281,
+      "step": 2020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245325668752766e-05,
+      "loss": 1.4473,
+      "step": 2021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624148519583386e-05,
+      "loss": 1.4502,
+      "step": 2022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.623764321427375e-05,
+      "loss": 1.5522,
+      "step": 2023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233799725001094e-05,
+      "loss": 1.4785,
+      "step": 2024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6229954728944895e-05,
+      "loss": 1.334,
+      "step": 2025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622610822703455e-05,
+      "loss": 1.5479,
+      "step": 2026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222260220199794e-05,
+      "loss": 1.5371,
+      "step": 2027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6218410709370735e-05,
+      "loss": 1.4268,
+      "step": 2028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6214559695477855e-05,
+      "loss": 1.2241,
+      "step": 2029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210707179451983e-05,
+      "loss": 1.3921,
+      "step": 2030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620685316222432e-05,
+      "loss": 1.4321,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202997644726425e-05,
+      "loss": 1.5049,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199140627890228e-05,
+      "loss": 1.291,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195282112648007e-05,
+      "loss": 1.127,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619142209993242e-05,
+      "loss": 1.2725,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187560590676475e-05,
+      "loss": 1.2612,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183697585813544e-05,
+      "loss": 1.5352,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617983308627736e-05,
+      "loss": 1.2734,
+      "step": 2038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617596709300202e-05,
+      "loss": 1.3042,
+      "step": 2039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617209960692198e-05,
+      "loss": 1.4678,
+      "step": 2040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6168230628972058e-05,
+      "loss": 1.2378,
+      "step": 2041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164360160087436e-05,
+      "loss": 1.4512,
+      "step": 2042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6160488201203643e-05,
+      "loss": 1.3203,
+      "step": 2043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156614753256583e-05,
+      "loss": 1.2983,
+      "step": 2044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6152739817182516e-05,
+      "loss": 1.3882,
+      "step": 2045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614886339391806e-05,
+      "loss": 1.4678,
+      "step": 2046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144985484400187e-05,
+      "loss": 1.376,
+      "step": 2047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614110608956624e-05,
+      "loss": 1.2285,
+      "step": 2048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613722521035391e-05,
+      "loss": 1.2803,
+      "step": 2049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6133342847701255e-05,
+      "loss": 1.5654,
+      "step": 2050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612945900254669e-05,
+      "loss": 1.3867,
+      "step": 2051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6125573675828984e-05,
+      "loss": 1.3721,
+      "step": 2052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6121686868487262e-05,
+      "loss": 1.3662,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6117798581461018e-05,
+      "loss": 1.3838,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6113908815690096e-05,
+      "loss": 1.4697,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.61100175721147e-05,
+      "loss": 1.2939,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6106124851675386e-05,
+      "loss": 1.1826,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6102230655313076e-05,
+      "loss": 1.3696,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6098334983969035e-05,
+      "loss": 1.3921,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094437838584906e-05,
+      "loss": 1.3911,
+      "step": 2060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6090539220102658e-05,
+      "loss": 1.4058,
+      "step": 2061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608663912946465e-05,
+      "loss": 1.3936,
+      "step": 2062
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082737567613572e-05,
+      "loss": 1.4341,
+      "step": 2063
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6078834535492477e-05,
+      "loss": 1.196,
+      "step": 2064
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6074930034044785e-05,
+      "loss": 1.5342,
+      "step": 2065
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6071024064214243e-05,
+      "loss": 1.4131,
+      "step": 2066
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6067116626944983e-05,
+      "loss": 1.457,
+      "step": 2067
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063207723181475e-05,
+      "loss": 1.3706,
+      "step": 2068
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605929735386855e-05,
+      "loss": 1.6221,
+      "step": 2069
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605538551995139e-05,
+      "loss": 1.2646,
+      "step": 2070
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051472222375528e-05,
+      "loss": 1.3667,
+      "step": 2071
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604755746208686e-05,
+      "loss": 1.3555,
+      "step": 2072
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6043641240031623e-05,
+      "loss": 1.4336,
+      "step": 2073
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603972355715642e-05,
+      "loss": 1.4521,
+      "step": 2074
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6035804414408204e-05,
+      "loss": 1.4883,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603188381273427e-05,
+      "loss": 1.6113,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027961753082282e-05,
+      "loss": 1.5107,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6024038236400246e-05,
+      "loss": 1.4951,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020113263636523e-05,
+      "loss": 1.5176,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6016186835739822e-05,
+      "loss": 1.3066,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601225895365921e-05,
+      "loss": 1.5928,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.60083296183441e-05,
+      "loss": 1.3711,
+      "step": 2082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004398830744266e-05,
+      "loss": 1.3027,
+      "step": 2083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000466591809822e-05,
+      "loss": 1.1909,
+      "step": 2084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5996532902491236e-05,
+      "loss": 1.397,
+      "step": 2085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992597763739327e-05,
+      "loss": 1.3711,
+      "step": 2086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988661176505266e-05,
+      "loss": 1.4102,
+      "step": 2087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5984723141740578e-05,
+      "loss": 1.3887,
+      "step": 2088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598078366039712e-05,
+      "loss": 1.3438,
+      "step": 2089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5976842733427127e-05,
+      "loss": 1.3477,
+      "step": 2090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5972900361783158e-05,
+      "loss": 1.4062,
+      "step": 2091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596895654641813e-05,
+      "loss": 1.4043,
+      "step": 2092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965011288285316e-05,
+      "loss": 1.2383,
+      "step": 2093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596106458833833e-05,
+      "loss": 1.376,
+      "step": 2094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957116447531132e-05,
+      "loss": 1.5039,
+      "step": 2095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953166866818043e-05,
+      "loss": 1.3633,
+      "step": 2096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949215847153718e-05,
+      "loss": 1.4512,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5945263389493166e-05,
+      "loss": 1.3291,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941309494791742e-05,
+      "loss": 1.2759,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5937354164005154e-05,
+      "loss": 1.333,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933397398089447e-05,
+      "loss": 1.3096,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592943919800102e-05,
+      "loss": 1.542,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592547956469662e-05,
+      "loss": 1.5215,
+      "step": 2103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5921518499133333e-05,
+      "loss": 1.4209,
+      "step": 2104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59175560022686e-05,
+      "loss": 1.5469,
+      "step": 2105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59135920750602e-05,
+      "loss": 1.4199,
+      "step": 2106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909626718466264e-05,
+      "loss": 1.2969,
+      "step": 2107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.590565993344526e-05,
+      "loss": 1.4609,
+      "step": 2108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901691720956016e-05,
+      "loss": 1.4414,
+      "step": 2109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589772208195769e-05,
+      "loss": 1.2466,
+      "step": 2110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5893751017409793e-05,
+      "loss": 1.373,
+      "step": 2111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588977852827218e-05,
+      "loss": 1.3467,
+      "step": 2112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885804615505047e-05,
+      "loss": 1.3232,
+      "step": 2113
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5881829280068934e-05,
+      "loss": 1.3174,
+      "step": 2114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 1.2051,
+      "step": 2115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873874345033666e-05,
+      "loss": 1.291,
+      "step": 2116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869894747357314e-05,
+      "loss": 1.3721,
+      "step": 2117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865913730857583e-05,
+      "loss": 1.2642,
+      "step": 2118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586193129649674e-05,
+      "loss": 1.3154,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5857947445237385e-05,
+      "loss": 1.2583,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585396217804246e-05,
+      "loss": 1.332,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5849975495875257e-05,
+      "loss": 1.3491,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5845987399699396e-05,
+      "loss": 1.3989,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841997890478853e-05,
+      "loss": 1.4385,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583800696917794e-05,
+      "loss": 1.3701,
+      "step": 2125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5834014636761307e-05,
+      "loss": 1.3384,
+      "step": 2126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583002089419395e-05,
+      "loss": 1.4043,
+      "step": 2127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582602574244121e-05,
+      "loss": 1.4053,
+      "step": 2128
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822029182468747e-05,
+      "loss": 1.5146,
+      "step": 2129
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5818031215242596e-05,
+      "loss": 1.3857,
+      "step": 2130
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814031841729097e-05,
+      "loss": 1.5732,
+      "step": 2131
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581003106289496e-05,
+      "loss": 1.4443,
+      "step": 2132
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580602887970721e-05,
+      "loss": 1.3223,
+      "step": 2133
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802025293133225e-05,
+      "loss": 1.3325,
+      "step": 2134
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5798020304140724e-05,
+      "loss": 1.3535,
+      "step": 2135
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5794013913697758e-05,
+      "loss": 1.4492,
+      "step": 2136
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790006122772716e-05,
+      "loss": 1.4546,
+      "step": 2137
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5785996932334332e-05,
+      "loss": 1.4219,
+      "step": 2138
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578198634335167e-05,
+      "loss": 1.3096,
+      "step": 2139
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577797435679415e-05,
+      "loss": 1.4297,
+      "step": 2140
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57739609736315e-05,
+      "loss": 1.2632,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5769946194833816e-05,
+      "loss": 1.4082,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576593002137151e-05,
+      "loss": 1.3906,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761912454215345e-05,
+      "loss": 1.498,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757893494336408e-05,
+      "loss": 1.3037,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5753873142706134e-05,
+      "loss": 1.2427,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749851400296284e-05,
+      "loss": 1.4199,
+      "step": 2147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574582826807897e-05,
+      "loss": 1.3018,
+      "step": 2148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5741803747026624e-05,
+      "loss": 1.4453,
+      "step": 2149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5737777838112024e-05,
+      "loss": 1.4229,
+      "step": 2150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573375054230828e-05,
+      "loss": 1.4209,
+      "step": 2151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5729721860588832e-05,
+      "loss": 1.3032,
+      "step": 2152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572569179392747e-05,
+      "loss": 1.5371,
+      "step": 2153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57216603432983e-05,
+      "loss": 1.5312,
+      "step": 2154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717627509675775e-05,
+      "loss": 1.3516,
+      "step": 2155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571359329403468e-05,
+      "loss": 1.2939,
+      "step": 2156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5709557697350132e-05,
+      "loss": 1.4062,
+      "step": 2157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570552072059758e-05,
+      "loss": 1.3574,
+      "step": 2158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5701482364752814e-05,
+      "loss": 1.356,
+      "step": 2159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569744263079195e-05,
+      "loss": 1.4702,
+      "step": 2160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569340151969144e-05,
+      "loss": 1.4404,
+      "step": 2161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.568935903242807e-05,
+      "loss": 1.3936,
+      "step": 2162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685315169978955e-05,
+      "loss": 1.2866,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5681269933321546e-05,
+      "loss": 1.4365,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5677223323433625e-05,
+      "loss": 1.25,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673175341293304e-05,
+      "loss": 1.4028,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669125987879033e-05,
+      "loss": 1.4297,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566507526416958e-05,
+      "loss": 1.3994,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566102317114406e-05,
+      "loss": 1.395,
+      "step": 2169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565696970978192e-05,
+      "loss": 1.5723,
+      "step": 2170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652914881062908e-05,
+      "loss": 1.3403,
+      "step": 2171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5648858685967142e-05,
+      "loss": 1.3711,
+      "step": 2172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564480112547505e-05,
+      "loss": 1.3384,
+      "step": 2173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640742200567386e-05,
+      "loss": 1.4341,
+      "step": 2174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636681912225245e-05,
+      "loss": 1.3862,
+      "step": 2175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563262026143005e-05,
+      "loss": 1.3047,
+      "step": 2176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5628557249163544e-05,
+      "loss": 1.4697,
+      "step": 2177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562449287640781e-05,
+      "loss": 1.2729,
+      "step": 2178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562042714414525e-05,
+      "loss": 1.4297,
+      "step": 2179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5616360053358606e-05,
+      "loss": 1.373,
+      "step": 2180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5612291605030937e-05,
+      "loss": 1.4199,
+      "step": 2181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608221800145643e-05,
+      "loss": 1.3301,
+      "step": 2182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5604150639686436e-05,
+      "loss": 1.4126,
+      "step": 2183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5600078124637364e-05,
+      "loss": 1.332,
+      "step": 2184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5596004255982808e-05,
+      "loss": 1.2131,
+      "step": 2185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 1.4863,
+      "step": 2186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5587852461796375e-05,
+      "loss": 1.2554,
+      "step": 2187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5583774538234883e-05,
+      "loss": 1.4209,
+      "step": 2188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579695265008673e-05,
+      "loss": 1.4224,
+      "step": 2189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5575614643103757e-05,
+      "loss": 1.417,
+      "step": 2190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5571532673506467e-05,
+      "loss": 1.3232,
+      "step": 2191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5567449357203467e-05,
+      "loss": 1.3594,
+      "step": 2192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556336469518174e-05,
+      "loss": 1.4834,
+      "step": 2193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559278688428605e-05,
+      "loss": 1.376,
+      "step": 2194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5555191337931687e-05,
+      "loss": 1.4585,
+      "step": 2195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555110264467895e-05,
+      "loss": 1.2051,
+      "step": 2196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547012609658687e-05,
+      "loss": 1.4717,
+      "step": 2197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5542921233859502e-05,
+      "loss": 1.4121,
+      "step": 2198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5538828518270334e-05,
+      "loss": 1.2544,
+      "step": 2199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553473446388043e-05,
+      "loss": 1.335,
+      "step": 2200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530639071679384e-05,
+      "loss": 1.3623,
+      "step": 2201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55265423426571e-05,
+      "loss": 1.208,
+      "step": 2202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5522444277803797e-05,
+      "loss": 1.2241,
+      "step": 2203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551834487811003e-05,
+      "loss": 1.4131,
+      "step": 2204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5514244144566676e-05,
+      "loss": 1.2754,
+      "step": 2205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510142078164933e-05,
+      "loss": 1.3994,
+      "step": 2206
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5506038679896312e-05,
+      "loss": 1.2573,
+      "step": 2207
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5501933950752655e-05,
+      "loss": 1.3633,
+      "step": 2208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549782789172613e-05,
+      "loss": 1.3594,
+      "step": 2209
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493720503809212e-05,
+      "loss": 1.1785,
+      "step": 2210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5489611787994712e-05,
+      "loss": 1.2686,
+      "step": 2211
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548550174527575e-05,
+      "loss": 1.4102,
+      "step": 2212
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5481390376645777e-05,
+      "loss": 1.2734,
+      "step": 2213
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477277683098555e-05,
+      "loss": 1.3804,
+      "step": 2214
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547316366562818e-05,
+      "loss": 1.4121,
+      "step": 2215
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5469048325229046e-05,
+      "loss": 1.2798,
+      "step": 2216
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546493166289589e-05,
+      "loss": 1.3838,
+      "step": 2217
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460813679623748e-05,
+      "loss": 1.3711,
+      "step": 2218
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5456694376408e-05,
+      "loss": 1.2041,
+      "step": 2219
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452573754244314e-05,
+      "loss": 1.1299,
+      "step": 2220
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544845181412871e-05,
+      "loss": 1.334,
+      "step": 2221
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54443285570575e-05,
+      "loss": 1.4199,
+      "step": 2222
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440203984027323e-05,
+      "loss": 1.3306,
+      "step": 2223
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5436078096035145e-05,
+      "loss": 1.3833,
+      "step": 2224
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5431950894078235e-05,
+      "loss": 1.2234,
+      "step": 2225
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542782237915419e-05,
+      "loss": 1.3516,
+      "step": 2226
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5423692552260922e-05,
+      "loss": 1.3452,
+      "step": 2227
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5419561414396657e-05,
+      "loss": 1.1912,
+      "step": 2228
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541542896655994e-05,
+      "loss": 1.2651,
+      "step": 2229
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5411295209749636e-05,
+      "loss": 1.1616,
+      "step": 2230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540716014496492e-05,
+      "loss": 1.4409,
+      "step": 2231
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5403023773205286e-05,
+      "loss": 1.2485,
+      "step": 2232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5398886095470548e-05,
+      "loss": 1.2969,
+      "step": 2233
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539474711276083e-05,
+      "loss": 1.4033,
+      "step": 2234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539060682607657e-05,
+      "loss": 1.3291,
+      "step": 2235
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538646523641853e-05,
+      "loss": 1.438,
+      "step": 2236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538232234478778e-05,
+      "loss": 1.2959,
+      "step": 2237
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378178152185703e-05,
+      "loss": 1.4805,
+      "step": 2238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5374032659614e-05,
+      "loss": 1.3086,
+      "step": 2239
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536988586807469e-05,
+      "loss": 1.3237,
+      "step": 2240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5365737778570102e-05,
+      "loss": 1.3198,
+      "step": 2241
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5361588392102874e-05,
+      "loss": 1.4844,
+      "step": 2242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357437709675964e-05,
+      "loss": 1.2954,
+      "step": 2243
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535328573229264e-05,
+      "loss": 1.2651,
+      "step": 2244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5349132460956488e-05,
+      "loss": 1.186,
+      "step": 2245
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53449778966714e-05,
+      "loss": 1.4648,
+      "step": 2246
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5340822040441582e-05,
+      "loss": 1.4014,
+      "step": 2247
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533666489327156e-05,
+      "loss": 1.3105,
+      "step": 2248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332506456166164e-05,
+      "loss": 1.2749,
+      "step": 2249
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532834673013053e-05,
+      "loss": 1.3403,
+      "step": 2250
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5324185716170125e-05,
+      "loss": 1.1973,
+      "step": 2251
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320023415290705e-05,
+      "loss": 1.082,
+      "step": 2252
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5315859828498353e-05,
+      "loss": 1.123,
+      "step": 2253
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531169495679946e-05,
+      "loss": 1.3574,
+      "step": 2254
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307528801200718e-05,
+      "loss": 1.3115,
+      "step": 2255
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5303361362709144e-05,
+      "loss": 1.1284,
+      "step": 2256
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529919264233205e-05,
+      "loss": 1.1045,
+      "step": 2257
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5295022641077074e-05,
+      "loss": 1.3418,
+      "step": 2258
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290851359952145e-05,
+      "loss": 1.3755,
+      "step": 2259
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528667879996552e-05,
+      "loss": 1.375,
+      "step": 2260
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282504962125747e-05,
+      "loss": 1.3555,
+      "step": 2261
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52783298474417e-05,
+      "loss": 1.4814,
+      "step": 2262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5274153456922554e-05,
+      "loss": 1.3633,
+      "step": 2263
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5269975791577788e-05,
+      "loss": 1.4014,
+      "step": 2264
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5265796852417195e-05,
+      "loss": 1.2998,
+      "step": 2265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5261616640450875e-05,
+      "loss": 1.4482,
+      "step": 2266
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257435156689235e-05,
+      "loss": 1.2886,
+      "step": 2267
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253252402142989e-05,
+      "loss": 1.4434,
+      "step": 2268
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524906837782316e-05,
+      "loss": 1.394,
+      "step": 2269
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5244883084741071e-05,
+      "loss": 1.3652,
+      "step": 2270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5240696523908366e-05,
+      "loss": 1.3223,
+      "step": 2271
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236508696336976e-05,
+      "loss": 1.3252,
+      "step": 2272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232319603039156e-05,
+      "loss": 1.4121,
+      "step": 2273
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228129245027458e-05,
+      "loss": 1.1755,
+      "step": 2274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223937623314742e-05,
+      "loss": 1.3999,
+      "step": 2275
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5219744738914175e-05,
+      "loss": 1.3291,
+      "step": 2276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5215550592839219e-05,
+      "loss": 1.3408,
+      "step": 2277
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211355186103655e-05,
+      "loss": 1.3848,
+      "step": 2278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5207158519721565e-05,
+      "loss": 1.3452,
+      "step": 2279
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5202960594707325e-05,
+      "loss": 1.27,
+      "step": 2280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198761412075631e-05,
+      "loss": 1.4785,
+      "step": 2281
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194560972841474e-05,
+      "loss": 1.2573,
+      "step": 2282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5190359278020151e-05,
+      "loss": 1.3262,
+      "step": 2283
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5186156328627255e-05,
+      "loss": 1.3975,
+      "step": 2284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5181952125678697e-05,
+      "loss": 1.3618,
+      "step": 2285
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177746670190674e-05,
+      "loss": 1.3701,
+      "step": 2286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173539963179704e-05,
+      "loss": 1.2285,
+      "step": 2287
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169332005662591e-05,
+      "loss": 1.3115,
+      "step": 2288
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5165122798656455e-05,
+      "loss": 1.1714,
+      "step": 2289
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5160912343178704e-05,
+      "loss": 1.4873,
+      "step": 2290
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156700640247062e-05,
+      "loss": 1.2329,
+      "step": 2291
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5152487690879546e-05,
+      "loss": 1.4268,
+      "step": 2292
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5148273496094468e-05,
+      "loss": 1.4053,
+      "step": 2293
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514405805691046e-05,
+      "loss": 1.2466,
+      "step": 2294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513984137434644e-05,
+      "loss": 1.3955,
+      "step": 2295
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135623449421629e-05,
+      "loss": 1.3652,
+      "step": 2296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513140428315555e-05,
+      "loss": 1.2031,
+      "step": 2297
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127183876568024e-05,
+      "loss": 1.2334,
+      "step": 2298
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5122962230679178e-05,
+      "loss": 1.2056,
+      "step": 2299
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118739346509428e-05,
+      "loss": 1.2144,
+      "step": 2300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5114515225079502e-05,
+      "loss": 1.2583,
+      "step": 2301
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5110289867410417e-05,
+      "loss": 1.4092,
+      "step": 2302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106063274523493e-05,
+      "loss": 1.1465,
+      "step": 2303
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101835447440346e-05,
+      "loss": 1.418,
+      "step": 2304
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097606387182894e-05,
+      "loss": 1.3662,
+      "step": 2305
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093376094773347e-05,
+      "loss": 1.4229,
+      "step": 2306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089144571234222e-05,
+      "loss": 1.1587,
+      "step": 2307
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084911817588327e-05,
+      "loss": 1.4023,
+      "step": 2308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508067783485877e-05,
+      "loss": 1.2959,
+      "step": 2309
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5076442624068954e-05,
+      "loss": 1.2773,
+      "step": 2310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507220618624258e-05,
+      "loss": 1.1943,
+      "step": 2311
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5067968522403645e-05,
+      "loss": 1.1914,
+      "step": 2312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506372963357644e-05,
+      "loss": 1.3096,
+      "step": 2313
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.505948952078556e-05,
+      "loss": 1.144,
+      "step": 2314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5055248185055886e-05,
+      "loss": 1.2095,
+      "step": 2315
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051005627412606e-05,
+      "loss": 1.1958,
+      "step": 2316
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5046761848881188e-05,
+      "loss": 1.4238,
+      "step": 2317
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504251685048741e-05,
+      "loss": 1.2935,
+      "step": 2318
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5038270633257336e-05,
+      "loss": 1.3369,
+      "step": 2319
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034023198217328e-05,
+      "loss": 1.2085,
+      "step": 2320
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5029774546394043e-05,
+      "loss": 1.3018,
+      "step": 2321
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025524678814428e-05,
+      "loss": 1.2905,
+      "step": 2322
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021273596505724e-05,
+      "loss": 1.3242,
+      "step": 2323
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017021300495476e-05,
+      "loss": 1.2773,
+      "step": 2324
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012767791811506e-05,
+      "loss": 1.2788,
+      "step": 2325
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008513071481947e-05,
+      "loss": 1.311,
+      "step": 2326
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004257140535208e-05,
+      "loss": 1.2593,
+      "step": 2327
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3447,
+      "step": 2328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995741650905326e-05,
+      "loss": 1.2632,
+      "step": 2329
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499148209428048e-05,
+      "loss": 1.3994,
+      "step": 2330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4987221331155044e-05,
+      "loss": 1.3643,
+      "step": 2331
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982959362558897e-05,
+      "loss": 1.4541,
+      "step": 2332
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4978696189522208e-05,
+      "loss": 1.3423,
+      "step": 2333
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4974431813075439e-05,
+      "loss": 1.2803,
+      "step": 2334
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970166234249335e-05,
+      "loss": 1.2017,
+      "step": 2335
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965899454074941e-05,
+      "loss": 1.4414,
+      "step": 2336
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961631473583589e-05,
+      "loss": 1.3242,
+      "step": 2337
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957362293806896e-05,
+      "loss": 1.2642,
+      "step": 2338
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495309191577678e-05,
+      "loss": 1.2686,
+      "step": 2339
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948820340525439e-05,
+      "loss": 1.4551,
+      "step": 2340
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494454756908536e-05,
+      "loss": 1.3345,
+      "step": 2341
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4940273602489326e-05,
+      "loss": 1.2949,
+      "step": 2342
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935998441770408e-05,
+      "loss": 1.2725,
+      "step": 2343
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931722087961962e-05,
+      "loss": 1.4443,
+      "step": 2344
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492744454209763e-05,
+      "loss": 1.2773,
+      "step": 2345
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.492316580521135e-05,
+      "loss": 1.4365,
+      "step": 2346
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491888587833734e-05,
+      "loss": 1.2949,
+      "step": 2347
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.491460476251011e-05,
+      "loss": 1.2114,
+      "step": 2348
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910322458764461e-05,
+      "loss": 1.2529,
+      "step": 2349
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490603896813547e-05,
+      "loss": 1.3198,
+      "step": 2350
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4901754291658512e-05,
+      "loss": 1.4336,
+      "step": 2351
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897468430369245e-05,
+      "loss": 1.3345,
+      "step": 2352
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489318138530361e-05,
+      "loss": 1.2456,
+      "step": 2353
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888893157497836e-05,
+      "loss": 1.2173,
+      "step": 2354
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4884603747988442e-05,
+      "loss": 1.3057,
+      "step": 2355
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4880313157812227e-05,
+      "loss": 1.2461,
+      "step": 2356
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876021388006276e-05,
+      "loss": 1.2148,
+      "step": 2357
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871728439607967e-05,
+      "loss": 1.1953,
+      "step": 2358
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867434313654951e-05,
+      "loss": 1.3105,
+      "step": 2359
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486313901118517e-05,
+      "loss": 1.3369,
+      "step": 2360
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858842533236853e-05,
+      "loss": 1.2212,
+      "step": 2361
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854544880848507e-05,
+      "loss": 1.269,
+      "step": 2362
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4850246055058923e-05,
+      "loss": 1.335,
+      "step": 2363
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845946056907187e-05,
+      "loss": 1.2217,
+      "step": 2364
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841644887432652e-05,
+      "loss": 1.3887,
+      "step": 2365
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4837342547674962e-05,
+      "loss": 1.3584,
+      "step": 2366
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833039038674049e-05,
+      "loss": 1.3555,
+      "step": 2367
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4828734361470116e-05,
+      "loss": 1.3125,
+      "step": 2368
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482442851710366e-05,
+      "loss": 1.3125,
+      "step": 2369
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820121506615452e-05,
+      "loss": 1.4668,
+      "step": 2370
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481581333104655e-05,
+      "loss": 1.2861,
+      "step": 2371
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4811503991438287e-05,
+      "loss": 1.1716,
+      "step": 2372
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4807193488832284e-05,
+      "loss": 1.1553,
+      "step": 2373
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4802881824270443e-05,
+      "loss": 1.2529,
+      "step": 2374
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479856899879494e-05,
+      "loss": 1.4023,
+      "step": 2375
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4794255013448242e-05,
+      "loss": 1.3037,
+      "step": 2376
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4789939869273082e-05,
+      "loss": 1.1235,
+      "step": 2377
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785623567312492e-05,
+      "loss": 1.3867,
+      "step": 2378
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781306108609768e-05,
+      "loss": 1.2014,
+      "step": 2379
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477698749420849e-05,
+      "loss": 1.165,
+      "step": 2380
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477266772515252e-05,
+      "loss": 1.3145,
+      "step": 2381
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768346802485999e-05,
+      "loss": 1.4727,
+      "step": 2382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764024727253343e-05,
+      "loss": 1.2964,
+      "step": 2383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4759701500499249e-05,
+      "loss": 1.3296,
+      "step": 2384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4755377123268693e-05,
+      "loss": 1.3188,
+      "step": 2385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4751051596606929e-05,
+      "loss": 1.3315,
+      "step": 2386
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4746724921559491e-05,
+      "loss": 1.2109,
+      "step": 2387
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742397099172183e-05,
+      "loss": 1.375,
+      "step": 2388
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738068130491092e-05,
+      "loss": 1.29,
+      "step": 2389
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733738016562584e-05,
+      "loss": 1.1831,
+      "step": 2390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729406758433296e-05,
+      "loss": 1.2271,
+      "step": 2391
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4725074357150146e-05,
+      "loss": 1.3955,
+      "step": 2392
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4720740813760326e-05,
+      "loss": 1.4297,
+      "step": 2393
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716406129311307e-05,
+      "loss": 1.1062,
+      "step": 2394
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471207030485083e-05,
+      "loss": 1.3721,
+      "step": 2395
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707733341426921e-05,
+      "loss": 1.3247,
+      "step": 2396
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470339524008787e-05,
+      "loss": 1.5469,
+      "step": 2397
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4699056001882252e-05,
+      "loss": 1.5156,
+      "step": 2398
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469471562785891e-05,
+      "loss": 1.3613,
+      "step": 2399
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690374119066964e-05,
+      "loss": 1.4482,
+      "step": 2400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4686031476555811e-05,
+      "loss": 1.3691,
+      "step": 2401
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681687701375113e-05,
+      "loss": 1.2729,
+      "step": 2402
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677342794574819e-05,
+      "loss": 1.2412,
+      "step": 2403
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4672996757205143e-05,
+      "loss": 1.3481,
+      "step": 2404
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668649590316573e-05,
+      "loss": 1.4072,
+      "step": 2405
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664301294959868e-05,
+      "loss": 1.2734,
+      "step": 2406
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4659951872186066e-05,
+      "loss": 1.3594,
+      "step": 2407
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655601323046477e-05,
+      "loss": 1.4102,
+      "step": 2408
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4651249648592676e-05,
+      "loss": 1.1772,
+      "step": 2409
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4646896849876518e-05,
+      "loss": 1.2329,
+      "step": 2410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464254292795012e-05,
+      "loss": 1.334,
+      "step": 2411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638187883865881e-05,
+      "loss": 1.3164,
+      "step": 2412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4633831718676468e-05,
+      "loss": 1.0713,
+      "step": 2413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4629474433434818e-05,
+      "loss": 1.3789,
+      "step": 2414
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625116029194134e-05,
+      "loss": 1.4102,
+      "step": 2415
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4620756507007898e-05,
+      "loss": 1.0342,
+      "step": 2416
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616395867929862e-05,
+      "loss": 1.415,
+      "step": 2417
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4612034113014036e-05,
+      "loss": 1.2617,
+      "step": 2418
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4607671243314717e-05,
+      "loss": 1.394,
+      "step": 2419
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4603307259886453e-05,
+      "loss": 1.2842,
+      "step": 2420
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4598942163784073e-05,
+      "loss": 1.0845,
+      "step": 2421
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4594575956062678e-05,
+      "loss": 1.3789,
+      "step": 2422
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590208637777632e-05,
+      "loss": 1.4072,
+      "step": 2423
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4585840209984561e-05,
+      "loss": 1.2598,
+      "step": 2424
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581470673739372e-05,
+      "loss": 1.2832,
+      "step": 2425
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4577100030098233e-05,
+      "loss": 1.3389,
+      "step": 2426
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572728280117577e-05,
+      "loss": 1.4004,
+      "step": 2427
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4568355424854113e-05,
+      "loss": 1.3594,
+      "step": 2428
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563981465364807e-05,
+      "loss": 1.3296,
+      "step": 2429
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45596064027069e-05,
+      "loss": 1.291,
+      "step": 2430
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555230237937897e-05,
+      "loss": 1.3262,
+      "step": 2431
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550852972115568e-05,
+      "loss": 1.4004,
+      "step": 2432
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546474606297951e-05,
+      "loss": 1.3228,
+      "step": 2433
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4542095141543346e-05,
+      "loss": 1.3232,
+      "step": 2434
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4537714578910329e-05,
+      "loss": 1.3838,
+      "step": 2435
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533332919457728e-05,
+      "loss": 1.3408,
+      "step": 2436
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528950164244644e-05,
+      "loss": 1.2793,
+      "step": 2437
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524566314330439e-05,
+      "loss": 1.0955,
+      "step": 2438
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520181370774742e-05,
+      "loss": 1.2725,
+      "step": 2439
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4515795334637451e-05,
+      "loss": 1.3452,
+      "step": 2440
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511408206978714e-05,
+      "loss": 1.2656,
+      "step": 2441
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507019988858964e-05,
+      "loss": 1.4229,
+      "step": 2442
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4502630681338876e-05,
+      "loss": 1.1963,
+      "step": 2443
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4498240285479401e-05,
+      "loss": 1.4326,
+      "step": 2444
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449384880234175e-05,
+      "loss": 1.3047,
+      "step": 2445
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4489456232987395e-05,
+      "loss": 1.3457,
+      "step": 2446
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4485062578478076e-05,
+      "loss": 1.3271,
+      "step": 2447
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4480667839875786e-05,
+      "loss": 1.2539,
+      "step": 2448
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447627201824279e-05,
+      "loss": 1.4141,
+      "step": 2449
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447187511464161e-05,
+      "loss": 1.4092,
+      "step": 2450
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467477130135029e-05,
+      "loss": 1.2822,
+      "step": 2451
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4463078065786091e-05,
+      "loss": 1.1196,
+      "step": 2452
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458677922658107e-05,
+      "loss": 1.2104,
+      "step": 2453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454276701814636e-05,
+      "loss": 1.2305,
+      "step": 2454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4449874404319508e-05,
+      "loss": 1.2556,
+      "step": 2455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4445471031236818e-05,
+      "loss": 1.3604,
+      "step": 2456
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441066583630905e-05,
+      "loss": 1.3887,
+      "step": 2457
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443666106256638e-05,
+      "loss": 1.2539,
+      "step": 2458
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432254469108107e-05,
+      "loss": 1.3066,
+      "step": 2459
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442784680432122e-05,
+      "loss": 1.4307,
+      "step": 2460
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4423438069271092e-05,
+      "loss": 1.3232,
+      "step": 2461
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4419028265023375e-05,
+      "loss": 1.3071,
+      "step": 2462
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441461739264397e-05,
+      "loss": 1.3379,
+      "step": 2463
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4410205453199036e-05,
+      "loss": 1.2891,
+      "step": 2464
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405792447754992e-05,
+      "loss": 1.1182,
+      "step": 2465
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401378377378512e-05,
+      "loss": 1.2817,
+      "step": 2466
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4396963243136535e-05,
+      "loss": 1.3955,
+      "step": 2467
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4392547046096241e-05,
+      "loss": 1.1011,
+      "step": 2468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438812978732509e-05,
+      "loss": 1.2861,
+      "step": 2469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383711467890776e-05,
+      "loss": 1.3857,
+      "step": 2470
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4379292088861265e-05,
+      "loss": 1.2812,
+      "step": 2471
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437487165130477e-05,
+      "loss": 1.2612,
+      "step": 2472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4370450156289763e-05,
+      "loss": 1.2354,
+      "step": 2473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366027604884976e-05,
+      "loss": 1.2559,
+      "step": 2474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361603998159388e-05,
+      "loss": 1.3359,
+      "step": 2475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4357179337182244e-05,
+      "loss": 1.4404,
+      "step": 2476
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4352753623023028e-05,
+      "loss": 1.2842,
+      "step": 2477
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4348326856751496e-05,
+      "loss": 1.3979,
+      "step": 2478
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343899039437643e-05,
+      "loss": 1.3789,
+      "step": 2479
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.433947017215173e-05,
+      "loss": 1.2607,
+      "step": 2480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335040255964265e-05,
+      "loss": 1.1592,
+      "step": 2481
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330609291946012e-05,
+      "loss": 1.2817,
+      "step": 2482
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4326177281167986e-05,
+      "loss": 1.3662,
+      "step": 2483
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432174422470146e-05,
+      "loss": 1.4795,
+      "step": 2484
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4317310123617953e-05,
+      "loss": 1.3052,
+      "step": 2485
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312874978989244e-05,
+      "loss": 1.4648,
+      "step": 2486
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4308438791887358e-05,
+      "loss": 1.1699,
+      "step": 2487
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4304001563384573e-05,
+      "loss": 1.251,
+      "step": 2488
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429956329455342e-05,
+      "loss": 1.2288,
+      "step": 2489
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295123986466684e-05,
+      "loss": 1.2568,
+      "step": 2490
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4290683640197394e-05,
+      "loss": 1.1904,
+      "step": 2491
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428624225681884e-05,
+      "loss": 1.3047,
+      "step": 2492
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4281799837404553e-05,
+      "loss": 1.3555,
+      "step": 2493
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4277356383028322e-05,
+      "loss": 1.1519,
+      "step": 2494
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4272911894764179e-05,
+      "loss": 1.417,
+      "step": 2495
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268466373686414e-05,
+      "loss": 1.4355,
+      "step": 2496
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264019820869556e-05,
+      "loss": 1.4067,
+      "step": 2497
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.42595722373884e-05,
+      "loss": 1.4316,
+      "step": 2498
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4255123624317968e-05,
+      "loss": 1.2183,
+      "step": 2499
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.425067398273355e-05,
+      "loss": 1.2607,
+      "step": 2500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246223313710674e-05,
+      "loss": 1.355,
+      "step": 2501
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4241771618325123e-05,
+      "loss": 1.2451,
+      "step": 2502
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4237318897652921e-05,
+      "loss": 1.2085,
+      "step": 2503
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4232865152770345e-05,
+      "loss": 1.3096,
+      "step": 2504
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4228410384753923e-05,
+      "loss": 1.397,
+      "step": 2505
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223954594680416e-05,
+      "loss": 1.2666,
+      "step": 2506
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421949778362685e-05,
+      "loss": 1.1851,
+      "step": 2507
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215039952670482e-05,
+      "loss": 1.3555,
+      "step": 2508
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210581102888827e-05,
+      "loss": 1.4673,
+      "step": 2509
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206121235359643e-05,
+      "loss": 1.2661,
+      "step": 2510
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420166035116093e-05,
+      "loss": 1.415,
+      "step": 2511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419719845137094e-05,
+      "loss": 1.4092,
+      "step": 2512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419273553706816e-05,
+      "loss": 1.1504,
+      "step": 2513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4188271609331341e-05,
+      "loss": 1.1221,
+      "step": 2514
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4183806669239459e-05,
+      "loss": 1.4019,
+      "step": 2515
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179340717871749e-05,
+      "loss": 1.543,
+      "step": 2516
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174873756307675e-05,
+      "loss": 1.248,
+      "step": 2517
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417040578562696e-05,
+      "loss": 1.4541,
+      "step": 2518
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416593680690957e-05,
+      "loss": 1.3188,
+      "step": 2519
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4161466821235705e-05,
+      "loss": 1.2725,
+      "step": 2520
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156995829685817e-05,
+      "loss": 1.2817,
+      "step": 2521
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152523833340595e-05,
+      "loss": 1.3018,
+      "step": 2522
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4148050833280977e-05,
+      "loss": 1.2417,
+      "step": 2523
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143576830588137e-05,
+      "loss": 1.4297,
+      "step": 2524
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139101826343496e-05,
+      "loss": 1.3018,
+      "step": 2525
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4134625821628713e-05,
+      "loss": 1.1567,
+      "step": 2526
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4130148817525696e-05,
+      "loss": 1.3691,
+      "step": 2527
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412567081511659e-05,
+      "loss": 1.2061,
+      "step": 2528
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121191815483777e-05,
+      "loss": 1.3042,
+      "step": 2529
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.411671181970989e-05,
+      "loss": 1.3018,
+      "step": 2530
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112230828877791e-05,
+      "loss": 1.1748,
+      "step": 2531
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4107748844070597e-05,
+      "loss": 1.3442,
+      "step": 2532
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4103265866371646e-05,
+      "loss": 1.0898,
+      "step": 2533
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4098781896864537e-05,
+      "loss": 1.2705,
+      "step": 2534
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094296936633096e-05,
+      "loss": 1.2373,
+      "step": 2535
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089810986761387e-05,
+      "loss": 1.2476,
+      "step": 2536
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4085324048333726e-05,
+      "loss": 1.3965,
+      "step": 2537
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408083612243465e-05,
+      "loss": 1.4355,
+      "step": 2538
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407634721014895e-05,
+      "loss": 1.3745,
+      "step": 2539
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4071857312561648e-05,
+      "loss": 1.3726,
+      "step": 2540
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 1.3403,
+      "step": 2541
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062874565823519e-05,
+      "loss": 1.3354,
+      "step": 2542
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405838171884393e-05,
+      "loss": 1.2598,
+      "step": 2543
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.405388789090521e-05,
+      "loss": 1.2812,
+      "step": 2544
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049393083093574e-05,
+      "loss": 1.3896,
+      "step": 2545
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044897296495472e-05,
+      "loss": 1.4141,
+      "step": 2546
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040400532197582e-05,
+      "loss": 1.269,
+      "step": 2547
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035902791286834e-05,
+      "loss": 1.231,
+      "step": 2548
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031404074850374e-05,
+      "loss": 1.2998,
+      "step": 2549
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026904383975604e-05,
+      "loss": 1.2695,
+      "step": 2550
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022403719750153e-05,
+      "loss": 1.2905,
+      "step": 2551
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401790208326188e-05,
+      "loss": 1.3477,
+      "step": 2552
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401339947559889e-05,
+      "loss": 1.2754,
+      "step": 2553
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008895897849512e-05,
+      "loss": 1.3179,
+      "step": 2554
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004391351102317e-05,
+      "loss": 1.2793,
+      "step": 2555
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999885836446106e-05,
+      "loss": 1.2109,
+      "step": 2556
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995379354969918e-05,
+      "loss": 1.3457,
+      "step": 2557
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399087190776302e-05,
+      "loss": 1.4092,
+      "step": 2558
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3986363495914917e-05,
+      "loss": 1.2695,
+      "step": 2559
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981854120515346e-05,
+      "loss": 1.3105,
+      "step": 2560
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3977343782654274e-05,
+      "loss": 1.2573,
+      "step": 2561
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972832483421907e-05,
+      "loss": 1.1606,
+      "step": 2562
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968320223908678e-05,
+      "loss": 1.1216,
+      "step": 2563
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963807005205254e-05,
+      "loss": 1.3882,
+      "step": 2564
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3959292828402534e-05,
+      "loss": 1.4473,
+      "step": 2565
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395477769459165e-05,
+      "loss": 1.2769,
+      "step": 2566
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395026160486396e-05,
+      "loss": 1.2583,
+      "step": 2567
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945744560311056e-05,
+      "loss": 1.1685,
+      "step": 2568
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941226562024767e-05,
+      "loss": 1.2754,
+      "step": 2569
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393670761109714e-05,
+      "loss": 1.2642,
+      "step": 2570
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3932187708620464e-05,
+      "loss": 1.2915,
+      "step": 2571
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927666855687251e-05,
+      "loss": 1.3115,
+      "step": 2572
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3923145053390247e-05,
+      "loss": 1.377,
+      "step": 2573
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918622302822425e-05,
+      "loss": 1.2563,
+      "step": 2574
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3914098605076987e-05,
+      "loss": 1.2271,
+      "step": 2575
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390957396124736e-05,
+      "loss": 1.2324,
+      "step": 2576
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3905048372427212e-05,
+      "loss": 1.3618,
+      "step": 2577
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900521839710428e-05,
+      "loss": 1.4023,
+      "step": 2578
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895994364191124e-05,
+      "loss": 1.2939,
+      "step": 2579
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3891465946963645e-05,
+      "loss": 1.3496,
+      "step": 2580
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3886936589122564e-05,
+      "loss": 1.2729,
+      "step": 2581
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3882406291762681e-05,
+      "loss": 1.2891,
+      "step": 2582
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877875055979021e-05,
+      "loss": 1.2383,
+      "step": 2583
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387334288286684e-05,
+      "loss": 1.3857,
+      "step": 2584
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386880977352162e-05,
+      "loss": 1.3887,
+      "step": 2585
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386427572903906e-05,
+      "loss": 1.1836,
+      "step": 2586
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.38597407505151e-05,
+      "loss": 1.293,
+      "step": 2587
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3855204839045894e-05,
+      "loss": 1.3467,
+      "step": 2588
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850667995727827e-05,
+      "loss": 1.3564,
+      "step": 2589
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846130221657506e-05,
+      "loss": 1.5156,
+      "step": 2590
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3841591517931769e-05,
+      "loss": 1.2539,
+      "step": 2591
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3837051885647673e-05,
+      "loss": 1.2773,
+      "step": 2592
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832511325902498e-05,
+      "loss": 1.4199,
+      "step": 2593
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827969839793757e-05,
+      "loss": 1.2939,
+      "step": 2594
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3823427428419178e-05,
+      "loss": 1.2803,
+      "step": 2595
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3818884092876714e-05,
+      "loss": 1.4668,
+      "step": 2596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3814339834264545e-05,
+      "loss": 1.4766,
+      "step": 2597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809794653681075e-05,
+      "loss": 1.3408,
+      "step": 2598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805248552224923e-05,
+      "loss": 1.1777,
+      "step": 2599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3800701530994942e-05,
+      "loss": 1.3403,
+      "step": 2600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796153591090195e-05,
+      "loss": 1.2412,
+      "step": 2601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3791604733609978e-05,
+      "loss": 1.3008,
+      "step": 2602
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.37870549596538e-05,
+      "loss": 1.1616,
+      "step": 2603
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782504270321397e-05,
+      "loss": 1.4316,
+      "step": 2604
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777952666712726e-05,
+      "loss": 1.335,
+      "step": 2605
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773400149927962e-05,
+      "loss": 1.1714,
+      "step": 2606
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768846721067507e-05,
+      "loss": 1.1909,
+      "step": 2607
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376429238123197e-05,
+      "loss": 1.1626,
+      "step": 2608
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759737131522197e-05,
+      "loss": 1.3003,
+      "step": 2609
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3755180973039241e-05,
+      "loss": 1.1904,
+      "step": 2610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750623906884383e-05,
+      "loss": 1.4424,
+      "step": 2611
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3746065934159123e-05,
+      "loss": 1.0776,
+      "step": 2612
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3741507055965169e-05,
+      "loss": 1.2642,
+      "step": 2613
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3736947273404462e-05,
+      "loss": 1.374,
+      "step": 2614
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3732386587579152e-05,
+      "loss": 1.3701,
+      "step": 2615
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727824999591614e-05,
+      "loss": 1.2515,
+      "step": 2616
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723262510544437e-05,
+      "loss": 1.2466,
+      "step": 2617
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371869912154043e-05,
+      "loss": 1.29,
+      "step": 2618
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714134833682618e-05,
+      "loss": 1.1865,
+      "step": 2619
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370956964807424e-05,
+      "loss": 1.3076,
+      "step": 2620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3705003565818763e-05,
+      "loss": 1.4014,
+      "step": 2621
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3700436588019854e-05,
+      "loss": 1.2432,
+      "step": 2622
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3695868715781416e-05,
+      "loss": 1.2427,
+      "step": 2623
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3691299950207548e-05,
+      "loss": 1.377,
+      "step": 2624
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3686730292402583e-05,
+      "loss": 1.4551,
+      "step": 2625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682159743471059e-05,
+      "loss": 1.4434,
+      "step": 2626
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3677588304517731e-05,
+      "loss": 1.3076,
+      "step": 2627
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367301597664757e-05,
+      "loss": 1.3057,
+      "step": 2628
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3668442760965763e-05,
+      "loss": 1.374,
+      "step": 2629
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663868658577712e-05,
+      "loss": 1.3906,
+      "step": 2630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365929367058903e-05,
+      "loss": 1.3291,
+      "step": 2631
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3654717798105544e-05,
+      "loss": 1.3989,
+      "step": 2632
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3650141042233299e-05,
+      "loss": 1.3843,
+      "step": 2633
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3645563404078554e-05,
+      "loss": 1.4736,
+      "step": 2634
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640984884747774e-05,
+      "loss": 1.3223,
+      "step": 2635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3636405485347645e-05,
+      "loss": 1.3662,
+      "step": 2636
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631825206985063e-05,
+      "loss": 1.4902,
+      "step": 2637
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362724405076713e-05,
+      "loss": 1.2681,
+      "step": 2638
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622662017801174e-05,
+      "loss": 1.1914,
+      "step": 2639
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618079109194717e-05,
+      "loss": 1.3306,
+      "step": 2640
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361349532605551e-05,
+      "loss": 1.2058,
+      "step": 2641
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608910669491508e-05,
+      "loss": 1.3301,
+      "step": 2642
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604325140610872e-05,
+      "loss": 1.2549,
+      "step": 2643
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3599738740521987e-05,
+      "loss": 1.335,
+      "step": 2644
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359515147033343e-05,
+      "loss": 1.3047,
+      "step": 2645
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590563331154008e-05,
+      "loss": 1.2778,
+      "step": 2646
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3585974324092721e-05,
+      "loss": 1.1436,
+      "step": 2647
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581384450258796e-05,
+      "loss": 1.3193,
+      "step": 2648
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3576793710761649e-05,
+      "loss": 1.2466,
+      "step": 2649
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572202106710924e-05,
+      "loss": 1.3916,
+      "step": 2650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567609639216463e-05,
+      "loss": 1.3184,
+      "step": 2651
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563016309388322e-05,
+      "loss": 1.3799,
+      "step": 2652
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3558422118336765e-05,
+      "loss": 1.2444,
+      "step": 2653
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553827067172257e-05,
+      "loss": 1.2441,
+      "step": 2654
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549231157005482e-05,
+      "loss": 1.2163,
+      "step": 2655
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544634388947321e-05,
+      "loss": 1.2676,
+      "step": 2656
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3540036764108875e-05,
+      "loss": 1.2646,
+      "step": 2657
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535438283601437e-05,
+      "loss": 1.2939,
+      "step": 2658
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3530838948536517e-05,
+      "loss": 1.4072,
+      "step": 2659
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352623876002583e-05,
+      "loss": 1.4561,
+      "step": 2660
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521637719181295e-05,
+      "loss": 1.3301,
+      "step": 2661
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3517035827115042e-05,
+      "loss": 1.3335,
+      "step": 2662
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512433084939398e-05,
+      "loss": 1.3848,
+      "step": 2663
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3507829493766903e-05,
+      "loss": 1.2339,
+      "step": 2664
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35032250547103e-05,
+      "loss": 1.2646,
+      "step": 2665
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498619768882537e-05,
+      "loss": 1.2329,
+      "step": 2666
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349401363739676e-05,
+      "loss": 1.3257,
+      "step": 2667
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3489406661366337e-05,
+      "loss": 1.3652,
+      "step": 2668
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348479884190482e-05,
+      "loss": 1.3564,
+      "step": 2669
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480190180125979e-05,
+      "loss": 1.3896,
+      "step": 2670
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347558067714378e-05,
+      "loss": 1.3113,
+      "step": 2671
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3470970334072393e-05,
+      "loss": 1.3628,
+      "step": 2672
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3466359152026197e-05,
+      "loss": 1.2192,
+      "step": 2673
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3461747132119764e-05,
+      "loss": 1.3975,
+      "step": 2674
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3457134275467878e-05,
+      "loss": 1.293,
+      "step": 2675
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3452520583185522e-05,
+      "loss": 1.314,
+      "step": 2676
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447906056387876e-05,
+      "loss": 1.3691,
+      "step": 2677
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443290696190335e-05,
+      "loss": 1.1714,
+      "step": 2678
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3438674503708474e-05,
+      "loss": 1.3711,
+      "step": 2679
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3434057480058091e-05,
+      "loss": 1.3828,
+      "step": 2680
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3429439626355171e-05,
+      "loss": 1.4053,
+      "step": 2681
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342482094371591e-05,
+      "loss": 1.2217,
+      "step": 2682
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 1.209,
+      "step": 2683
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3415581096094106e-05,
+      "loss": 1.3271,
+      "step": 2684
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410959933344948e-05,
+      "loss": 1.2051,
+      "step": 2685
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3406337946126207e-05,
+      "loss": 1.2329,
+      "step": 2686
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401715135555072e-05,
+      "loss": 1.3359,
+      "step": 2687
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339709150274893e-05,
+      "loss": 1.377,
+      "step": 2688
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339246704882537e-05,
+      "loss": 1.2642,
+      "step": 2689
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3387841774902172e-05,
+      "loss": 1.1978,
+      "step": 2690
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338321568209733e-05,
+      "loss": 1.4385,
+      "step": 2691
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3378588771529018e-05,
+      "loss": 1.2656,
+      "step": 2692
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3373961044315615e-05,
+      "loss": 1.4141,
+      "step": 2693
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33693325015757e-05,
+      "loss": 1.292,
+      "step": 2694
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364703144428046e-05,
+      "loss": 1.311,
+      "step": 2695
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3360072973991627e-05,
+      "loss": 1.2256,
+      "step": 2696
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3355441991385606e-05,
+      "loss": 1.2251,
+      "step": 2697
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350810197729352e-05,
+      "loss": 1.4287,
+      "step": 2698
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346177594142416e-05,
+      "loss": 1.1934,
+      "step": 2699
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3341544181744559e-05,
+      "loss": 1.2803,
+      "step": 2700
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.333690996165573e-05,
+      "loss": 1.3174,
+      "step": 2701
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332274934996077e-05,
+      "loss": 1.1489,
+      "step": 2702
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327639102885939e-05,
+      "loss": 1.3125,
+      "step": 2703
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3323002466445853e-05,
+      "loss": 1.3101,
+      "step": 2704
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318365026796553e-05,
+      "loss": 1.3975,
+      "step": 2705
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3313726785058955e-05,
+      "loss": 1.1226,
+      "step": 2706
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3309087742354185e-05,
+      "loss": 1.3691,
+      "step": 2707
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330444789980355e-05,
+      "loss": 1.4341,
+      "step": 2708
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3299807258528557e-05,
+      "loss": 1.4019,
+      "step": 2709
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3295165819650903e-05,
+      "loss": 1.1987,
+      "step": 2710
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290523584292479e-05,
+      "loss": 1.4756,
+      "step": 2711
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3285880553575367e-05,
+      "loss": 1.3438,
+      "step": 2712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3281236728621849e-05,
+      "loss": 1.1909,
+      "step": 2713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276592110554389e-05,
+      "loss": 1.3042,
+      "step": 2714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3271946700495644e-05,
+      "loss": 1.3105,
+      "step": 2715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267300499568468e-05,
+      "loss": 1.1826,
+      "step": 2716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262653508895901e-05,
+      "loss": 1.3818,
+      "step": 2717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258005729601178e-05,
+      "loss": 1.312,
+      "step": 2718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325335716280772e-05,
+      "loss": 1.2183,
+      "step": 2719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248707809639143e-05,
+      "loss": 1.4551,
+      "step": 2720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244057671219248e-05,
+      "loss": 1.3159,
+      "step": 2721
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3239406748672034e-05,
+      "loss": 1.292,
+      "step": 2722
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3234755043121681e-05,
+      "loss": 1.3975,
+      "step": 2723
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230102555692561e-05,
+      "loss": 1.3711,
+      "step": 2724
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3225449287509238e-05,
+      "loss": 1.2627,
+      "step": 2725
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3220795239696458e-05,
+      "loss": 1.3882,
+      "step": 2726
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3216140413379166e-05,
+      "loss": 1.1545,
+      "step": 2727
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211484809682482e-05,
+      "loss": 1.2651,
+      "step": 2728
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3206828429731724e-05,
+      "loss": 1.3652,
+      "step": 2729
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202171274652396e-05,
+      "loss": 1.3301,
+      "step": 2730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197513345570183e-05,
+      "loss": 1.3125,
+      "step": 2731
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3192854643610971e-05,
+      "loss": 1.2959,
+      "step": 2732
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3188195169900813e-05,
+      "loss": 1.1685,
+      "step": 2733
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183534925565966e-05,
+      "loss": 1.3574,
+      "step": 2734
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3178873911732866e-05,
+      "loss": 1.3926,
+      "step": 2735
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317421212952813e-05,
+      "loss": 1.3438,
+      "step": 2736
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169549580078573e-05,
+      "loss": 1.3311,
+      "step": 2737
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3164886264511183e-05,
+      "loss": 1.3701,
+      "step": 2738
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3160222183953144e-05,
+      "loss": 1.4111,
+      "step": 2739
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155557339531817e-05,
+      "loss": 1.2808,
+      "step": 2740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3150891732374751e-05,
+      "loss": 1.2866,
+      "step": 2741
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146225363609677e-05,
+      "loss": 1.3359,
+      "step": 2742
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141558234364516e-05,
+      "loss": 1.1904,
+      "step": 2743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136890345767364e-05,
+      "loss": 1.2456,
+      "step": 2744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3132221698946508e-05,
+      "loss": 1.3726,
+      "step": 2745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3127552295030412e-05,
+      "loss": 1.2676,
+      "step": 2746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3122882135147732e-05,
+      "loss": 1.3457,
+      "step": 2747
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.31182112204273e-05,
+      "loss": 1.3354,
+      "step": 2748
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113539551998125e-05,
+      "loss": 1.3916,
+      "step": 2749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108867130989414e-05,
+      "loss": 1.4385,
+      "step": 2750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.310419395853054e-05,
+      "loss": 1.4385,
+      "step": 2751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099520035751067e-05,
+      "loss": 1.4619,
+      "step": 2752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3094845363780739e-05,
+      "loss": 1.4497,
+      "step": 2753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 1.3291,
+      "step": 2754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085493776787386e-05,
+      "loss": 1.2871,
+      "step": 2755
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3080816864024752e-05,
+      "loss": 1.1938,
+      "step": 2756
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3076139206592044e-05,
+      "loss": 1.4487,
+      "step": 2757
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30714608056199e-05,
+      "loss": 1.3525,
+      "step": 2758
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3066781662239154e-05,
+      "loss": 1.2866,
+      "step": 2759
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3062101777580802e-05,
+      "loss": 1.5264,
+      "step": 2760
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057421152776031e-05,
+      "loss": 1.2837,
+      "step": 2761
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3052739788956207e-05,
+      "loss": 1.3799,
+      "step": 2762
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3048057687252866e-05,
+      "loss": 1.3311,
+      "step": 2763
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043374848797732e-05,
+      "loss": 1.3633,
+      "step": 2764
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3038691274722702e-05,
+      "loss": 1.0032,
+      "step": 2765
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303400696615985e-05,
+      "loss": 1.2285,
+      "step": 2766
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3029321924241427e-05,
+      "loss": 1.335,
+      "step": 2767
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302463615009987e-05,
+      "loss": 1.2126,
+      "step": 2768
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301994964486778e-05,
+      "loss": 1.3433,
+      "step": 2769
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3015262409677942e-05,
+      "loss": 1.2007,
+      "step": 2770
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.301057444566332e-05,
+      "loss": 1.2964,
+      "step": 2771
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005885753957048e-05,
+      "loss": 1.4082,
+      "step": 2772
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300119633569244e-05,
+      "loss": 1.2651,
+      "step": 2773
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299650619200298e-05,
+      "loss": 1.3276,
+      "step": 2774
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.299181532402234e-05,
+      "loss": 1.291,
+      "step": 2775
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2987123732884349e-05,
+      "loss": 1.4277,
+      "step": 2776
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982431419723027e-05,
+      "loss": 1.2798,
+      "step": 2777
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977738385672558e-05,
+      "loss": 1.1826,
+      "step": 2778
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973044631867305e-05,
+      "loss": 1.5049,
+      "step": 2779
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2968350159441807e-05,
+      "loss": 1.2651,
+      "step": 2780
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963654969530773e-05,
+      "loss": 1.46,
+      "step": 2781
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295895906326908e-05,
+      "loss": 1.3896,
+      "step": 2782
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2954262441791792e-05,
+      "loss": 1.3721,
+      "step": 2783
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2949565106234138e-05,
+      "loss": 1.4092,
+      "step": 2784
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944867057731514e-05,
+      "loss": 1.2236,
+      "step": 2785
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29401682974195e-05,
+      "loss": 1.4434,
+      "step": 2786
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2935468826433839e-05,
+      "loss": 1.2539,
+      "step": 2787
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293076864591045e-05,
+      "loss": 1.2969,
+      "step": 2788
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926067756985423e-05,
+      "loss": 1.3013,
+      "step": 2789
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2921366160795017e-05,
+      "loss": 1.4482,
+      "step": 2790
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916663858475666e-05,
+      "loss": 1.1582,
+      "step": 2791
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291196085116397e-05,
+      "loss": 1.3643,
+      "step": 2792
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2907257139996704e-05,
+      "loss": 1.2285,
+      "step": 2793
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2902552726110811e-05,
+      "loss": 1.3125,
+      "step": 2794
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897847610643401e-05,
+      "loss": 1.2568,
+      "step": 2795
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289314179473176e-05,
+      "loss": 1.2234,
+      "step": 2796
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2888435279513338e-05,
+      "loss": 1.3291,
+      "step": 2797
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2883728066125758e-05,
+      "loss": 1.4834,
+      "step": 2798
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2879020155706803e-05,
+      "loss": 1.1978,
+      "step": 2799
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874311549394434e-05,
+      "loss": 1.4033,
+      "step": 2800
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.286960224832678e-05,
+      "loss": 1.3198,
+      "step": 2801
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864892253642137e-05,
+      "loss": 1.4824,
+      "step": 2802
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860181566478958e-05,
+      "loss": 1.2661,
+      "step": 2803
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2855470187975878e-05,
+      "loss": 1.3154,
+      "step": 2804
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850758119271696e-05,
+      "loss": 1.3369,
+      "step": 2805
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846045361505366e-05,
+      "loss": 1.4414,
+      "step": 2806
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2841331915816029e-05,
+      "loss": 1.3799,
+      "step": 2807
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836617783342968e-05,
+      "loss": 1.3633,
+      "step": 2808
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2831902965225655e-05,
+      "loss": 1.3652,
+      "step": 2809
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2827187462603715e-05,
+      "loss": 1.3643,
+      "step": 2810
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282247127661694e-05,
+      "loss": 1.2202,
+      "step": 2811
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2817754408405288e-05,
+      "loss": 1.3521,
+      "step": 2812
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813036859108883e-05,
+      "loss": 1.417,
+      "step": 2813
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808318629868014e-05,
+      "loss": 1.2358,
+      "step": 2814
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280359972182313e-05,
+      "loss": 1.374,
+      "step": 2815
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798880136114851e-05,
+      "loss": 1.0938,
+      "step": 2816
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279415987388395e-05,
+      "loss": 1.3369,
+      "step": 2817
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278943893627138e-05,
+      "loss": 1.375,
+      "step": 2818
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2784717324418243e-05,
+      "loss": 1.2256,
+      "step": 2819
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779995039465809e-05,
+      "loss": 1.3652,
+      "step": 2820
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277527208255551e-05,
+      "loss": 1.3584,
+      "step": 2821
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.277054845482894e-05,
+      "loss": 1.1875,
+      "step": 2822
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276582415742786e-05,
+      "loss": 1.417,
+      "step": 2823
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2761099191494184e-05,
+      "loss": 1.3062,
+      "step": 2824
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 1.3999,
+      "step": 2825
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751647258597531e-05,
+      "loss": 1.1465,
+      "step": 2826
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2746920293919202e-05,
+      "loss": 1.3438,
+      "step": 2827
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2742192665277569e-05,
+      "loss": 1.3584,
+      "step": 2828
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737464373815351e-05,
+      "loss": 1.3643,
+      "step": 2829
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732735420675439e-05,
+      "loss": 1.3159,
+      "step": 2830
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2728005807000872e-05,
+      "loss": 1.2427,
+      "step": 2831
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272327553393486e-05,
+      "loss": 1.3574,
+      "step": 2832
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271854460262076e-05,
+      "loss": 1.2979,
+      "step": 2833
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2713813014202094e-05,
+      "loss": 1.2808,
+      "step": 2834
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709080769822549e-05,
+      "loss": 1.375,
+      "step": 2835
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2704347870625962e-05,
+      "loss": 1.2681,
+      "step": 2836
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269961431775633e-05,
+      "loss": 1.2285,
+      "step": 2837
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694880112357809e-05,
+      "loss": 1.3262,
+      "step": 2838
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2690145255574718e-05,
+      "loss": 1.2021,
+      "step": 2839
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2685409748551518e-05,
+      "loss": 1.2036,
+      "step": 2840
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2680673592432846e-05,
+      "loss": 1.2495,
+      "step": 2841
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2675936788363483e-05,
+      "loss": 1.3413,
+      "step": 2842
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2671199337488374e-05,
+      "loss": 1.4238,
+      "step": 2843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2666461240952615e-05,
+      "loss": 1.3555,
+      "step": 2844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266172249990146e-05,
+      "loss": 1.3926,
+      "step": 2845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265698311548032e-05,
+      "loss": 1.415,
+      "step": 2846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2652243088834762e-05,
+      "loss": 1.1133,
+      "step": 2847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2647502421110505e-05,
+      "loss": 1.2979,
+      "step": 2848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642761113453421e-05,
+      "loss": 1.3799,
+      "step": 2849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263801916700955e-05,
+      "loss": 1.2158,
+      "step": 2850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263327658292507e-05,
+      "loss": 1.2964,
+      "step": 2851
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2628533362346319e-05,
+      "loss": 1.3945,
+      "step": 2852
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623789506419792e-05,
+      "loss": 1.3491,
+      "step": 2853
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619045016292135e-05,
+      "loss": 1.3477,
+      "step": 2854
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2614299893110152e-05,
+      "loss": 1.3716,
+      "step": 2855
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260955413802079e-05,
+      "loss": 1.3086,
+      "step": 2856
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2604807752171158e-05,
+      "loss": 1.2734,
+      "step": 2857
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260006073670851e-05,
+      "loss": 1.3135,
+      "step": 2858
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595313092780259e-05,
+      "loss": 1.2944,
+      "step": 2859
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2590564821533967e-05,
+      "loss": 1.3604,
+      "step": 2860
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2585815924117347e-05,
+      "loss": 1.3477,
+      "step": 2861
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581066401678261e-05,
+      "loss": 1.2646,
+      "step": 2862
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576316255364732e-05,
+      "loss": 1.5488,
+      "step": 2863
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571565486324921e-05,
+      "loss": 1.3008,
+      "step": 2864
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2566814095707147e-05,
+      "loss": 1.1982,
+      "step": 2865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.256206208465988e-05,
+      "loss": 1.2461,
+      "step": 2866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557309454331734e-05,
+      "loss": 1.2686,
+      "step": 2867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255255620587148e-05,
+      "loss": 1.3535,
+      "step": 2868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254780234042803e-05,
+      "loss": 1.3281,
+      "step": 2869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2543047859150452e-05,
+      "loss": 1.1187,
+      "step": 2870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253829276318796e-05,
+      "loss": 1.4883,
+      "step": 2871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2533537053689921e-05,
+      "loss": 1.269,
+      "step": 2872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528780731805842e-05,
+      "loss": 1.3682,
+      "step": 2873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2524023798685386e-05,
+      "loss": 1.2759,
+      "step": 2874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251926625547836e-05,
+      "loss": 1.2583,
+      "step": 2875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514508103334715e-05,
+      "loss": 1.1929,
+      "step": 2876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2509749343404555e-05,
+      "loss": 1.2881,
+      "step": 2877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250498997683813e-05,
+      "loss": 1.3857,
+      "step": 2878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2500230004785834e-05,
+      "loss": 1.4336,
+      "step": 2879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249546942839821e-05,
+      "loss": 1.2529,
+      "step": 2880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2490708248825944e-05,
+      "loss": 1.1367,
+      "step": 2881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485946467219876e-05,
+      "loss": 1.1953,
+      "step": 2882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2481184084730977e-05,
+      "loss": 1.3574,
+      "step": 2883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476421102510377e-05,
+      "loss": 1.3574,
+      "step": 2884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2471657521709341e-05,
+      "loss": 1.3066,
+      "step": 2885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2466893343479289e-05,
+      "loss": 1.292,
+      "step": 2886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.246212856897177e-05,
+      "loss": 1.2231,
+      "step": 2887
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2457363199338497e-05,
+      "loss": 1.3779,
+      "step": 2888
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2452597235731312e-05,
+      "loss": 1.3647,
+      "step": 2889
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2447830679302206e-05,
+      "loss": 1.3809,
+      "step": 2890
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2443063531203313e-05,
+      "loss": 1.3105,
+      "step": 2891
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438295792586905e-05,
+      "loss": 1.3438,
+      "step": 2892
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2433527464605408e-05,
+      "loss": 1.4287,
+      "step": 2893
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428758548411377e-05,
+      "loss": 1.1733,
+      "step": 2894
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2423989045157518e-05,
+      "loss": 1.3047,
+      "step": 2895
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2419218955996677e-05,
+      "loss": 1.1064,
+      "step": 2896
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414448282081843e-05,
+      "loss": 1.4189,
+      "step": 2897
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2409677024566145e-05,
+      "loss": 1.2153,
+      "step": 2898
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240490518460285e-05,
+      "loss": 1.2578,
+      "step": 2899
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.240013276334537e-05,
+      "loss": 1.2437,
+      "step": 2900
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395359761947252e-05,
+      "loss": 1.2021,
+      "step": 2901
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2390586181562195e-05,
+      "loss": 1.2363,
+      "step": 2902
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385812023344022e-05,
+      "loss": 1.3237,
+      "step": 2903
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238103728844671e-05,
+      "loss": 1.3467,
+      "step": 2904
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376261978024364e-05,
+      "loss": 1.2256,
+      "step": 2905
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371486093231238e-05,
+      "loss": 1.2812,
+      "step": 2906
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236670963522172e-05,
+      "loss": 1.2153,
+      "step": 2907
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361932605150331e-05,
+      "loss": 1.3545,
+      "step": 2908
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2357155004171743e-05,
+      "loss": 1.3008,
+      "step": 2909
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2352376833440755e-05,
+      "loss": 1.1753,
+      "step": 2910
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234759809411231e-05,
+      "loss": 1.5674,
+      "step": 2911
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342818787341482e-05,
+      "loss": 1.3218,
+      "step": 2912
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233803891428349e-05,
+      "loss": 1.4355,
+      "step": 2913
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2333258476093685e-05,
+      "loss": 1.3672,
+      "step": 2914
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328477473927556e-05,
+      "loss": 1.416,
+      "step": 2915
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323695908940729e-05,
+      "loss": 1.457,
+      "step": 2916
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318913782288963e-05,
+      "loss": 1.4805,
+      "step": 2917
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314131095128159e-05,
+      "loss": 1.3945,
+      "step": 2918
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2309347848614344e-05,
+      "loss": 1.4023,
+      "step": 2919
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2304564043903688e-05,
+      "loss": 1.1289,
+      "step": 2920
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299779682152493e-05,
+      "loss": 1.2012,
+      "step": 2921
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2294994764517197e-05,
+      "loss": 1.2764,
+      "step": 2922
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2290209292154372e-05,
+      "loss": 1.2563,
+      "step": 2923
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285423266220721e-05,
+      "loss": 1.4717,
+      "step": 2924
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2280636687873089e-05,
+      "loss": 1.1865,
+      "step": 2925
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.227584955826844e-05,
+      "loss": 1.3281,
+      "step": 2926
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271061878563888e-05,
+      "loss": 1.1309,
+      "step": 2927
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2266273649916669e-05,
+      "loss": 1.3174,
+      "step": 2928
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2261484873484157e-05,
+      "loss": 1.3232,
+      "step": 2929
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2256695550423851e-05,
+      "loss": 1.2534,
+      "step": 2930
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2251905681893392e-05,
+      "loss": 1.4062,
+      "step": 2931
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2247115269050547e-05,
+      "loss": 1.2417,
+      "step": 2932
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242324313053213e-05,
+      "loss": 1.3965,
+      "step": 2933
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237532815059427e-05,
+      "loss": 1.2163,
+      "step": 2934
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232740776227347e-05,
+      "loss": 1.0884,
+      "step": 2935
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2227948197715269e-05,
+      "loss": 1.3877,
+      "step": 2936
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222315508068161e-05,
+      "loss": 1.4336,
+      "step": 2937
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2218361426284928e-05,
+      "loss": 1.1699,
+      "step": 2938
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2213567235683908e-05,
+      "loss": 1.1865,
+      "step": 2939
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2208772510037356e-05,
+      "loss": 1.2241,
+      "step": 2940
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220397725050422e-05,
+      "loss": 1.2239,
+      "step": 2941
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199181458243569e-05,
+      "loss": 1.1899,
+      "step": 2942
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2194385134414608e-05,
+      "loss": 1.1895,
+      "step": 2943
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2189588280176659e-05,
+      "loss": 1.2725,
+      "step": 2944
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218479089668918e-05,
+      "loss": 1.4624,
+      "step": 2945
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2179992985111757e-05,
+      "loss": 1.4775,
+      "step": 2946
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175194546604102e-05,
+      "loss": 1.2871,
+      "step": 2947
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170395582326055e-05,
+      "loss": 1.3408,
+      "step": 2948
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2165596093437582e-05,
+      "loss": 1.3569,
+      "step": 2949
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2160796081098778e-05,
+      "loss": 1.3379,
+      "step": 2950
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2155995546469858e-05,
+      "loss": 1.3433,
+      "step": 2951
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2151194490711177e-05,
+      "loss": 1.3438,
+      "step": 2952
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146392914983203e-05,
+      "loss": 1.373,
+      "step": 2953
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141590820446531e-05,
+      "loss": 1.252,
+      "step": 2954
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2136788208261887e-05,
+      "loss": 1.2803,
+      "step": 2955
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213198507959012e-05,
+      "loss": 1.3169,
+      "step": 2956
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127181435592203e-05,
+      "loss": 1.2593,
+      "step": 2957
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2122377277429231e-05,
+      "loss": 0.9639,
+      "step": 2958
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117572606262433e-05,
+      "loss": 1.2783,
+      "step": 2959
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211276742325315e-05,
+      "loss": 1.3848,
+      "step": 2960
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107961729562852e-05,
+      "loss": 1.1367,
+      "step": 2961
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2103155526353133e-05,
+      "loss": 1.417,
+      "step": 2962
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209834881478571e-05,
+      "loss": 1.4561,
+      "step": 2963
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209354159602242e-05,
+      "loss": 1.3291,
+      "step": 2964
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088733871225228e-05,
+      "loss": 1.3926,
+      "step": 2965
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2083925641556218e-05,
+      "loss": 1.373,
+      "step": 2966
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 1.27,
+      "step": 2967
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074307672251688e-05,
+      "loss": 1.457,
+      "step": 2968
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069497934940944e-05,
+      "loss": 1.3887,
+      "step": 2969
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2064687697407939e-05,
+      "loss": 1.2808,
+      "step": 2970
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205987696081536e-05,
+      "loss": 1.4268,
+      "step": 2971
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205506572632602e-05,
+      "loss": 1.3125,
+      "step": 2972
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050253995102855e-05,
+      "loss": 1.2881,
+      "step": 2973
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045441768308913e-05,
+      "loss": 1.4463,
+      "step": 2974
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2040629047107367e-05,
+      "loss": 1.332,
+      "step": 2975
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035815832661509e-05,
+      "loss": 1.2866,
+      "step": 2976
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2031002126134754e-05,
+      "loss": 1.4053,
+      "step": 2977
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202618792869063e-05,
+      "loss": 1.3389,
+      "step": 2978
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021373241492786e-05,
+      "loss": 1.3809,
+      "step": 2979
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2016558065704986e-05,
+      "loss": 1.3813,
+      "step": 2980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011742402491117e-05,
+      "loss": 1.2041,
+      "step": 2981
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006926253015187e-05,
+      "loss": 1.2666,
+      "step": 2982
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.200210961844131e-05,
+      "loss": 1.3584,
+      "step": 2983
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997292499933729e-05,
+      "loss": 1.4155,
+      "step": 2984
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1992474898656791e-05,
+      "loss": 1.3691,
+      "step": 2985
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987656815774977e-05,
+      "loss": 1.332,
+      "step": 2986
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1982838252452868e-05,
+      "loss": 1.1782,
+      "step": 2987
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1978019209855174e-05,
+      "loss": 1.3291,
+      "step": 2988
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197319968914671e-05,
+      "loss": 1.293,
+      "step": 2989
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196837969149241e-05,
+      "loss": 1.3135,
+      "step": 2990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1963559218057334e-05,
+      "loss": 1.2891,
+      "step": 2991
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1958738270006638e-05,
+      "loss": 1.29,
+      "step": 2992
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1953916848505605e-05,
+      "loss": 1.2773,
+      "step": 2993
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949094954719631e-05,
+      "loss": 1.3584,
+      "step": 2994
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944272589814227e-05,
+      "loss": 1.2979,
+      "step": 2995
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1939449754955015e-05,
+      "loss": 1.4766,
+      "step": 2996
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934626451307727e-05,
+      "loss": 1.2661,
+      "step": 2997
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192980268003822e-05,
+      "loss": 1.2578,
+      "step": 2998
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1924978442312452e-05,
+      "loss": 1.3525,
+      "step": 2999
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.19201537392965e-05,
+      "loss": 1.3062,
+      "step": 3000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915328572156552e-05,
+      "loss": 1.3926,
+      "step": 3001
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1910502942058908e-05,
+      "loss": 1.2021,
+      "step": 3002
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190567685016998e-05,
+      "loss": 1.3555,
+      "step": 3003
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900850297656295e-05,
+      "loss": 1.3984,
+      "step": 3004
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1896023285684481e-05,
+      "loss": 1.3076,
+      "step": 3005
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1891195815421293e-05,
+      "loss": 1.1011,
+      "step": 3006
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.188636788803358e-05,
+      "loss": 1.2026,
+      "step": 3007
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1881539504688315e-05,
+      "loss": 1.2383,
+      "step": 3008
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876710666552578e-05,
+      "loss": 1.4131,
+      "step": 3009
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871881374793546e-05,
+      "loss": 1.3701,
+      "step": 3010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867051630578526e-05,
+      "loss": 1.1899,
+      "step": 3011
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1862221435074918e-05,
+      "loss": 1.3745,
+      "step": 3012
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857390789450245e-05,
+      "loss": 1.3521,
+      "step": 3013
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1852559694872125e-05,
+      "loss": 1.2739,
+      "step": 3014
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1847728152508294e-05,
+      "loss": 1.1191,
+      "step": 3015
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1842896163526594e-05,
+      "loss": 1.3042,
+      "step": 3016
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1838063729094973e-05,
+      "loss": 1.3145,
+      "step": 3017
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1833230850381488e-05,
+      "loss": 1.3506,
+      "step": 3018
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.18283975285543e-05,
+      "loss": 1.3945,
+      "step": 3019
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1823563764781688e-05,
+      "loss": 1.1201,
+      "step": 3020
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1818729560232021e-05,
+      "loss": 1.3188,
+      "step": 3021
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1813894916073792e-05,
+      "loss": 1.4272,
+      "step": 3022
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1809059833475587e-05,
+      "loss": 1.311,
+      "step": 3023
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1804224313606104e-05,
+      "loss": 1.4453,
+      "step": 3024
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799388357634147e-05,
+      "loss": 1.2065,
+      "step": 3025
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1794551966728623e-05,
+      "loss": 1.1685,
+      "step": 3026
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789715142058546e-05,
+      "loss": 1.3086,
+      "step": 3027
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178487788479303e-05,
+      "loss": 1.27,
+      "step": 3028
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1780040196101304e-05,
+      "loss": 1.2295,
+      "step": 3029
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177520207715269e-05,
+      "loss": 1.2495,
+      "step": 3030
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770363529116622e-05,
+      "loss": 1.2217,
+      "step": 3031
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1765524553162635e-05,
+      "loss": 1.3232,
+      "step": 3032
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760685150460363e-05,
+      "loss": 1.3701,
+      "step": 3033
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175584532217955e-05,
+      "loss": 1.2515,
+      "step": 3034
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1751005069490043e-05,
+      "loss": 1.3008,
+      "step": 3035
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746164393561781e-05,
+      "loss": 1.1567,
+      "step": 3036
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741323295564817e-05,
+      "loss": 1.1816,
+      "step": 3037
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 1.3242,
+      "step": 3038
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1731639838045492e-05,
+      "loss": 1.3418,
+      "step": 3039
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172679748086374e-05,
+      "loss": 1.1743,
+      "step": 3040
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721954706294496e-05,
+      "loss": 1.3237,
+      "step": 3041
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171711151550832e-05,
+      "loss": 1.3794,
+      "step": 3042
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1712267909675868e-05,
+      "loss": 1.4834,
+      "step": 3043
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707423889967897e-05,
+      "loss": 1.4473,
+      "step": 3044
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702579457555265e-05,
+      "loss": 1.2798,
+      "step": 3045
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1697734613608929e-05,
+      "loss": 1.2769,
+      "step": 3046
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692889359299946e-05,
+      "loss": 1.3105,
+      "step": 3047
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168804369579947e-05,
+      "loss": 1.2061,
+      "step": 3048
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1683197624278754e-05,
+      "loss": 1.3794,
+      "step": 3049
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1678351145909158e-05,
+      "loss": 1.1372,
+      "step": 3050
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673504261862125e-05,
+      "loss": 1.3252,
+      "step": 3051
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668656973309212e-05,
+      "loss": 1.3369,
+      "step": 3052
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663809281422058e-05,
+      "loss": 1.2358,
+      "step": 3053
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658961187372415e-05,
+      "loss": 1.2007,
+      "step": 3054
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165411269233212e-05,
+      "loss": 1.2646,
+      "step": 3055
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1649263797473116e-05,
+      "loss": 1.2861,
+      "step": 3056
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644414503967436e-05,
+      "loss": 1.3389,
+      "step": 3057
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1639564812987215e-05,
+      "loss": 1.3643,
+      "step": 3058
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1634714725704677e-05,
+      "loss": 1.2246,
+      "step": 3059
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629864243292149e-05,
+      "loss": 1.0991,
+      "step": 3060
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625013366922047e-05,
+      "loss": 1.2783,
+      "step": 3061
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1620162097766888e-05,
+      "loss": 1.3105,
+      "step": 3062
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.161531043699928e-05,
+      "loss": 1.3877,
+      "step": 3063
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1610458385791927e-05,
+      "loss": 1.3179,
+      "step": 3064
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605605945317629e-05,
+      "loss": 1.2527,
+      "step": 3065
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1600753116749279e-05,
+      "loss": 1.2705,
+      "step": 3066
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1595899901259862e-05,
+      "loss": 1.2812,
+      "step": 3067
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1591046300022459e-05,
+      "loss": 1.3428,
+      "step": 3068
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158619231421024e-05,
+      "loss": 1.2764,
+      "step": 3069
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1581337944996475e-05,
+      "loss": 1.2607,
+      "step": 3070
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1576483193554522e-05,
+      "loss": 1.2026,
+      "step": 3071
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571628061057829e-05,
+      "loss": 1.3804,
+      "step": 3072
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1566772548679943e-05,
+      "loss": 1.4165,
+      "step": 3073
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561916657594497e-05,
+      "loss": 1.4336,
+      "step": 3074
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1557060388975221e-05,
+      "loss": 1.2402,
+      "step": 3075
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1552203743995927e-05,
+      "loss": 1.3384,
+      "step": 3076
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154734672383053e-05,
+      "loss": 1.3506,
+      "step": 3077
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1542489329653024e-05,
+      "loss": 1.3047,
+      "step": 3078
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1537631562637506e-05,
+      "loss": 1.3145,
+      "step": 3079
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532773423958146e-05,
+      "loss": 1.3613,
+      "step": 3080
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527914914789222e-05,
+      "loss": 1.4941,
+      "step": 3081
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523056036305089e-05,
+      "loss": 1.3486,
+      "step": 3082
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15181967896802e-05,
+      "loss": 1.2944,
+      "step": 3083
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513337176089091e-05,
+      "loss": 1.2969,
+      "step": 3084
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1508477196706387e-05,
+      "loss": 1.2573,
+      "step": 3085
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503616852706804e-05,
+      "loss": 1.3701,
+      "step": 3086
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1498756145265145e-05,
+      "loss": 1.1636,
+      "step": 3087
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493895075556301e-05,
+      "loss": 1.3101,
+      "step": 3088
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1489033644755247e-05,
+      "loss": 1.1997,
+      "step": 3089
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1484171854037052e-05,
+      "loss": 1.3896,
+      "step": 3090
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1479309704576866e-05,
+      "loss": 1.3584,
+      "step": 3091
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474447197549933e-05,
+      "loss": 1.3154,
+      "step": 3092
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1469584334131578e-05,
+      "loss": 1.1702,
+      "step": 3093
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464721115497208e-05,
+      "loss": 1.2529,
+      "step": 3094
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459857542822327e-05,
+      "loss": 1.1855,
+      "step": 3095
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1454993617282513e-05,
+      "loss": 1.2427,
+      "step": 3096
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.145012934005344e-05,
+      "loss": 1.1704,
+      "step": 3097
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1445264712310855e-05,
+      "loss": 1.2871,
+      "step": 3098
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440399735230599e-05,
+      "loss": 1.2305,
+      "step": 3099
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1435534409988597e-05,
+      "loss": 1.2935,
+      "step": 3100
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1430668737760855e-05,
+      "loss": 1.4453,
+      "step": 3101
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1425802719723467e-05,
+      "loss": 1.2488,
+      "step": 3102
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1420936357052598e-05,
+      "loss": 1.3062,
+      "step": 3103
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1416069650924517e-05,
+      "loss": 1.3945,
+      "step": 3104
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411202602515556e-05,
+      "loss": 1.2139,
+      "step": 3105
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1406335213002144e-05,
+      "loss": 1.1519,
+      "step": 3106
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1401467483560781e-05,
+      "loss": 1.3169,
+      "step": 3107
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1396599415368062e-05,
+      "loss": 1.3516,
+      "step": 3108
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1391731009600655e-05,
+      "loss": 1.3325,
+      "step": 3109
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138686226743531e-05,
+      "loss": 1.4209,
+      "step": 3110
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1381993190048862e-05,
+      "loss": 1.3232,
+      "step": 3111
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137712377861822e-05,
+      "loss": 1.2661,
+      "step": 3112
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1372254034320385e-05,
+      "loss": 1.3174,
+      "step": 3113
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367383958332427e-05,
+      "loss": 1.4199,
+      "step": 3114
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1362513551831502e-05,
+      "loss": 1.3989,
+      "step": 3115
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357642815994849e-05,
+      "loss": 1.25,
+      "step": 3116
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135277175199978e-05,
+      "loss": 1.2842,
+      "step": 3117
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347900361023689e-05,
+      "loss": 1.3281,
+      "step": 3118
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1343028644244052e-05,
+      "loss": 1.3159,
+      "step": 3119
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338156602838418e-05,
+      "loss": 1.1746,
+      "step": 3120
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1333284237984415e-05,
+      "loss": 1.2539,
+      "step": 3121
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328411550859758e-05,
+      "loss": 1.2021,
+      "step": 3122
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1323538542642227e-05,
+      "loss": 1.4102,
+      "step": 3123
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1318665214509688e-05,
+      "loss": 1.2695,
+      "step": 3124
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313791567640084e-05,
+      "loss": 1.4487,
+      "step": 3125
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1308917603211433e-05,
+      "loss": 1.2217,
+      "step": 3126
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1304043322401832e-05,
+      "loss": 1.2632,
+      "step": 3127
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129916872638945e-05,
+      "loss": 1.2871,
+      "step": 3128
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294293816352535e-05,
+      "loss": 1.4521,
+      "step": 3129
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128941859346941e-05,
+      "loss": 1.3364,
+      "step": 3130
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284543058918476e-05,
+      "loss": 1.4697,
+      "step": 3131
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1279667213878206e-05,
+      "loss": 1.4058,
+      "step": 3132
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127479105952715e-05,
+      "loss": 1.2754,
+      "step": 3133
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1269914597043936e-05,
+      "loss": 1.4893,
+      "step": 3134
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265037827607257e-05,
+      "loss": 1.3184,
+      "step": 3135
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1260160752395892e-05,
+      "loss": 1.2734,
+      "step": 3136
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255283372588686e-05,
+      "loss": 1.458,
+      "step": 3137
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1250405689364561e-05,
+      "loss": 1.4434,
+      "step": 3138
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1245527703902506e-05,
+      "loss": 1.1733,
+      "step": 3139
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240649417381593e-05,
+      "loss": 1.3711,
+      "step": 3140
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123577083098096e-05,
+      "loss": 1.231,
+      "step": 3141
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230891945879817e-05,
+      "loss": 1.3311,
+      "step": 3142
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226012763257452e-05,
+      "loss": 1.3145,
+      "step": 3143
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221133284293223e-05,
+      "loss": 1.335,
+      "step": 3144
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1216253510166554e-05,
+      "loss": 1.3892,
+      "step": 3145
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211373442056948e-05,
+      "loss": 1.2856,
+      "step": 3146
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1206493081143972e-05,
+      "loss": 1.1816,
+      "step": 3147
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.120161242860727e-05,
+      "loss": 1.2095,
+      "step": 3148
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119673148562655e-05,
+      "loss": 1.2163,
+      "step": 3149
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1191850253381602e-05,
+      "loss": 1.3369,
+      "step": 3150
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186968733052271e-05,
+      "loss": 1.0894,
+      "step": 3151
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1182086925818476e-05,
+      "loss": 1.2178,
+      "step": 3152
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1177204832860214e-05,
+      "loss": 1.2959,
+      "step": 3153
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172322455357545e-05,
+      "loss": 1.3442,
+      "step": 3154
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167439794490595e-05,
+      "loss": 1.3872,
+      "step": 3155
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1162556851439563e-05,
+      "loss": 1.3833,
+      "step": 3156
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157673627384711e-05,
+      "loss": 1.2456,
+      "step": 3157
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152790123506374e-05,
+      "loss": 1.4229,
+      "step": 3158
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1147906340984955e-05,
+      "loss": 1.332,
+      "step": 3159
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143022281000922e-05,
+      "loss": 1.291,
+      "step": 3160
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138137944734807e-05,
+      "loss": 1.4688,
+      "step": 3161
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1133253333367213e-05,
+      "loss": 1.1953,
+      "step": 3162
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1128368448078814e-05,
+      "loss": 1.354,
+      "step": 3163
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123483290050339e-05,
+      "loss": 1.2266,
+      "step": 3164
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1118597860462593e-05,
+      "loss": 1.3345,
+      "step": 3165
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113712160496438e-05,
+      "loss": 1.4326,
+      "step": 3166
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1108826191332807e-05,
+      "loss": 1.2998,
+      "step": 3167
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11039399541527e-05,
+      "loss": 1.187,
+      "step": 3168
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1099053450137176e-05,
+      "loss": 1.3008,
+      "step": 3169
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.109416668046736e-05,
+      "loss": 1.3535,
+      "step": 3170
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1089279646324445e-05,
+      "loss": 1.3354,
+      "step": 3171
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084392348889686e-05,
+      "loss": 1.2705,
+      "step": 3172
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079504789344395e-05,
+      "loss": 1.2603,
+      "step": 3173
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.107461696886996e-05,
+      "loss": 1.3594,
+      "step": 3174
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069728888647827e-05,
+      "loss": 1.1445,
+      "step": 3175
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064840549859492e-05,
+      "loss": 1.3862,
+      "step": 3176
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1059951953686537e-05,
+      "loss": 1.4048,
+      "step": 3177
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055063101310581e-05,
+      "loss": 1.0747,
+      "step": 3178
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050173993913332e-05,
+      "loss": 1.2383,
+      "step": 3179
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 1.3623,
+      "step": 3180
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040395018782014e-05,
+      "loss": 1.3438,
+      "step": 3181
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103550515341164e-05,
+      "loss": 1.2896,
+      "step": 3182
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1030615037747354e-05,
+      "loss": 1.1489,
+      "step": 3183
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1025724672971159e-05,
+      "loss": 1.3672,
+      "step": 3184
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102083406026511e-05,
+      "loss": 1.1401,
+      "step": 3185
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015943200811325e-05,
+      "loss": 1.3066,
+      "step": 3186
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1011052095791987e-05,
+      "loss": 1.3013,
+      "step": 3187
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1006160746389333e-05,
+      "loss": 1.3423,
+      "step": 3188
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1001269153785661e-05,
+      "loss": 1.3691,
+      "step": 3189
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996377319163326e-05,
+      "loss": 1.29,
+      "step": 3190
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099148524370474e-05,
+      "loss": 1.3682,
+      "step": 3191
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986592928592382e-05,
+      "loss": 1.1548,
+      "step": 3192
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981700375008777e-05,
+      "loss": 1.2441,
+      "step": 3193
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0976807584136515e-05,
+      "loss": 1.2891,
+      "step": 3194
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0971914557158243e-05,
+      "loss": 1.3374,
+      "step": 3195
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967021295256664e-05,
+      "loss": 1.3218,
+      "step": 3196
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962127799614537e-05,
+      "loss": 1.186,
+      "step": 3197
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0957234071414675e-05,
+      "loss": 1.3442,
+      "step": 3198
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0952340111839958e-05,
+      "loss": 1.458,
+      "step": 3199
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0947445922073305e-05,
+      "loss": 1.4287,
+      "step": 3200
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0942551503297704e-05,
+      "loss": 1.2964,
+      "step": 3201
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937656856696199e-05,
+      "loss": 1.3311,
+      "step": 3202
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0932761983451879e-05,
+      "loss": 1.3662,
+      "step": 3203
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092786688474789e-05,
+      "loss": 1.2715,
+      "step": 3204
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922971561767442e-05,
+      "loss": 1.375,
+      "step": 3205
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091807601569379e-05,
+      "loss": 1.2139,
+      "step": 3206
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0913180247710245e-05,
+      "loss": 1.2002,
+      "step": 3207
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908284259000173e-05,
+      "loss": 1.1519,
+      "step": 3208
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0903388050746998e-05,
+      "loss": 1.0688,
+      "step": 3209
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089849162413418e-05,
+      "loss": 1.3262,
+      "step": 3210
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893594980345254e-05,
+      "loss": 1.2256,
+      "step": 3211
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0888698120563794e-05,
+      "loss": 1.2676,
+      "step": 3212
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883801045973425e-05,
+      "loss": 1.2681,
+      "step": 3213
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878903757757835e-05,
+      "loss": 1.1011,
+      "step": 3214
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874006257100753e-05,
+      "loss": 1.3916,
+      "step": 3215
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0869108545185963e-05,
+      "loss": 1.2554,
+      "step": 3216
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.08642106231973e-05,
+      "loss": 1.4219,
+      "step": 3217
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0859312492318655e-05,
+      "loss": 1.1807,
+      "step": 3218
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854414153733956e-05,
+      "loss": 1.1333,
+      "step": 3219
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849515608627199e-05,
+      "loss": 1.2998,
+      "step": 3220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844616858182411e-05,
+      "loss": 1.2827,
+      "step": 3221
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839717903583684e-05,
+      "loss": 1.2671,
+      "step": 3222
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834818746015155e-05,
+      "loss": 1.3779,
+      "step": 3223
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0829919386661006e-05,
+      "loss": 1.168,
+      "step": 3224
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825019826705467e-05,
+      "loss": 1.248,
+      "step": 3225
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0820120067332827e-05,
+      "loss": 1.2793,
+      "step": 3226
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0815220109727412e-05,
+      "loss": 1.2979,
+      "step": 3227
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.08103199550736e-05,
+      "loss": 1.3633,
+      "step": 3228
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805419604555817e-05,
+      "loss": 1.2495,
+      "step": 3229
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0800519059358539e-05,
+      "loss": 1.4395,
+      "step": 3230
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0795618320666282e-05,
+      "loss": 1.4414,
+      "step": 3231
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790717389663612e-05,
+      "loss": 1.3257,
+      "step": 3232
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785816267535146e-05,
+      "loss": 1.3223,
+      "step": 3233
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780914955465545e-05,
+      "loss": 1.373,
+      "step": 3234
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776013454639507e-05,
+      "loss": 1.1484,
+      "step": 3235
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077111176624179e-05,
+      "loss": 1.2998,
+      "step": 3236
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076620989145719e-05,
+      "loss": 1.4854,
+      "step": 3237
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761307831470544e-05,
+      "loss": 1.1953,
+      "step": 3238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.075640558746674e-05,
+      "loss": 1.4092,
+      "step": 3239
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.075150316063071e-05,
+      "loss": 1.3145,
+      "step": 3240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746600552147429e-05,
+      "loss": 1.2075,
+      "step": 3241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0741697763201913e-05,
+      "loss": 1.3237,
+      "step": 3242
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073679479497923e-05,
+      "loss": 1.2979,
+      "step": 3243
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0731891648664477e-05,
+      "loss": 1.5615,
+      "step": 3244
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0726988325442811e-05,
+      "loss": 1.3691,
+      "step": 3245
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072208482649942e-05,
+      "loss": 1.3262,
+      "step": 3246
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717181153019539e-05,
+      "loss": 1.272,
+      "step": 3247
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0712277306188439e-05,
+      "loss": 1.2144,
+      "step": 3248
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070737328719145e-05,
+      "loss": 1.2285,
+      "step": 3249
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702469097213919e-05,
+      "loss": 1.3652,
+      "step": 3250
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0697564737441254e-05,
+      "loss": 1.311,
+      "step": 3251
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0692660209058896e-05,
+      "loss": 1.3989,
+      "step": 3252
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0687755513252326e-05,
+      "loss": 1.4336,
+      "step": 3253
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0682850651207072e-05,
+      "loss": 1.3271,
+      "step": 3254
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677945624108691e-05,
+      "loss": 1.3525,
+      "step": 3255
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673040433142792e-05,
+      "loss": 1.4102,
+      "step": 3256
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0668135079495012e-05,
+      "loss": 1.3779,
+      "step": 3257
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066322956435104e-05,
+      "loss": 1.3564,
+      "step": 3258
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658323888896594e-05,
+      "loss": 1.3945,
+      "step": 3259
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653418054317432e-05,
+      "loss": 1.3828,
+      "step": 3260
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0648512061799359e-05,
+      "loss": 1.2622,
+      "step": 3261
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643605912528206e-05,
+      "loss": 1.1641,
+      "step": 3262
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0638699607689848e-05,
+      "loss": 1.3486,
+      "step": 3263
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633793148470199e-05,
+      "loss": 1.5176,
+      "step": 3264
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0628886536055208e-05,
+      "loss": 1.3413,
+      "step": 3265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623979771630862e-05,
+      "loss": 1.3887,
+      "step": 3266
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619072856383181e-05,
+      "loss": 1.2039,
+      "step": 3267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0614165791498232e-05,
+      "loss": 1.1816,
+      "step": 3268
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0609258578162105e-05,
+      "loss": 1.2773,
+      "step": 3269
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604351217560935e-05,
+      "loss": 1.4336,
+      "step": 3270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0599443710880888e-05,
+      "loss": 1.1948,
+      "step": 3271
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0594536059308165e-05,
+      "loss": 1.1074,
+      "step": 3272
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589628264029005e-05,
+      "loss": 1.2939,
+      "step": 3273
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058472032622968e-05,
+      "loss": 1.1279,
+      "step": 3274
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05798122470965e-05,
+      "loss": 1.4219,
+      "step": 3275
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0574904027815802e-05,
+      "loss": 1.2432,
+      "step": 3276
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0569995669573966e-05,
+      "loss": 1.2236,
+      "step": 3277
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0565087173557396e-05,
+      "loss": 1.2998,
+      "step": 3278
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560178540952537e-05,
+      "loss": 1.3042,
+      "step": 3279
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.055526977294586e-05,
+      "loss": 1.1738,
+      "step": 3280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0550360870723876e-05,
+      "loss": 1.2466,
+      "step": 3281
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545451835473121e-05,
+      "loss": 1.2964,
+      "step": 3282
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0540542668380173e-05,
+      "loss": 1.1821,
+      "step": 3283
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0535633370631631e-05,
+      "loss": 1.0764,
+      "step": 3284
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0530723943414135e-05,
+      "loss": 1.3779,
+      "step": 3285
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052581438791435e-05,
+      "loss": 1.2534,
+      "step": 3286
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.052090470531897e-05,
+      "loss": 1.3701,
+      "step": 3287
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0515994896814731e-05,
+      "loss": 1.3047,
+      "step": 3288
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0511084963588389e-05,
+      "loss": 1.2861,
+      "step": 3289
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0506174906826731e-05,
+      "loss": 1.3154,
+      "step": 3290
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501264727716578e-05,
+      "loss": 1.4248,
+      "step": 3291
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049635442744478e-05,
+      "loss": 1.2505,
+      "step": 3292
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0491444007198211e-05,
+      "loss": 1.3877,
+      "step": 3293
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486533468163782e-05,
+      "loss": 1.2056,
+      "step": 3294
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0481622811528426e-05,
+      "loss": 1.1875,
+      "step": 3295
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476712038479111e-05,
+      "loss": 1.1914,
+      "step": 3296
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471801150202824e-05,
+      "loss": 1.3594,
+      "step": 3297
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.046689014788659e-05,
+      "loss": 1.3569,
+      "step": 3298
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461979032717452e-05,
+      "loss": 1.3789,
+      "step": 3299
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045706780588249e-05,
+      "loss": 1.2988,
+      "step": 3300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452156468568802e-05,
+      "loss": 1.2319,
+      "step": 3301
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0447245021963515e-05,
+      "loss": 1.2349,
+      "step": 3302
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044233346725379e-05,
+      "loss": 1.2449,
+      "step": 3303
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0437421805626804e-05,
+      "loss": 1.4023,
+      "step": 3304
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0432510038269765e-05,
+      "loss": 1.2305,
+      "step": 3305
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0427598166369908e-05,
+      "loss": 1.2598,
+      "step": 3306
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422686191114486e-05,
+      "loss": 1.2134,
+      "step": 3307
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0417774113690788e-05,
+      "loss": 1.2485,
+      "step": 3308
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412861935286115e-05,
+      "loss": 1.0889,
+      "step": 3309
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.04079496570878e-05,
+      "loss": 1.168,
+      "step": 3310
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0403037280283204e-05,
+      "loss": 1.416,
+      "step": 3311
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398124806059702e-05,
+      "loss": 1.2725,
+      "step": 3312
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0393212235604701e-05,
+      "loss": 1.3813,
+      "step": 3313
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0388299570105624e-05,
+      "loss": 1.4053,
+      "step": 3314
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383386810749926e-05,
+      "loss": 1.2397,
+      "step": 3315
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0378473958725075e-05,
+      "loss": 1.3105,
+      "step": 3316
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373561015218566e-05,
+      "loss": 1.4219,
+      "step": 3317
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368647981417917e-05,
+      "loss": 1.292,
+      "step": 3318
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0363734858510664e-05,
+      "loss": 1.4619,
+      "step": 3319
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0358821647684372e-05,
+      "loss": 1.3008,
+      "step": 3320
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035390835012662e-05,
+      "loss": 1.2544,
+      "step": 3321
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0348994967025012e-05,
+      "loss": 1.3027,
+      "step": 3322
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0344081499567167e-05,
+      "loss": 1.3286,
+      "step": 3323
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339167948940734e-05,
+      "loss": 1.2505,
+      "step": 3324
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0334254316333371e-05,
+      "loss": 1.2627,
+      "step": 3325
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329340602932767e-05,
+      "loss": 1.3516,
+      "step": 3326
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.032442680992662e-05,
+      "loss": 1.3877,
+      "step": 3327
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0319512938502654e-05,
+      "loss": 1.375,
+      "step": 3328
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0314598989848615e-05,
+      "loss": 1.2754,
+      "step": 3329
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309684965152254e-05,
+      "loss": 1.2998,
+      "step": 3330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0304770865601357e-05,
+      "loss": 1.2905,
+      "step": 3331
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299856692383714e-05,
+      "loss": 1.2573,
+      "step": 3332
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0294942446687147e-05,
+      "loss": 1.4199,
+      "step": 3333
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0290028129699479e-05,
+      "loss": 1.3984,
+      "step": 3334
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285113742608562e-05,
+      "loss": 1.2739,
+      "step": 3335
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280199286602262e-05,
+      "loss": 1.3604,
+      "step": 3336
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0275284762868463e-05,
+      "loss": 1.1724,
+      "step": 3337
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270370172595064e-05,
+      "loss": 1.2124,
+      "step": 3338
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0265455516969978e-05,
+      "loss": 1.3057,
+      "step": 3339
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0260540797181137e-05,
+      "loss": 1.2676,
+      "step": 3340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255626014416484e-05,
+      "loss": 1.29,
+      "step": 3341
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0250711169863987e-05,
+      "loss": 1.4878,
+      "step": 3342
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0245796264711616e-05,
+      "loss": 1.3291,
+      "step": 3343
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240881300147363e-05,
+      "loss": 1.3994,
+      "step": 3344
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0235966277359238e-05,
+      "loss": 1.3433,
+      "step": 3345
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0231051197535253e-05,
+      "loss": 1.3105,
+      "step": 3346
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022613606186345e-05,
+      "loss": 1.3706,
+      "step": 3347
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022122087153187e-05,
+      "loss": 1.4365,
+      "step": 3348
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0216305627728575e-05,
+      "loss": 1.1694,
+      "step": 3349
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211390331641635e-05,
+      "loss": 1.3315,
+      "step": 3350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206474984459139e-05,
+      "loss": 1.3779,
+      "step": 3351
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0201559587369185e-05,
+      "loss": 1.4316,
+      "step": 3352
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019664414155988e-05,
+      "loss": 1.3662,
+      "step": 3353
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191728648219345e-05,
+      "loss": 1.3496,
+      "step": 3354
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186813108535718e-05,
+      "loss": 1.2729,
+      "step": 3355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181897523697142e-05,
+      "loss": 1.3066,
+      "step": 3356
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0176981894891768e-05,
+      "loss": 1.4224,
+      "step": 3357
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0172066223307769e-05,
+      "loss": 1.29,
+      "step": 3358
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167150510133314e-05,
+      "loss": 1.3169,
+      "step": 3359
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162234756556593e-05,
+      "loss": 1.2305,
+      "step": 3360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0157318963765805e-05,
+      "loss": 1.3535,
+      "step": 3361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152403132949149e-05,
+      "loss": 1.4268,
+      "step": 3362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0147487265294844e-05,
+      "loss": 1.4854,
+      "step": 3363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0142571361991111e-05,
+      "loss": 1.251,
+      "step": 3364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137655424226189e-05,
+      "loss": 1.2202,
+      "step": 3365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013273945318831e-05,
+      "loss": 1.375,
+      "step": 3366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127823450065728e-05,
+      "loss": 1.2925,
+      "step": 3367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0122907416046697e-05,
+      "loss": 1.4463,
+      "step": 3368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.011799135231948e-05,
+      "loss": 1.4873,
+      "step": 3369
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0113075260072355e-05,
+      "loss": 1.2578,
+      "step": 3370
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0108159140493591e-05,
+      "loss": 1.3867,
+      "step": 3371
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0103242994771477e-05,
+      "loss": 1.3306,
+      "step": 3372
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098326824094303e-05,
+      "loss": 1.4053,
+      "step": 3373
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093410629650367e-05,
+      "loss": 1.3779,
+      "step": 3374
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0088494412627969e-05,
+      "loss": 1.2207,
+      "step": 3375
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083578174215422e-05,
+      "loss": 1.3345,
+      "step": 3376
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078661915601034e-05,
+      "loss": 1.2959,
+      "step": 3377
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0073745637973125e-05,
+      "loss": 1.3247,
+      "step": 3378
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006882934252002e-05,
+      "loss": 1.5293,
+      "step": 3379
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063913030430044e-05,
+      "loss": 1.4043,
+      "step": 3380
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0058996702891527e-05,
+      "loss": 1.4004,
+      "step": 3381
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0054080361092806e-05,
+      "loss": 1.2998,
+      "step": 3382
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049164006222218e-05,
+      "loss": 1.3696,
+      "step": 3383
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0044247639468106e-05,
+      "loss": 1.4258,
+      "step": 3384
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039331262018812e-05,
+      "loss": 1.3218,
+      "step": 3385
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034414875062689e-05,
+      "loss": 1.4404,
+      "step": 3386
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029498479788076e-05,
+      "loss": 1.4072,
+      "step": 3387
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002458207738333e-05,
+      "loss": 1.4048,
+      "step": 3388
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019665669036804e-05,
+      "loss": 1.1914,
+      "step": 3389
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0014749255936853e-05,
+      "loss": 1.189,
+      "step": 3390
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009832839271831e-05,
+      "loss": 1.2695,
+      "step": 3391
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0004916420230096e-05,
+      "loss": 1.333,
+      "step": 3392
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 1.2622,
+      "step": 3393
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.995083579769907e-06,
+      "loss": 1.2588,
+      "step": 3394
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.99016716072817e-06,
+      "loss": 1.4854,
+      "step": 3395
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.985250744063152e-06,
+      "loss": 1.4229,
+      "step": 3396
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980334330963197e-06,
+      "loss": 1.3213,
+      "step": 3397
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975417922616671e-06,
+      "loss": 1.1997,
+      "step": 3398
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970501520211925e-06,
+      "loss": 1.3564,
+      "step": 3399
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965585124937318e-06,
+      "loss": 1.1348,
+      "step": 3400
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960668737981191e-06,
+      "loss": 1.1489,
+      "step": 3401
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.955752360531896e-06,
+      "loss": 1.0498,
+      "step": 3402
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950835993777784e-06,
+      "loss": 1.3125,
+      "step": 3403
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.945919638907195e-06,
+      "loss": 1.2188,
+      "step": 3404
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.941003297108475e-06,
+      "loss": 1.2178,
+      "step": 3405
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93608696956996e-06,
+      "loss": 1.3901,
+      "step": 3406
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.931170657479981e-06,
+      "loss": 1.2695,
+      "step": 3407
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.926254362026875e-06,
+      "loss": 1.2275,
+      "step": 3408
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.921338084398971e-06,
+      "loss": 1.1594,
+      "step": 3409
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.916421825784583e-06,
+      "loss": 1.2412,
+      "step": 3410
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.911505587372033e-06,
+      "loss": 1.1216,
+      "step": 3411
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906589370349636e-06,
+      "loss": 1.311,
+      "step": 3412
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901673175905699e-06,
+      "loss": 1.1621,
+      "step": 3413
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896757005228527e-06,
+      "loss": 1.2998,
+      "step": 3414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891840859506412e-06,
+      "loss": 1.376,
+      "step": 3415
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.886924739927648e-06,
+      "loss": 1.3818,
+      "step": 3416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.882008647680519e-06,
+      "loss": 1.3833,
+      "step": 3417
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.877092583953308e-06,
+      "loss": 1.2075,
+      "step": 3418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872176549934275e-06,
+      "loss": 1.2544,
+      "step": 3419
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.867260546811694e-06,
+      "loss": 1.3564,
+      "step": 3420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862344575773815e-06,
+      "loss": 1.248,
+      "step": 3421
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857428638008889e-06,
+      "loss": 1.3677,
+      "step": 3422
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.852512734705159e-06,
+      "loss": 1.2197,
+      "step": 3423
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847596867050854e-06,
+      "loss": 1.3809,
+      "step": 3424
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.842681036234199e-06,
+      "loss": 1.3828,
+      "step": 3425
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837765243443407e-06,
+      "loss": 1.4746,
+      "step": 3426
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832849489866686e-06,
+      "loss": 1.2715,
+      "step": 3427
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.827933776692236e-06,
+      "loss": 1.2651,
+      "step": 3428
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.823018105108234e-06,
+      "loss": 1.2324,
+      "step": 3429
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818102476302861e-06,
+      "loss": 1.3848,
+      "step": 3430
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.813186891464282e-06,
+      "loss": 1.2568,
+      "step": 3431
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808271351780657e-06,
+      "loss": 1.3936,
+      "step": 3432
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803355858440123e-06,
+      "loss": 1.2056,
+      "step": 3433
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798440412630818e-06,
+      "loss": 1.334,
+      "step": 3434
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793525015540861e-06,
+      "loss": 1.3301,
+      "step": 3435
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788609668358365e-06,
+      "loss": 1.1519,
+      "step": 3436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.783694372271429e-06,
+      "loss": 1.4199,
+      "step": 3437
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.778779128468133e-06,
+      "loss": 1.2603,
+      "step": 3438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773863938136552e-06,
+      "loss": 1.1926,
+      "step": 3439
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.768948802464747e-06,
+      "loss": 1.1156,
+      "step": 3440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.764033722640767e-06,
+      "loss": 1.2617,
+      "step": 3441
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759118699852638e-06,
+      "loss": 1.3511,
+      "step": 3442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.754203735288387e-06,
+      "loss": 1.1489,
+      "step": 3443
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.749288830136016e-06,
+      "loss": 1.1826,
+      "step": 3444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744373985583516e-06,
+      "loss": 1.2485,
+      "step": 3445
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.739459202818868e-06,
+      "loss": 1.3896,
+      "step": 3446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734544483030027e-06,
+      "loss": 1.2529,
+      "step": 3447
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729629827404938e-06,
+      "loss": 1.4229,
+      "step": 3448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.724715237131539e-06,
+      "loss": 1.2139,
+      "step": 3449
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.719800713397741e-06,
+      "loss": 1.3301,
+      "step": 3450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.714886257391441e-06,
+      "loss": 1.5645,
+      "step": 3451
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.709971870300525e-06,
+      "loss": 1.311,
+      "step": 3452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.705057553312857e-06,
+      "loss": 1.3501,
+      "step": 3453
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700143307616286e-06,
+      "loss": 1.1816,
+      "step": 3454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.695229134398648e-06,
+      "loss": 1.3301,
+      "step": 3455
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690315034847749e-06,
+      "loss": 1.3613,
+      "step": 3456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685401010151388e-06,
+      "loss": 1.2627,
+      "step": 3457
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.680487061497348e-06,
+      "loss": 1.3384,
+      "step": 3458
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.675573190073382e-06,
+      "loss": 1.4668,
+      "step": 3459
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.670659397067236e-06,
+      "loss": 1.269,
+      "step": 3460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66574568366663e-06,
+      "loss": 1.376,
+      "step": 3461
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.66083205105927e-06,
+      "loss": 1.4307,
+      "step": 3462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.655918500432833e-06,
+      "loss": 1.2246,
+      "step": 3463
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.651005032974994e-06,
+      "loss": 1.3242,
+      "step": 3464
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.646091649873384e-06,
+      "loss": 1.3994,
+      "step": 3465
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.641178352315631e-06,
+      "loss": 1.2744,
+      "step": 3466
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.636265141489338e-06,
+      "loss": 1.103,
+      "step": 3467
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631352018582088e-06,
+      "loss": 1.2041,
+      "step": 3468
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626438984781439e-06,
+      "loss": 1.3853,
+      "step": 3469
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.621526041274928e-06,
+      "loss": 1.3525,
+      "step": 3470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616613189250075e-06,
+      "loss": 1.377,
+      "step": 3471
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.611700429894376e-06,
+      "loss": 1.2378,
+      "step": 3472
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.606787764395304e-06,
+      "loss": 1.2944,
+      "step": 3473
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601875193940301e-06,
+      "loss": 1.3135,
+      "step": 3474
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.5969627197168e-06,
+      "loss": 1.4492,
+      "step": 3475
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.592050342912202e-06,
+      "loss": 1.4346,
+      "step": 3476
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58713806471389e-06,
+      "loss": 1.3447,
+      "step": 3477
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582225886309217e-06,
+      "loss": 1.236,
+      "step": 3478
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.577313808885517e-06,
+      "loss": 1.1426,
+      "step": 3479
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572401833630094e-06,
+      "loss": 1.3418,
+      "step": 3480
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.567489961730235e-06,
+      "loss": 1.2817,
+      "step": 3481
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.5625781943732e-06,
+      "loss": 1.1597,
+      "step": 3482
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557666532746214e-06,
+      "loss": 1.2744,
+      "step": 3483
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.552754978036488e-06,
+      "loss": 1.3809,
+      "step": 3484
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547843531431202e-06,
+      "loss": 1.311,
+      "step": 3485
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.542932194117514e-06,
+      "loss": 1.2861,
+      "step": 3486
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.538020967282551e-06,
+      "loss": 1.2378,
+      "step": 3487
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.533109852113413e-06,
+      "loss": 1.3325,
+      "step": 3488
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.528198849797177e-06,
+      "loss": 1.3691,
+      "step": 3489
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.52328796152089e-06,
+      "loss": 1.3174,
+      "step": 3490
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.518377188471579e-06,
+      "loss": 1.4033,
+      "step": 3491
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513466531836223e-06,
+      "loss": 1.1157,
+      "step": 3492
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.508555992801792e-06,
+      "loss": 1.3242,
+      "step": 3493
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.503645572555224e-06,
+      "loss": 1.3125,
+      "step": 3494
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.498735272283423e-06,
+      "loss": 1.2212,
+      "step": 3495
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.493825093173272e-06,
+      "loss": 1.4307,
+      "step": 3496
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488915036411615e-06,
+      "loss": 1.2681,
+      "step": 3497
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48400510318527e-06,
+      "loss": 1.0654,
+      "step": 3498
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47909529468103e-06,
+      "loss": 1.4199,
+      "step": 3499
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.474185612085656e-06,
+      "loss": 1.3433,
+      "step": 3500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469276056585869e-06,
+      "loss": 1.2969,
+      "step": 3501
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46436662936837e-06,
+      "loss": 1.3125,
+      "step": 3502
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45945733161983e-06,
+      "loss": 1.4072,
+      "step": 3503
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45454816452688e-06,
+      "loss": 1.3809,
+      "step": 3504
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.449639129276129e-06,
+      "loss": 1.2808,
+      "step": 3505
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444730227054144e-06,
+      "loss": 1.3291,
+      "step": 3506
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439821459047468e-06,
+      "loss": 1.3286,
+      "step": 3507
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.434912826442604e-06,
+      "loss": 1.2563,
+      "step": 3508
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.430004330426039e-06,
+      "loss": 1.2925,
+      "step": 3509
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.4250959721842e-06,
+      "loss": 1.4268,
+      "step": 3510
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420187752903502e-06,
+      "loss": 1.4502,
+      "step": 3511
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.415279673770322e-06,
+      "loss": 1.2693,
+      "step": 3512
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410371735970996e-06,
+      "loss": 1.252,
+      "step": 3513
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.405463940691839e-06,
+      "loss": 1.1382,
+      "step": 3514
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.400556289119115e-06,
+      "loss": 1.2124,
+      "step": 3515
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395648782439067e-06,
+      "loss": 1.3809,
+      "step": 3516
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.390741421837895e-06,
+      "loss": 1.2769,
+      "step": 3517
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.385834208501772e-06,
+      "loss": 1.2754,
+      "step": 3518
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.38092714361682e-06,
+      "loss": 1.3076,
+      "step": 3519
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376020228369142e-06,
+      "loss": 1.334,
+      "step": 3520
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.371113463944795e-06,
+      "loss": 1.2632,
+      "step": 3521
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366206851529803e-06,
+      "loss": 1.1821,
+      "step": 3522
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.361300392310157e-06,
+      "loss": 1.2559,
+      "step": 3523
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.3563940874718e-06,
+      "loss": 1.3408,
+      "step": 3524
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.351487938200645e-06,
+      "loss": 1.5,
+      "step": 3525
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346581945682568e-06,
+      "loss": 1.2954,
+      "step": 3526
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341676111103411e-06,
+      "loss": 1.1797,
+      "step": 3527
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.336770435648963e-06,
+      "loss": 1.3228,
+      "step": 3528
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.331864920504991e-06,
+      "loss": 1.2305,
+      "step": 3529
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.326959566857212e-06,
+      "loss": 1.3696,
+      "step": 3530
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322054375891308e-06,
+      "loss": 1.2441,
+      "step": 3531
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.317149348792933e-06,
+      "loss": 1.2988,
+      "step": 3532
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.312244486747675e-06,
+      "loss": 1.4844,
+      "step": 3533
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307339790941107e-06,
+      "loss": 1.3223,
+      "step": 3534
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.302435262558748e-06,
+      "loss": 1.1431,
+      "step": 3535
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.297530902786086e-06,
+      "loss": 1.3193,
+      "step": 3536
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292626712808557e-06,
+      "loss": 1.3877,
+      "step": 3537
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.287722693811563e-06,
+      "loss": 1.3682,
+      "step": 3538
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282818846980463e-06,
+      "loss": 1.1604,
+      "step": 3539
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.277915173500582e-06,
+      "loss": 1.3228,
+      "step": 3540
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.273011674557194e-06,
+      "loss": 1.0054,
+      "step": 3541
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.268108351335526e-06,
+      "loss": 1.2959,
+      "step": 3542
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263205205020776e-06,
+      "loss": 1.2637,
+      "step": 3543
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.258302236798089e-06,
+      "loss": 1.2622,
+      "step": 3544
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253399447852573e-06,
+      "loss": 1.3242,
+      "step": 3545
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.248496839369293e-06,
+      "loss": 1.2544,
+      "step": 3546
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.243594412533263e-06,
+      "loss": 1.2266,
+      "step": 3547
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.238692168529458e-06,
+      "loss": 1.3618,
+      "step": 3548
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.233790108542812e-06,
+      "loss": 1.3823,
+      "step": 3549
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.228888233758213e-06,
+      "loss": 1.2119,
+      "step": 3550
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223986545360496e-06,
+      "loss": 1.3657,
+      "step": 3551
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21908504453446e-06,
+      "loss": 1.3027,
+      "step": 3552
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.214183732464856e-06,
+      "loss": 1.3247,
+      "step": 3553
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20928261033639e-06,
+      "loss": 1.2793,
+      "step": 3554
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.204381679333723e-06,
+      "loss": 1.3726,
+      "step": 3555
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.199480940641464e-06,
+      "loss": 1.3164,
+      "step": 3556
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194580395444184e-06,
+      "loss": 1.3613,
+      "step": 3557
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.189680044926402e-06,
+      "loss": 1.4102,
+      "step": 3558
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.184779890272593e-06,
+      "loss": 1.2051,
+      "step": 3559
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179879932667178e-06,
+      "loss": 1.4072,
+      "step": 3560
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174980173294535e-06,
+      "loss": 1.2949,
+      "step": 3561
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.170080613338999e-06,
+      "loss": 1.3271,
+      "step": 3562
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165181253984846e-06,
+      "loss": 1.1577,
+      "step": 3563
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.160282096416317e-06,
+      "loss": 1.293,
+      "step": 3564
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15538314181759e-06,
+      "loss": 1.2988,
+      "step": 3565
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150484391372804e-06,
+      "loss": 1.3398,
+      "step": 3566
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145585846266044e-06,
+      "loss": 1.2979,
+      "step": 3567
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14068750768135e-06,
+      "loss": 1.2539,
+      "step": 3568
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135789376802701e-06,
+      "loss": 1.2876,
+      "step": 3569
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.130891454814039e-06,
+      "loss": 1.3564,
+      "step": 3570
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12599374289925e-06,
+      "loss": 1.2959,
+      "step": 3571
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121096242242167e-06,
+      "loss": 1.3955,
+      "step": 3572
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116198954026577e-06,
+      "loss": 1.3389,
+      "step": 3573
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.11130187943621e-06,
+      "loss": 1.3101,
+      "step": 3574
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106405019654747e-06,
+      "loss": 1.2637,
+      "step": 3575
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10150837586582e-06,
+      "loss": 1.3789,
+      "step": 3576
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.096611949253009e-06,
+      "loss": 1.2568,
+      "step": 3577
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091715740999829e-06,
+      "loss": 1.4043,
+      "step": 3578
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.086819752289757e-06,
+      "loss": 1.4834,
+      "step": 3579
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081923984306212e-06,
+      "loss": 1.373,
+      "step": 3580
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07702843823256e-06,
+      "loss": 1.2676,
+      "step": 3581
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072133115252113e-06,
+      "loss": 1.2021,
+      "step": 3582
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.067238016548126e-06,
+      "loss": 1.4365,
+      "step": 3583
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062343143303803e-06,
+      "loss": 1.4717,
+      "step": 3584
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.057448496702294e-06,
+      "loss": 1.4419,
+      "step": 3585
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.0525540779267e-06,
+      "loss": 1.1274,
+      "step": 3586
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.047659888160047e-06,
+      "loss": 1.2651,
+      "step": 3587
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.042765928585327e-06,
+      "loss": 1.0278,
+      "step": 3588
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037872200385466e-06,
+      "loss": 1.4053,
+      "step": 3589
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.032978704743338e-06,
+      "loss": 1.2744,
+      "step": 3590
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.02808544284176e-06,
+      "loss": 1.2695,
+      "step": 3591
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.023192415863487e-06,
+      "loss": 1.3936,
+      "step": 3592
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.018299624991226e-06,
+      "loss": 1.2734,
+      "step": 3593
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01340707140762e-06,
+      "loss": 1.0879,
+      "step": 3594
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008514756295263e-06,
+      "loss": 1.2876,
+      "step": 3595
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003622680836679e-06,
+      "loss": 1.418,
+      "step": 3596
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.998730846214342e-06,
+      "loss": 1.4253,
+      "step": 3597
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993839253610669e-06,
+      "loss": 1.2139,
+      "step": 3598
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.988947904208013e-06,
+      "loss": 1.5293,
+      "step": 3599
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984056799188678e-06,
+      "loss": 1.3218,
+      "step": 3600
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.979165939734894e-06,
+      "loss": 1.4004,
+      "step": 3601
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974275327028843e-06,
+      "loss": 1.1343,
+      "step": 3602
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.969384962252646e-06,
+      "loss": 1.4062,
+      "step": 3603
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96449484658836e-06,
+      "loss": 1.3975,
+      "step": 3604
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959604981217991e-06,
+      "loss": 1.3257,
+      "step": 3605
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 1.3086,
+      "step": 3606
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949826006086672e-06,
+      "loss": 1.4697,
+      "step": 3607
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.944936898689419e-06,
+      "loss": 1.3271,
+      "step": 3608
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94004804631347e-06,
+      "loss": 1.3486,
+      "step": 3609
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935159450140511e-06,
+      "loss": 1.4365,
+      "step": 3610
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930271111352178e-06,
+      "loss": 1.293,
+      "step": 3611
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92538303113004e-06,
+      "loss": 1.3135,
+      "step": 3612
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.920495210655605e-06,
+      "loss": 1.2744,
+      "step": 3613
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915607651110319e-06,
+      "loss": 1.3496,
+      "step": 3614
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.910720353675558e-06,
+      "loss": 1.3589,
+      "step": 3615
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905833319532643e-06,
+      "loss": 1.1763,
+      "step": 3616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.900946549862828e-06,
+      "loss": 1.3276,
+      "step": 3617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.896060045847305e-06,
+      "loss": 1.2124,
+      "step": 3618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.891173808667196e-06,
+      "loss": 1.4517,
+      "step": 3619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886287839503566e-06,
+      "loss": 1.4014,
+      "step": 3620
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88140213953741e-06,
+      "loss": 1.2988,
+      "step": 3621
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876516709949661e-06,
+      "loss": 1.3506,
+      "step": 3622
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.871631551921191e-06,
+      "loss": 1.2349,
+      "step": 3623
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.86674666663279e-06,
+      "loss": 1.4775,
+      "step": 3624
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861862055265197e-06,
+      "loss": 1.4736,
+      "step": 3625
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856977718999083e-06,
+      "loss": 1.2988,
+      "step": 3626
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85209365901505e-06,
+      "loss": 1.1475,
+      "step": 3627
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.84720987649363e-06,
+      "loss": 1.5137,
+      "step": 3628
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842326372615294e-06,
+      "loss": 1.3672,
+      "step": 3629
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83744314856044e-06,
+      "loss": 1.2998,
+      "step": 3630
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832560205509407e-06,
+      "loss": 1.2788,
+      "step": 3631
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827677544642458e-06,
+      "loss": 1.2432,
+      "step": 3632
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.822795167139788e-06,
+      "loss": 1.3047,
+      "step": 3633
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817913074181527e-06,
+      "loss": 1.3789,
+      "step": 3634
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813031266947734e-06,
+      "loss": 1.4307,
+      "step": 3635
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.808149746618403e-06,
+      "loss": 1.0024,
+      "step": 3636
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.803268514373451e-06,
+      "loss": 1.3213,
+      "step": 3637
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798387571392732e-06,
+      "loss": 1.4834,
+      "step": 3638
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79350691885603e-06,
+      "loss": 1.246,
+      "step": 3639
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788626557943054e-06,
+      "loss": 1.3203,
+      "step": 3640
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78374648983345e-06,
+      "loss": 1.2153,
+      "step": 3641
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778866715706782e-06,
+      "loss": 1.2085,
+      "step": 3642
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.77398723674255e-06,
+      "loss": 1.3691,
+      "step": 3643
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769108054120185e-06,
+      "loss": 1.3652,
+      "step": 3644
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.764229169019047e-06,
+      "loss": 1.3301,
+      "step": 3645
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759350582618412e-06,
+      "loss": 1.4043,
+      "step": 3646
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.754472296097498e-06,
+      "loss": 1.332,
+      "step": 3647
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.749594310635442e-06,
+      "loss": 1.2725,
+      "step": 3648
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744716627411316e-06,
+      "loss": 1.335,
+      "step": 3649
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.739839247604111e-06,
+      "loss": 1.2202,
+      "step": 3650
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734962172392746e-06,
+      "loss": 1.2583,
+      "step": 3651
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730085402956067e-06,
+      "loss": 1.2671,
+      "step": 3652
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725208940472852e-06,
+      "loss": 1.3359,
+      "step": 3653
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.7203327861218e-06,
+      "loss": 1.2368,
+      "step": 3654
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715456941081529e-06,
+      "loss": 1.4219,
+      "step": 3655
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.710581406530594e-06,
+      "loss": 1.4521,
+      "step": 3656
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705706183647468e-06,
+      "loss": 1.2627,
+      "step": 3657
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700831273610552e-06,
+      "loss": 1.3481,
+      "step": 3658
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695956677598173e-06,
+      "loss": 1.4028,
+      "step": 3659
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.691082396788569e-06,
+      "loss": 1.4199,
+      "step": 3660
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686208432359918e-06,
+      "loss": 1.335,
+      "step": 3661
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.681334785490314e-06,
+      "loss": 1.3584,
+      "step": 3662
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.676461457357777e-06,
+      "loss": 1.1909,
+      "step": 3663
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.671588449140248e-06,
+      "loss": 1.3281,
+      "step": 3664
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.666715762015587e-06,
+      "loss": 1.3228,
+      "step": 3665
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661843397161586e-06,
+      "loss": 1.3027,
+      "step": 3666
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65697135575595e-06,
+      "loss": 1.3066,
+      "step": 3667
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.652099638976314e-06,
+      "loss": 1.3062,
+      "step": 3668
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.647228248000223e-06,
+      "loss": 1.3296,
+      "step": 3669
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642357184005153e-06,
+      "loss": 1.1978,
+      "step": 3670
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6374864481685e-06,
+      "loss": 1.2979,
+      "step": 3671
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632616041667577e-06,
+      "loss": 1.4717,
+      "step": 3672
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.62774596567962e-06,
+      "loss": 1.3008,
+      "step": 3673
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.622876221381782e-06,
+      "loss": 1.3633,
+      "step": 3674
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.618006809951142e-06,
+      "loss": 1.1958,
+      "step": 3675
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.613137732564691e-06,
+      "loss": 1.2671,
+      "step": 3676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60826899039935e-06,
+      "loss": 1.2466,
+      "step": 3677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60340058463194e-06,
+      "loss": 1.2676,
+      "step": 3678
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.59853251643922e-06,
+      "loss": 1.3193,
+      "step": 3679
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.593664786997858e-06,
+      "loss": 1.4463,
+      "step": 3680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588797397484446e-06,
+      "loss": 1.4209,
+      "step": 3681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58393034907549e-06,
+      "loss": 1.2646,
+      "step": 3682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.579063642947404e-06,
+      "loss": 1.4268,
+      "step": 3683
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574197280276538e-06,
+      "loss": 1.3574,
+      "step": 3684
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.569331262239145e-06,
+      "loss": 1.4385,
+      "step": 3685
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.564465590011408e-06,
+      "loss": 1.4824,
+      "step": 3686
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.559600264769406e-06,
+      "loss": 1.2417,
+      "step": 3687
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.554735287689149e-06,
+      "loss": 1.3018,
+      "step": 3688
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.549870659946565e-06,
+      "loss": 1.2676,
+      "step": 3689
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.545006382717487e-06,
+      "loss": 1.3506,
+      "step": 3690
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540142457177678e-06,
+      "loss": 1.3281,
+      "step": 3691
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535278884502794e-06,
+      "loss": 1.2866,
+      "step": 3692
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.530415665868424e-06,
+      "loss": 1.3569,
+      "step": 3693
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.525552802450067e-06,
+      "loss": 1.3447,
+      "step": 3694
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.520690295423136e-06,
+      "loss": 1.3315,
+      "step": 3695
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515828145962953e-06,
+      "loss": 1.0957,
+      "step": 3696
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.510966355244756e-06,
+      "loss": 1.2949,
+      "step": 3697
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506104924443704e-06,
+      "loss": 1.3066,
+      "step": 3698
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501243854734857e-06,
+      "loss": 1.2744,
+      "step": 3699
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496383147293201e-06,
+      "loss": 1.1597,
+      "step": 3700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.491522803293618e-06,
+      "loss": 1.3906,
+      "step": 3701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486662823910912e-06,
+      "loss": 1.2744,
+      "step": 3702
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.481803210319801e-06,
+      "loss": 1.4658,
+      "step": 3703
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.476943963694913e-06,
+      "loss": 1.2734,
+      "step": 3704
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472085085210781e-06,
+      "loss": 1.312,
+      "step": 3705
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467226576041856e-06,
+      "loss": 1.2202,
+      "step": 3706
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.462368437362497e-06,
+      "loss": 1.2012,
+      "step": 3707
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.457510670346976e-06,
+      "loss": 1.2349,
+      "step": 3708
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.452653276169474e-06,
+      "loss": 1.2979,
+      "step": 3709
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.447796256004075e-06,
+      "loss": 1.0947,
+      "step": 3710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442939611024782e-06,
+      "loss": 1.2852,
+      "step": 3711
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.438083342405503e-06,
+      "loss": 1.2485,
+      "step": 3712
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.433227451320059e-06,
+      "loss": 1.1836,
+      "step": 3713
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428371938942173e-06,
+      "loss": 1.25,
+      "step": 3714
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42351680644548e-06,
+      "loss": 1.2637,
+      "step": 3715
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.418662055003527e-06,
+      "loss": 1.2808,
+      "step": 3716
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41380768578976e-06,
+      "loss": 1.3071,
+      "step": 3717
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.408953699977546e-06,
+      "loss": 1.3584,
+      "step": 3718
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.404100098740141e-06,
+      "loss": 1.3599,
+      "step": 3719
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.399246883250724e-06,
+      "loss": 1.4609,
+      "step": 3720
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394394054682373e-06,
+      "loss": 1.3652,
+      "step": 3721
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.389541614208075e-06,
+      "loss": 1.2876,
+      "step": 3722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.384689563000724e-06,
+      "loss": 1.2617,
+      "step": 3723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.379837902233116e-06,
+      "loss": 1.1851,
+      "step": 3724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.374986633077955e-06,
+      "loss": 1.3564,
+      "step": 3725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370135756707853e-06,
+      "loss": 1.2104,
+      "step": 3726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.365285274295326e-06,
+      "loss": 1.4312,
+      "step": 3727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.360435187012789e-06,
+      "loss": 1.2588,
+      "step": 3728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355585496032566e-06,
+      "loss": 1.3735,
+      "step": 3729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.350736202526885e-06,
+      "loss": 1.2358,
+      "step": 3730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.345887307667881e-06,
+      "loss": 1.4077,
+      "step": 3731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341038812627588e-06,
+      "loss": 1.3535,
+      "step": 3732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.336190718577945e-06,
+      "loss": 1.3047,
+      "step": 3733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331343026690791e-06,
+      "loss": 1.4043,
+      "step": 3734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326495738137875e-06,
+      "loss": 1.3774,
+      "step": 3735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.321648854090847e-06,
+      "loss": 1.2603,
+      "step": 3736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316802375721247e-06,
+      "loss": 1.3955,
+      "step": 3737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.311956304200532e-06,
+      "loss": 1.293,
+      "step": 3738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307110640700056e-06,
+      "loss": 1.3696,
+      "step": 3739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.302265386391071e-06,
+      "loss": 1.146,
+      "step": 3740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297420542444737e-06,
+      "loss": 1.2349,
+      "step": 3741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292576110032106e-06,
+      "loss": 1.2959,
+      "step": 3742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287732090324134e-06,
+      "loss": 1.2939,
+      "step": 3743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.282888484491682e-06,
+      "loss": 1.2024,
+      "step": 3744
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27804529370551e-06,
+      "loss": 1.333,
+      "step": 3745
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.273202519136266e-06,
+      "loss": 1.3604,
+      "step": 3746
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.26836016195451e-06,
+      "loss": 1.4238,
+      "step": 3747
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 1.4717,
+      "step": 3748
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258676704435183e-06,
+      "loss": 1.1709,
+      "step": 3749
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.253835606438224e-06,
+      "loss": 1.3374,
+      "step": 3750
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248994930509962e-06,
+      "loss": 1.3428,
+      "step": 3751
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.244154677820451e-06,
+      "loss": 1.2754,
+      "step": 3752
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239314849539639e-06,
+      "loss": 1.1562,
+      "step": 3753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23447544683737e-06,
+      "loss": 1.3218,
+      "step": 3754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229636470883381e-06,
+      "loss": 1.3823,
+      "step": 3755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.224797922847312e-06,
+      "loss": 1.4521,
+      "step": 3756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.219959803898698e-06,
+      "loss": 1.4248,
+      "step": 3757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.215122115206971e-06,
+      "loss": 1.3047,
+      "step": 3758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210284857941459e-06,
+      "loss": 1.3525,
+      "step": 3759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.205448033271382e-06,
+      "loss": 1.3516,
+      "step": 3760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200611642365856e-06,
+      "loss": 1.1616,
+      "step": 3761
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.195775686393898e-06,
+      "loss": 1.2344,
+      "step": 3762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190940166524418e-06,
+      "loss": 1.3975,
+      "step": 3763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.186105083926213e-06,
+      "loss": 1.1553,
+      "step": 3764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.181270439767982e-06,
+      "loss": 1.3486,
+      "step": 3765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.176436235218317e-06,
+      "loss": 1.4434,
+      "step": 3766
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171602471445703e-06,
+      "loss": 1.1948,
+      "step": 3767
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.166769149618517e-06,
+      "loss": 1.2349,
+      "step": 3768
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16193627090503e-06,
+      "loss": 1.1548,
+      "step": 3769
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.157103836473408e-06,
+      "loss": 1.3682,
+      "step": 3770
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152271847491706e-06,
+      "loss": 1.2246,
+      "step": 3771
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14744030512788e-06,
+      "loss": 1.3223,
+      "step": 3772
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.14260921054976e-06,
+      "loss": 1.4258,
+      "step": 3773
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.137778564925085e-06,
+      "loss": 1.2642,
+      "step": 3774
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132948369421478e-06,
+      "loss": 1.2954,
+      "step": 3775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128118625206455e-06,
+      "loss": 1.313,
+      "step": 3776
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123289333447429e-06,
+      "loss": 1.2495,
+      "step": 3777
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.118460495311687e-06,
+      "loss": 1.3096,
+      "step": 3778
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113632111966421e-06,
+      "loss": 1.3242,
+      "step": 3779
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.108804184578708e-06,
+      "loss": 1.2573,
+      "step": 3780
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.103976714315517e-06,
+      "loss": 1.4385,
+      "step": 3781
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09914970234371e-06,
+      "loss": 1.2485,
+      "step": 3782
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.094323149830024e-06,
+      "loss": 1.5342,
+      "step": 3783
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089497057941095e-06,
+      "loss": 1.1616,
+      "step": 3784
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.084671427843451e-06,
+      "loss": 1.3984,
+      "step": 3785
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.079846260703503e-06,
+      "loss": 1.4316,
+      "step": 3786
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075021557687552e-06,
+      "loss": 1.4521,
+      "step": 3787
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.070197319961783e-06,
+      "loss": 1.2559,
+      "step": 3788
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065373548692272e-06,
+      "loss": 1.1431,
+      "step": 3789
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.060550245044987e-06,
+      "loss": 1.3335,
+      "step": 3790
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055727410185776e-06,
+      "loss": 1.2749,
+      "step": 3791
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05090504528037e-06,
+      "loss": 1.3643,
+      "step": 3792
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046083151494397e-06,
+      "loss": 1.1768,
+      "step": 3793
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041261729993365e-06,
+      "loss": 1.1284,
+      "step": 3794
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036440781942671e-06,
+      "loss": 1.3936,
+      "step": 3795
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.031620308507591e-06,
+      "loss": 1.3945,
+      "step": 3796
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026800310853293e-06,
+      "loss": 1.3652,
+      "step": 3797
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.021980790144828e-06,
+      "loss": 1.2822,
+      "step": 3798
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01716174754713e-06,
+      "loss": 1.3818,
+      "step": 3799
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012343184225026e-06,
+      "loss": 1.2749,
+      "step": 3800
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00752510134321e-06,
+      "loss": 1.333,
+      "step": 3801
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002707500066276e-06,
+      "loss": 1.2432,
+      "step": 3802
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997890381558691e-06,
+      "loss": 1.2241,
+      "step": 3803
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993073746984816e-06,
+      "loss": 1.4482,
+      "step": 3804
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988257597508884e-06,
+      "loss": 1.2739,
+      "step": 3805
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983441934295016e-06,
+      "loss": 1.333,
+      "step": 3806
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978626758507218e-06,
+      "loss": 1.4971,
+      "step": 3807
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.973812071309372e-06,
+      "loss": 1.1665,
+      "step": 3808
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.96899787386525e-06,
+      "loss": 1.4336,
+      "step": 3809
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964184167338494e-06,
+      "loss": 1.4287,
+      "step": 3810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.959370952892638e-06,
+      "loss": 1.293,
+      "step": 3811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954558231691092e-06,
+      "loss": 1.3311,
+      "step": 3812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94974600489715e-06,
+      "loss": 1.3979,
+      "step": 3813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.944934273673982e-06,
+      "loss": 1.1084,
+      "step": 3814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940123039184644e-06,
+      "loss": 1.3389,
+      "step": 3815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.935312302592062e-06,
+      "loss": 1.4482,
+      "step": 3816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930502065059056e-06,
+      "loss": 1.1914,
+      "step": 3817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925692327748317e-06,
+      "loss": 1.3379,
+      "step": 3818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 1.3066,
+      "step": 3819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.916074358443787e-06,
+      "loss": 1.3945,
+      "step": 3820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.911266128774775e-06,
+      "loss": 1.374,
+      "step": 3821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.906458403977585e-06,
+      "loss": 1.291,
+      "step": 3822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.901651185214295e-06,
+      "loss": 1.0828,
+      "step": 3823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.896844473646873e-06,
+      "loss": 1.2935,
+      "step": 3824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892038270437153e-06,
+      "loss": 1.3623,
+      "step": 3825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.887232576746852e-06,
+      "loss": 1.249,
+      "step": 3826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882427393737573e-06,
+      "loss": 1.3174,
+      "step": 3827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.877622722570772e-06,
+      "loss": 1.3525,
+      "step": 3828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.8728185644078e-06,
+      "loss": 1.3389,
+      "step": 3829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868014920409882e-06,
+      "loss": 1.2705,
+      "step": 3830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.863211791738118e-06,
+      "loss": 1.248,
+      "step": 3831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.858409179553474e-06,
+      "loss": 1.2378,
+      "step": 3832
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853607085016802e-06,
+      "loss": 1.2466,
+      "step": 3833
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.848805509288824e-06,
+      "loss": 1.4258,
+      "step": 3834
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.844004453530142e-06,
+      "loss": 1.1797,
+      "step": 3835
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839203918901229e-06,
+      "loss": 1.3887,
+      "step": 3836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83440390656242e-06,
+      "loss": 1.5273,
+      "step": 3837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.829604417673947e-06,
+      "loss": 1.5117,
+      "step": 3838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.8248054533959e-06,
+      "loss": 1.3486,
+      "step": 3839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.820007014888243e-06,
+      "loss": 1.3262,
+      "step": 3840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.815209103310825e-06,
+      "loss": 1.1294,
+      "step": 3841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810411719823346e-06,
+      "loss": 1.4233,
+      "step": 3842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.805614865585397e-06,
+      "loss": 1.3076,
+      "step": 3843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800818541756431e-06,
+      "loss": 1.3789,
+      "step": 3844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796022749495783e-06,
+      "loss": 1.375,
+      "step": 3845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.791227489962645e-06,
+      "loss": 1.2451,
+      "step": 3846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786432764316097e-06,
+      "loss": 1.2715,
+      "step": 3847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.781638573715074e-06,
+      "loss": 1.2383,
+      "step": 3848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77684491931839e-06,
+      "loss": 1.248,
+      "step": 3849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.772051802284738e-06,
+      "loss": 1.2261,
+      "step": 3850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.767259223772656e-06,
+      "loss": 1.4561,
+      "step": 3851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.762467184940574e-06,
+      "loss": 1.4854,
+      "step": 3852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757675686946787e-06,
+      "loss": 1.1675,
+      "step": 3853
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.752884730949456e-06,
+      "loss": 1.2324,
+      "step": 3854
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.748094318106611e-06,
+      "loss": 1.2283,
+      "step": 3855
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74330444957615e-06,
+      "loss": 1.3501,
+      "step": 3856
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.738515126515846e-06,
+      "loss": 1.1606,
+      "step": 3857
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.733726350083331e-06,
+      "loss": 1.3652,
+      "step": 3858
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728938121436116e-06,
+      "loss": 1.2559,
+      "step": 3859
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.724150441731563e-06,
+      "loss": 1.2603,
+      "step": 3860
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.719363312126914e-06,
+      "loss": 1.375,
+      "step": 3861
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71457673377928e-06,
+      "loss": 1.2432,
+      "step": 3862
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.709790707845632e-06,
+      "loss": 1.3599,
+      "step": 3863
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.705005235482805e-06,
+      "loss": 1.2388,
+      "step": 3864
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700220317847508e-06,
+      "loss": 1.3628,
+      "step": 3865
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.695435956096313e-06,
+      "loss": 1.2881,
+      "step": 3866
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.690652151385658e-06,
+      "loss": 1.4141,
+      "step": 3867
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685868904871846e-06,
+      "loss": 1.1206,
+      "step": 3868
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681086217711039e-06,
+      "loss": 1.1978,
+      "step": 3869
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676304091059273e-06,
+      "loss": 1.1611,
+      "step": 3870
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.671522526072446e-06,
+      "loss": 1.3809,
+      "step": 3871
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.666741523906318e-06,
+      "loss": 1.4028,
+      "step": 3872
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661961085716513e-06,
+      "loss": 1.3828,
+      "step": 3873
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65718121265852e-06,
+      "loss": 1.3228,
+      "step": 3874
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.652401905887692e-06,
+      "loss": 1.3691,
+      "step": 3875
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.647623166559247e-06,
+      "loss": 1.3984,
+      "step": 3876
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642844995828262e-06,
+      "loss": 1.2993,
+      "step": 3877
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638067394849672e-06,
+      "loss": 1.457,
+      "step": 3878
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.633290364778284e-06,
+      "loss": 1.2881,
+      "step": 3879
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628513906768764e-06,
+      "loss": 1.3594,
+      "step": 3880
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623738021975638e-06,
+      "loss": 1.333,
+      "step": 3881
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.618962711553294e-06,
+      "loss": 1.1755,
+      "step": 3882
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.61418797665598e-06,
+      "loss": 1.458,
+      "step": 3883
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.609413818437807e-06,
+      "loss": 1.3789,
+      "step": 3884
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.604640238052749e-06,
+      "loss": 1.3477,
+      "step": 3885
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.599867236654635e-06,
+      "loss": 1.2983,
+      "step": 3886
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.595094815397154e-06,
+      "loss": 1.3389,
+      "step": 3887
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.590322975433857e-06,
+      "loss": 1.3774,
+      "step": 3888
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585551717918157e-06,
+      "loss": 1.3027,
+      "step": 3889
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.580781044003324e-06,
+      "loss": 1.252,
+      "step": 3890
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.576010954842486e-06,
+      "loss": 1.3613,
+      "step": 3891
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571241451588627e-06,
+      "loss": 1.2051,
+      "step": 3892
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.566472535394596e-06,
+      "loss": 1.2402,
+      "step": 3893
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561704207413095e-06,
+      "loss": 1.2671,
+      "step": 3894
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.556936468796693e-06,
+      "loss": 1.1567,
+      "step": 3895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5521693206977976e-06,
+      "loss": 1.2329,
+      "step": 3896
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54740276426869e-06,
+      "loss": 1.2393,
+      "step": 3897
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.542636800661505e-06,
+      "loss": 1.3467,
+      "step": 3898
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.53787143102823e-06,
+      "loss": 1.2681,
+      "step": 3899
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.533106656520716e-06,
+      "loss": 1.2646,
+      "step": 3900
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528342478290661e-06,
+      "loss": 1.4375,
+      "step": 3901
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.523578897489626e-06,
+      "loss": 1.1001,
+      "step": 3902
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.518815915269025e-06,
+      "loss": 1.3403,
+      "step": 3903
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5140535327801285e-06,
+      "loss": 1.4473,
+      "step": 3904
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.509291751174057e-06,
+      "loss": 1.3979,
+      "step": 3905
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504530571601792e-06,
+      "loss": 1.1821,
+      "step": 3906
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4997699952141675e-06,
+      "loss": 1.1777,
+      "step": 3907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.495010023161873e-06,
+      "loss": 1.5381,
+      "step": 3908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.490250656595448e-06,
+      "loss": 1.3872,
+      "step": 3909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.485491896665289e-06,
+      "loss": 1.0647,
+      "step": 3910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4807337445216425e-06,
+      "loss": 1.4521,
+      "step": 3911
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.475976201314614e-06,
+      "loss": 1.293,
+      "step": 3912
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471219268194161e-06,
+      "loss": 1.4375,
+      "step": 3913
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.466462946310082e-06,
+      "loss": 1.3271,
+      "step": 3914
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461707236812042e-06,
+      "loss": 1.2715,
+      "step": 3915
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.456952140849551e-06,
+      "loss": 1.1895,
+      "step": 3916
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.452197659571972e-06,
+      "loss": 1.3682,
+      "step": 3917
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.447443794128525e-06,
+      "loss": 1.5049,
+      "step": 3918
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442690545668267e-06,
+      "loss": 1.2417,
+      "step": 3919
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.437937915340121e-06,
+      "loss": 1.4023,
+      "step": 3920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.433185904292852e-06,
+      "loss": 1.4424,
+      "step": 3921
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.428434513675084e-06,
+      "loss": 1.5156,
+      "step": 3922
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423683744635273e-06,
+      "loss": 1.4707,
+      "step": 3923
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418933598321741e-06,
+      "loss": 1.2368,
+      "step": 3924
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.414184075882657e-06,
+      "loss": 1.4844,
+      "step": 3925
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.409435178466036e-06,
+      "loss": 1.3701,
+      "step": 3926
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4046869072197446e-06,
+      "loss": 1.4326,
+      "step": 3927
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3999392632914936e-06,
+      "loss": 1.2646,
+      "step": 3928
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.395192247828845e-06,
+      "loss": 1.3462,
+      "step": 3929
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3904458619792106e-06,
+      "loss": 1.3213,
+      "step": 3930
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.385700106889853e-06,
+      "loss": 1.2554,
+      "step": 3931
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380954983707866e-06,
+      "loss": 1.3027,
+      "step": 3932
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.376210493580211e-06,
+      "loss": 1.2441,
+      "step": 3933
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.371466637653685e-06,
+      "loss": 1.46,
+      "step": 3934
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.366723417074934e-06,
+      "loss": 1.2295,
+      "step": 3935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361980832990454e-06,
+      "loss": 1.3682,
+      "step": 3936
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3572388865465805e-06,
+      "loss": 1.2734,
+      "step": 3937
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3524975788895e-06,
+      "loss": 1.3018,
+      "step": 3938
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.34775691116524e-06,
+      "loss": 1.2563,
+      "step": 3939
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343016884519685e-06,
+      "loss": 1.3003,
+      "step": 3940
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338277500098545e-06,
+      "loss": 1.4512,
+      "step": 3941
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.33353875904739e-06,
+      "loss": 1.4043,
+      "step": 3942
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.328800662511629e-06,
+      "loss": 1.1733,
+      "step": 3943
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3240632116365185e-06,
+      "loss": 1.3618,
+      "step": 3944
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.319326407567158e-06,
+      "loss": 1.2124,
+      "step": 3945
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.314590251448486e-06,
+      "loss": 1.188,
+      "step": 3946
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309854744425286e-06,
+      "loss": 1.2974,
+      "step": 3947
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305119887642191e-06,
+      "loss": 1.2612,
+      "step": 3948
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300385682243676e-06,
+      "loss": 1.3105,
+      "step": 3949
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295652129374043e-06,
+      "loss": 1.2764,
+      "step": 3950
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290919230177455e-06,
+      "loss": 1.3447,
+      "step": 3951
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.28618698579791e-06,
+      "loss": 1.3115,
+      "step": 3952
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.281455397379244e-06,
+      "loss": 1.2778,
+      "step": 3953
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276724466065146e-06,
+      "loss": 1.3032,
+      "step": 3954
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27199419299913e-06,
+      "loss": 1.4414,
+      "step": 3955
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.267264579324565e-06,
+      "loss": 1.2671,
+      "step": 3956
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26253562618465e-06,
+      "loss": 1.2061,
+      "step": 3957
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2578073347224335e-06,
+      "loss": 1.3931,
+      "step": 3958
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.253079706080803e-06,
+      "loss": 1.3555,
+      "step": 3959
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2483527414024714e-06,
+      "loss": 1.2368,
+      "step": 3960
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.243626441830009e-06,
+      "loss": 1.2935,
+      "step": 3961
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.23890080850582e-06,
+      "loss": 1.4307,
+      "step": 3962
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234175842572146e-06,
+      "loss": 1.2251,
+      "step": 3963
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.229451545171062e-06,
+      "loss": 1.2788,
+      "step": 3964
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224727917444493e-06,
+      "loss": 1.1099,
+      "step": 3965
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220004960534193e-06,
+      "loss": 1.3398,
+      "step": 3966
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.215282675581758e-06,
+      "loss": 1.2783,
+      "step": 3967
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.210561063728623e-06,
+      "loss": 1.3901,
+      "step": 3968
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20584012611605e-06,
+      "loss": 1.269,
+      "step": 3969
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201119863885153e-06,
+      "loss": 1.3374,
+      "step": 3970
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1964002781768696e-06,
+      "loss": 1.3159,
+      "step": 3971
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.191681370131991e-06,
+      "loss": 1.1968,
+      "step": 3972
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186963140891119e-06,
+      "loss": 1.4199,
+      "step": 3973
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.182245591594714e-06,
+      "loss": 1.4072,
+      "step": 3974
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.177528723383063e-06,
+      "loss": 1.3105,
+      "step": 3975
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.172812537396286e-06,
+      "loss": 1.2637,
+      "step": 3976
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.168097034774349e-06,
+      "loss": 1.2617,
+      "step": 3977
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163382216657033e-06,
+      "loss": 1.334,
+      "step": 3978
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.158668084183975e-06,
+      "loss": 1.2856,
+      "step": 3979
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153954638494634e-06,
+      "loss": 1.2759,
+      "step": 3980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.149241880728311e-06,
+      "loss": 1.2627,
+      "step": 3981
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1445298120241244e-06,
+      "loss": 1.3965,
+      "step": 3982
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139818433521044e-06,
+      "loss": 1.1045,
+      "step": 3983
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135107746357867e-06,
+      "loss": 1.4019,
+      "step": 3984
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.13039775167322e-06,
+      "loss": 1.2788,
+      "step": 3985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.125688450605568e-06,
+      "loss": 1.3701,
+      "step": 3986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.120979844293201e-06,
+      "loss": 1.3047,
+      "step": 3987
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.116271933874246e-06,
+      "loss": 1.3271,
+      "step": 3988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1115647204866625e-06,
+      "loss": 1.4434,
+      "step": 3989
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.106858205268243e-06,
+      "loss": 1.4092,
+      "step": 3990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1021523893566e-06,
+      "loss": 1.1885,
+      "step": 3991
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0974472738891905e-06,
+      "loss": 1.3042,
+      "step": 3992
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0927428600032966e-06,
+      "loss": 1.4316,
+      "step": 3993
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.08803914883603e-06,
+      "loss": 1.3301,
+      "step": 3994
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083336141524337e-06,
+      "loss": 1.2817,
+      "step": 3995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.078633839204984e-06,
+      "loss": 1.2832,
+      "step": 3996
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0739322430145795e-06,
+      "loss": 1.3428,
+      "step": 3997
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069231354089552e-06,
+      "loss": 1.2778,
+      "step": 3998
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.064531173566166e-06,
+      "loss": 1.3545,
+      "step": 3999
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0598317025805045e-06,
+      "loss": 1.4736,
+      "step": 4000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.055132942268488e-06,
+      "loss": 1.2197,
+      "step": 4001
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.050434893765865e-06,
+      "loss": 1.3311,
+      "step": 4002
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.045737558208208e-06,
+      "loss": 1.2593,
+      "step": 4003
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041040936730923e-06,
+      "loss": 1.2339,
+      "step": 4004
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036345030469232e-06,
+      "loss": 1.3862,
+      "step": 4005
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.031649840558195e-06,
+      "loss": 1.2163,
+      "step": 4006
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026955368132695e-06,
+      "loss": 1.3008,
+      "step": 4007
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022261614327448e-06,
+      "loss": 1.2803,
+      "step": 4008
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.017568580276979e-06,
+      "loss": 1.3779,
+      "step": 4009
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0128762671156545e-06,
+      "loss": 1.1802,
+      "step": 4010
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.008184675977665e-06,
+      "loss": 1.335,
+      "step": 4011
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003493807997021e-06,
+      "loss": 1.5078,
+      "step": 4012
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998803664307565e-06,
+      "loss": 1.3486,
+      "step": 4013
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9941142460429555e-06,
+      "loss": 1.3174,
+      "step": 4014
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.989425554336682e-06,
+      "loss": 1.3135,
+      "step": 4015
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.984737590322059e-06,
+      "loss": 1.4209,
+      "step": 4016
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.980050355132222e-06,
+      "loss": 1.3496,
+      "step": 4017
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975363849900135e-06,
+      "loss": 1.1929,
+      "step": 4018
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970678075758576e-06,
+      "loss": 1.3076,
+      "step": 4019
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965993033840154e-06,
+      "loss": 1.4136,
+      "step": 4020
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.961308725277301e-06,
+      "loss": 1.2031,
+      "step": 4021
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.95662515120227e-06,
+      "loss": 1.3999,
+      "step": 4022
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.951942312747135e-06,
+      "loss": 1.2305,
+      "step": 4023
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947260211043795e-06,
+      "loss": 1.2676,
+      "step": 4024
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9425788472239685e-06,
+      "loss": 1.2275,
+      "step": 4025
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.937898222419199e-06,
+      "loss": 1.3027,
+      "step": 4026
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.933218337760852e-06,
+      "loss": 1.2695,
+      "step": 4027
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928539194380101e-06,
+      "loss": 1.2207,
+      "step": 4028
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.923860793407959e-06,
+      "loss": 1.2329,
+      "step": 4029
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.919183135975249e-06,
+      "loss": 1.2832,
+      "step": 4030
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914506223212618e-06,
+      "loss": 1.2656,
+      "step": 4031
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 1.2319,
+      "step": 4032
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.905154636219264e-06,
+      "loss": 1.3525,
+      "step": 4033
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900479964248935e-06,
+      "loss": 1.3232,
+      "step": 4034
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8958060414694616e-06,
+      "loss": 1.2627,
+      "step": 4035
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.891132869010592e-06,
+      "loss": 1.3628,
+      "step": 4036
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8864604480018784e-06,
+      "loss": 1.251,
+      "step": 4037
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.881788779572704e-06,
+      "loss": 1.332,
+      "step": 4038
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8771178648522695e-06,
+      "loss": 1.2617,
+      "step": 4039
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.87244770496959e-06,
+      "loss": 1.3252,
+      "step": 4040
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.867778301053496e-06,
+      "loss": 1.4541,
+      "step": 4041
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863109654232638e-06,
+      "loss": 1.3081,
+      "step": 4042
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8584417656354865e-06,
+      "loss": 1.2432,
+      "step": 4043
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853774636390322e-06,
+      "loss": 1.3877,
+      "step": 4044
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.849108267625253e-06,
+      "loss": 1.29,
+      "step": 4045
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844442660468187e-06,
+      "loss": 1.3047,
+      "step": 4046
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.839777816046859e-06,
+      "loss": 1.4482,
+      "step": 4047
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.835113735488819e-06,
+      "loss": 1.3154,
+      "step": 4048
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830450419921432e-06,
+      "loss": 1.2964,
+      "step": 4049
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825787870471873e-06,
+      "loss": 1.3848,
+      "step": 4050
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8211260882671385e-06,
+      "loss": 1.2769,
+      "step": 4051
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.816465074434036e-06,
+      "loss": 1.189,
+      "step": 4052
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.811804830099186e-06,
+      "loss": 1.2876,
+      "step": 4053
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.807145356389034e-06,
+      "loss": 1.2246,
+      "step": 4054
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802486654429818e-06,
+      "loss": 1.4697,
+      "step": 4055
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797828725347608e-06,
+      "loss": 1.0996,
+      "step": 4056
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.793171570268278e-06,
+      "loss": 1.2622,
+      "step": 4057
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.788515190317523e-06,
+      "loss": 1.2393,
+      "step": 4058
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.783859586620839e-06,
+      "loss": 1.2344,
+      "step": 4059
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.779204760303544e-06,
+      "loss": 1.3701,
+      "step": 4060
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.774550712490765e-06,
+      "loss": 1.2852,
+      "step": 4061
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769897444307441e-06,
+      "loss": 1.3525,
+      "step": 4062
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.765244956878323e-06,
+      "loss": 1.3818,
+      "step": 4063
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.760593251327969e-06,
+      "loss": 1.3433,
+      "step": 4064
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.755942328780754e-06,
+      "loss": 1.2173,
+      "step": 4065
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7512921903608605e-06,
+      "loss": 1.3613,
+      "step": 4066
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.746642837192285e-06,
+      "loss": 1.2822,
+      "step": 4067
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741994270398826e-06,
+      "loss": 1.4238,
+      "step": 4068
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.737346491104102e-06,
+      "loss": 1.3857,
+      "step": 4069
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732699500431535e-06,
+      "loss": 1.4058,
+      "step": 4070
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.728053299504358e-06,
+      "loss": 1.4697,
+      "step": 4071
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723407889445616e-06,
+      "loss": 1.3979,
+      "step": 4072
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.718763271378155e-06,
+      "loss": 1.4746,
+      "step": 4073
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.714119446424634e-06,
+      "loss": 1.3174,
+      "step": 4074
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7094764157075245e-06,
+      "loss": 1.3574,
+      "step": 4075
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7048341803491e-06,
+      "loss": 1.3896,
+      "step": 4076
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.700192741471447e-06,
+      "loss": 1.3198,
+      "step": 4077
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6955521001964526e-06,
+      "loss": 1.3271,
+      "step": 4078
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6909122576458165e-06,
+      "loss": 1.2383,
+      "step": 4079
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6862732149410446e-06,
+      "loss": 1.2671,
+      "step": 4080
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6816349732034525e-06,
+      "loss": 1.3979,
+      "step": 4081
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6769975335541485e-06,
+      "loss": 1.231,
+      "step": 4082
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672360897114063e-06,
+      "loss": 1.4443,
+      "step": 4083
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667725065003925e-06,
+      "loss": 1.2778,
+      "step": 4084
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.663090038344272e-06,
+      "loss": 1.2368,
+      "step": 4085
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.658455818255445e-06,
+      "loss": 1.3281,
+      "step": 4086
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653822405857588e-06,
+      "loss": 1.3682,
+      "step": 4087
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649189802270652e-06,
+      "loss": 1.4102,
+      "step": 4088
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6445580086143945e-06,
+      "loss": 1.2998,
+      "step": 4089
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639927026008377e-06,
+      "loss": 1.1147,
+      "step": 4090
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6352968555719556e-06,
+      "loss": 1.3516,
+      "step": 4091
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630667498424304e-06,
+      "loss": 1.5146,
+      "step": 4092
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.626038955684389e-06,
+      "loss": 1.2549,
+      "step": 4093
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.621411228470986e-06,
+      "loss": 1.1816,
+      "step": 4094
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616784317902673e-06,
+      "loss": 1.2881,
+      "step": 4095
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.612158225097829e-06,
+      "loss": 1.2183,
+      "step": 4096
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.607532951174633e-06,
+      "loss": 1.4326,
+      "step": 4097
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602908497251073e-06,
+      "loss": 1.3555,
+      "step": 4098
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598284864444934e-06,
+      "loss": 1.3643,
+      "step": 4099
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.593662053873797e-06,
+      "loss": 1.2661,
+      "step": 4100
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.589040066655056e-06,
+      "loss": 1.3193,
+      "step": 4101
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.584418903905898e-06,
+      "loss": 1.3115,
+      "step": 4102
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 1.2915,
+      "step": 4103
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.575179056284095e-06,
+      "loss": 1.4102,
+      "step": 4104
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.57056037364483e-06,
+      "loss": 1.2803,
+      "step": 4105
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56594251994191e-06,
+      "loss": 1.25,
+      "step": 4106
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.561325496291526e-06,
+      "loss": 1.4316,
+      "step": 4107
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556709303809672e-06,
+      "loss": 1.0972,
+      "step": 4108
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552093943612125e-06,
+      "loss": 1.3135,
+      "step": 4109
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.547479416814482e-06,
+      "loss": 1.4326,
+      "step": 4110
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.542865724532123e-06,
+      "loss": 1.3691,
+      "step": 4111
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5382528678802374e-06,
+      "loss": 1.3516,
+      "step": 4112
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533640847973809e-06,
+      "loss": 1.335,
+      "step": 4113
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529029665927611e-06,
+      "loss": 1.1567,
+      "step": 4114
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.524419322856223e-06,
+      "loss": 1.3799,
+      "step": 4115
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.519809819874024e-06,
+      "loss": 1.2598,
+      "step": 4116
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515201158095184e-06,
+      "loss": 1.2524,
+      "step": 4117
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.510593338633668e-06,
+      "loss": 1.3657,
+      "step": 4118
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.505986362603242e-06,
+      "loss": 1.2778,
+      "step": 4119
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501380231117466e-06,
+      "loss": 1.2642,
+      "step": 4120
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.496774945289701e-06,
+      "loss": 1.3486,
+      "step": 4121
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4921705062331e-06,
+      "loss": 1.2896,
+      "step": 4122
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487566915060604e-06,
+      "loss": 1.2266,
+      "step": 4123
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.48296417288496e-06,
+      "loss": 1.3877,
+      "step": 4124
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.478362280818704e-06,
+      "loss": 1.3193,
+      "step": 4125
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4737612399741746e-06,
+      "loss": 1.3271,
+      "step": 4126
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.469161051463485e-06,
+      "loss": 1.3438,
+      "step": 4127
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464561716398565e-06,
+      "loss": 1.4121,
+      "step": 4128
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.459963235891127e-06,
+      "loss": 1.1968,
+      "step": 4129
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.455365611052678e-06,
+      "loss": 1.1831,
+      "step": 4130
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.450768842994523e-06,
+      "loss": 1.2754,
+      "step": 4131
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446172932827746e-06,
+      "loss": 1.1831,
+      "step": 4132
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.441577881663239e-06,
+      "loss": 1.3193,
+      "step": 4133
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436983690611679e-06,
+      "loss": 1.2051,
+      "step": 4134
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.432390360783538e-06,
+      "loss": 1.3359,
+      "step": 4135
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427797893289078e-06,
+      "loss": 1.2812,
+      "step": 4136
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4232062892383526e-06,
+      "loss": 1.4106,
+      "step": 4137
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418615549741207e-06,
+      "loss": 1.3457,
+      "step": 4138
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.414025675907278e-06,
+      "loss": 1.2773,
+      "step": 4139
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409436668845997e-06,
+      "loss": 1.46,
+      "step": 4140
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.404848529666572e-06,
+      "loss": 1.29,
+      "step": 4141
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.400261259478017e-06,
+      "loss": 1.3003,
+      "step": 4142
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395674859389127e-06,
+      "loss": 1.168,
+      "step": 4143
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.391089330508493e-06,
+      "loss": 1.2437,
+      "step": 4144
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.386504673944491e-06,
+      "loss": 1.2812,
+      "step": 4145
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381920890805284e-06,
+      "loss": 1.2549,
+      "step": 4146
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.37733798219883e-06,
+      "loss": 1.3633,
+      "step": 4147
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.372755949232871e-06,
+      "loss": 1.4727,
+      "step": 4148
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368174793014943e-06,
+      "loss": 1.1494,
+      "step": 4149
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.363594514652358e-06,
+      "loss": 1.2153,
+      "step": 4150
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359015115252227e-06,
+      "loss": 1.3389,
+      "step": 4151
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.354436595921449e-06,
+      "loss": 1.2852,
+      "step": 4152
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.349858957766701e-06,
+      "loss": 1.4297,
+      "step": 4153
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34528220189446e-06,
+      "loss": 1.3428,
+      "step": 4154
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.340706329410975e-06,
+      "loss": 1.291,
+      "step": 4155
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336131341422291e-06,
+      "loss": 1.3403,
+      "step": 4156
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.331557239034238e-06,
+      "loss": 1.2671,
+      "step": 4157
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326984023352435e-06,
+      "loss": 1.2695,
+      "step": 4158
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.322411695482273e-06,
+      "loss": 1.2412,
+      "step": 4159
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317840256528944e-06,
+      "loss": 1.2969,
+      "step": 4160
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313269707597418e-06,
+      "loss": 1.2603,
+      "step": 4161
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3087000497924534e-06,
+      "loss": 1.373,
+      "step": 4162
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.304131284218589e-06,
+      "loss": 1.4482,
+      "step": 4163
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.299563411980148e-06,
+      "loss": 1.2803,
+      "step": 4164
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29499643418124e-06,
+      "loss": 1.2598,
+      "step": 4165
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.29043035192576e-06,
+      "loss": 1.2671,
+      "step": 4166
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285865166317386e-06,
+      "loss": 1.2036,
+      "step": 4167
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281300878459573e-06,
+      "loss": 1.4023,
+      "step": 4168
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276737489455564e-06,
+      "loss": 1.2881,
+      "step": 4169
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.272175000408387e-06,
+      "loss": 1.2783,
+      "step": 4170
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.26761341242085e-06,
+      "loss": 1.3408,
+      "step": 4171
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.263052726595542e-06,
+      "loss": 1.2256,
+      "step": 4172
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258492944034833e-06,
+      "loss": 1.2988,
+      "step": 4173
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.25393406584088e-06,
+      "loss": 1.2637,
+      "step": 4174
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249376093115615e-06,
+      "loss": 1.2876,
+      "step": 4175
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2448190269607625e-06,
+      "loss": 1.458,
+      "step": 4176
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240262868477807e-06,
+      "loss": 1.2603,
+      "step": 4177
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2357076187680325e-06,
+      "loss": 1.2432,
+      "step": 4178
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231153278932497e-06,
+      "loss": 1.3291,
+      "step": 4179
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226599850072038e-06,
+      "loss": 1.1348,
+      "step": 4180
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.222047333287277e-06,
+      "loss": 1.2451,
+      "step": 4181
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.217495729678604e-06,
+      "loss": 1.3398,
+      "step": 4182
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212945040346202e-06,
+      "loss": 1.1147,
+      "step": 4183
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.208395266390024e-06,
+      "loss": 1.3145,
+      "step": 4184
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2038464089098095e-06,
+      "loss": 1.3018,
+      "step": 4185
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.199298469005061e-06,
+      "loss": 1.4248,
+      "step": 4186
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194751447775079e-06,
+      "loss": 1.1738,
+      "step": 4187
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.190205346318927e-06,
+      "loss": 1.2314,
+      "step": 4188
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.185660165735457e-06,
+      "loss": 1.3135,
+      "step": 4189
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.181115907123289e-06,
+      "loss": 1.3906,
+      "step": 4190
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.176572571580826e-06,
+      "loss": 1.3193,
+      "step": 4191
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1720301602062446e-06,
+      "loss": 1.2549,
+      "step": 4192
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167488674097501e-06,
+      "loss": 1.4834,
+      "step": 4193
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.162948114352328e-06,
+      "loss": 1.229,
+      "step": 4194
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.158408482068235e-06,
+      "loss": 1.2607,
+      "step": 4195
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153869778342496e-06,
+      "loss": 1.3809,
+      "step": 4196
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149332004272177e-06,
+      "loss": 1.4424,
+      "step": 4197
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.144795160954109e-06,
+      "loss": 1.3105,
+      "step": 4198
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.140259249484904e-06,
+      "loss": 1.2314,
+      "step": 4199
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1357242709609425e-06,
+      "loss": 1.2388,
+      "step": 4200
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.131190226478383e-06,
+      "loss": 1.1997,
+      "step": 4201
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126657117133159e-06,
+      "loss": 1.3174,
+      "step": 4202
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.122124944020978e-06,
+      "loss": 1.4512,
+      "step": 4203
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.117593708237323e-06,
+      "loss": 1.2422,
+      "step": 4204
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.11306341087744e-06,
+      "loss": 1.3936,
+      "step": 4205
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.108534053036358e-06,
+      "loss": 1.3052,
+      "step": 4206
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.104005635808878e-06,
+      "loss": 1.4307,
+      "step": 4207
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.099478160289577e-06,
+      "loss": 1.2842,
+      "step": 4208
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094951627572792e-06,
+      "loss": 1.3252,
+      "step": 4209
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.090426038752641e-06,
+      "loss": 1.272,
+      "step": 4210
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.085901394923017e-06,
+      "loss": 1.2163,
+      "step": 4211
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081377697177576e-06,
+      "loss": 1.2588,
+      "step": 4212
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.076854946609756e-06,
+      "loss": 1.2083,
+      "step": 4213
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072333144312752e-06,
+      "loss": 1.251,
+      "step": 4214
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.067812291379539e-06,
+      "loss": 1.2373,
+      "step": 4215
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.063292388902862e-06,
+      "loss": 1.4199,
+      "step": 4216
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058773437975239e-06,
+      "loss": 1.0986,
+      "step": 4217
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054255439688947e-06,
+      "loss": 0.9971,
+      "step": 4218
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.049738395136045e-06,
+      "loss": 1.3252,
+      "step": 4219
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045222305408352e-06,
+      "loss": 1.4365,
+      "step": 4220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.040707171597466e-06,
+      "loss": 1.1587,
+      "step": 4221
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0361929947947495e-06,
+      "loss": 1.3662,
+      "step": 4222
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031679776091326e-06,
+      "loss": 1.2695,
+      "step": 4223
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027167516578097e-06,
+      "loss": 1.2754,
+      "step": 4224
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.022656217345729e-06,
+      "loss": 1.2803,
+      "step": 4225
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0181458794846605e-06,
+      "loss": 1.3828,
+      "step": 4226
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.013636504085088e-06,
+      "loss": 1.251,
+      "step": 4227
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.009128092236983e-06,
+      "loss": 1.1875,
+      "step": 4228
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0046206450300835e-06,
+      "loss": 1.2886,
+      "step": 4229
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.000114163553894e-06,
+      "loss": 1.3359,
+      "step": 4230
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995608648897686e-06,
+      "loss": 1.2803,
+      "step": 4231
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9911041021504915e-06,
+      "loss": 1.2002,
+      "step": 4232
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986600524401113e-06,
+      "loss": 1.2192,
+      "step": 4233
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.982097916738121e-06,
+      "loss": 1.2583,
+      "step": 4234
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977596280249851e-06,
+      "loss": 1.4492,
+      "step": 4235
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973095616024398e-06,
+      "loss": 1.2378,
+      "step": 4236
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.968595925149629e-06,
+      "loss": 1.4209,
+      "step": 4237
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.964097208713171e-06,
+      "loss": 1.1675,
+      "step": 4238
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959599467802418e-06,
+      "loss": 1.2998,
+      "step": 4239
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955102703504533e-06,
+      "loss": 1.4922,
+      "step": 4240
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.950606916906427e-06,
+      "loss": 1.3047,
+      "step": 4241
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94611210909479e-06,
+      "loss": 1.3975,
+      "step": 4242
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9416182811560715e-06,
+      "loss": 1.3604,
+      "step": 4243
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.937125434176485e-06,
+      "loss": 1.2051,
+      "step": 4244
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 1.208,
+      "step": 4245
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.928142687438356e-06,
+      "loss": 1.291,
+      "step": 4246
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923652789851052e-06,
+      "loss": 1.251,
+      "step": 4247
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919163877565351e-06,
+      "loss": 1.3096,
+      "step": 4248
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.91467595166628e-06,
+      "loss": 1.3115,
+      "step": 4249
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910189013238614e-06,
+      "loss": 1.1538,
+      "step": 4250
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.905703063366907e-06,
+      "loss": 1.3252,
+      "step": 4251
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.901218103135464e-06,
+      "loss": 1.3574,
+      "step": 4252
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.896734133628354e-06,
+      "loss": 1.4629,
+      "step": 4253
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89225115592941e-06,
+      "loss": 1.0063,
+      "step": 4254
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887769171122211e-06,
+      "loss": 1.3223,
+      "step": 4255
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883288180290115e-06,
+      "loss": 1.3193,
+      "step": 4256
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878808184516225e-06,
+      "loss": 1.354,
+      "step": 4257
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.874329184883416e-06,
+      "loss": 1.2954,
+      "step": 4258
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869851182474308e-06,
+      "loss": 1.3003,
+      "step": 4259
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.86537417837129e-06,
+      "loss": 1.3916,
+      "step": 4260
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8608981736565066e-06,
+      "loss": 1.3584,
+      "step": 4261
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856423169411865e-06,
+      "loss": 1.3887,
+      "step": 4262
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8519491667190264e-06,
+      "loss": 1.4219,
+      "step": 4263
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847476166659407e-06,
+      "loss": 1.2158,
+      "step": 4264
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843004170314185e-06,
+      "loss": 1.082,
+      "step": 4265
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8385331787642956e-06,
+      "loss": 1.2358,
+      "step": 4266
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834063193090432e-06,
+      "loss": 1.2026,
+      "step": 4267
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.829594214373043e-06,
+      "loss": 1.457,
+      "step": 4268
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825126243692328e-06,
+      "loss": 1.4575,
+      "step": 4269
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.820659282128258e-06,
+      "loss": 1.2354,
+      "step": 4270
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.816193330760542e-06,
+      "loss": 1.3203,
+      "step": 4271
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811728390668661e-06,
+      "loss": 1.2598,
+      "step": 4272
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.807264462931841e-06,
+      "loss": 1.4121,
+      "step": 4273
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.802801548629062e-06,
+      "loss": 1.2881,
+      "step": 4274
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.798339648839073e-06,
+      "loss": 1.2754,
+      "step": 4275
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.793878764640361e-06,
+      "loss": 1.3652,
+      "step": 4276
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7894188971111776e-06,
+      "loss": 1.2607,
+      "step": 4277
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.78496004732952e-06,
+      "loss": 1.292,
+      "step": 4278
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7805022163731565e-06,
+      "loss": 1.2852,
+      "step": 4279
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776045405319584e-06,
+      "loss": 1.2969,
+      "step": 4280
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77158961524608e-06,
+      "loss": 1.2905,
+      "step": 4281
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.767134847229656e-06,
+      "loss": 1.2812,
+      "step": 4282
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.762681102347079e-06,
+      "loss": 1.4512,
+      "step": 4283
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75822838167488e-06,
+      "loss": 1.0486,
+      "step": 4284
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.75377668628933e-06,
+      "loss": 1.2671,
+      "step": 4285
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.749326017266455e-06,
+      "loss": 1.127,
+      "step": 4286
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.744876375682034e-06,
+      "loss": 1.3848,
+      "step": 4287
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740427762611604e-06,
+      "loss": 1.29,
+      "step": 4288
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735980179130442e-06,
+      "loss": 1.3477,
+      "step": 4289
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731533626313588e-06,
+      "loss": 1.1362,
+      "step": 4290
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7270881052358226e-06,
+      "loss": 1.354,
+      "step": 4291
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.72264361697168e-06,
+      "loss": 1.1865,
+      "step": 4292
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.718200162595449e-06,
+      "loss": 1.3491,
+      "step": 4293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713757743181163e-06,
+      "loss": 1.1934,
+      "step": 4294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.709316359802611e-06,
+      "loss": 1.4248,
+      "step": 4295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704876013533319e-06,
+      "loss": 1.2612,
+      "step": 4296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700436705446584e-06,
+      "loss": 1.4756,
+      "step": 4297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695998436615429e-06,
+      "loss": 1.2798,
+      "step": 4298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.691561208112645e-06,
+      "loss": 1.2002,
+      "step": 4299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687125021010758e-06,
+      "loss": 1.3252,
+      "step": 4300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.682689876382046e-06,
+      "loss": 1.2866,
+      "step": 4301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.678255775298542e-06,
+      "loss": 1.3682,
+      "step": 4302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673822718832016e-06,
+      "loss": 1.3945,
+      "step": 4303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.669390708053993e-06,
+      "loss": 1.3994,
+      "step": 4304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664959744035738e-06,
+      "loss": 1.3252,
+      "step": 4305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660529827848273e-06,
+      "loss": 1.3872,
+      "step": 4306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.656100960562359e-06,
+      "loss": 1.2422,
+      "step": 4307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651673143248509e-06,
+      "loss": 1.248,
+      "step": 4308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.647246376976975e-06,
+      "loss": 1.2183,
+      "step": 4309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.642820662817759e-06,
+      "loss": 1.3135,
+      "step": 4310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638396001840613e-06,
+      "loss": 1.1708,
+      "step": 4311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633972395115024e-06,
+      "loss": 1.1401,
+      "step": 4312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.629549843710241e-06,
+      "loss": 1.2583,
+      "step": 4313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.625128348695233e-06,
+      "loss": 1.3311,
+      "step": 4314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6207079111387404e-06,
+      "loss": 1.3013,
+      "step": 4315
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.616288532109225e-06,
+      "loss": 1.0601,
+      "step": 4316
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.611870212674913e-06,
+      "loss": 1.4453,
+      "step": 4317
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6074529539037604e-06,
+      "loss": 1.2456,
+      "step": 4318
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6030367568634666e-06,
+      "loss": 1.4541,
+      "step": 4319
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598621622621489e-06,
+      "loss": 1.3799,
+      "step": 4320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594207552245008e-06,
+      "loss": 1.2549,
+      "step": 4321
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5897945468009685e-06,
+      "loss": 1.2588,
+      "step": 4322
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.585382607356032e-06,
+      "loss": 1.3091,
+      "step": 4323
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.580971734976628e-06,
+      "loss": 1.2485,
+      "step": 4324
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.576561930728909e-06,
+      "loss": 1.2134,
+      "step": 4325
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.572153195678785e-06,
+      "loss": 1.5029,
+      "step": 4326
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567745530891895e-06,
+      "loss": 1.2095,
+      "step": 4327
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.563338937433622e-06,
+      "loss": 1.2446,
+      "step": 4328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5589334163690975e-06,
+      "loss": 1.4707,
+      "step": 4329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.554528968763182e-06,
+      "loss": 1.4067,
+      "step": 4330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5501255956804955e-06,
+      "loss": 1.3369,
+      "step": 4331
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545723298185368e-06,
+      "loss": 1.3389,
+      "step": 4332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5413220773419e-06,
+      "loss": 1.208,
+      "step": 4333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53692193421391e-06,
+      "loss": 1.2324,
+      "step": 4334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532522869864974e-06,
+      "loss": 1.3369,
+      "step": 4335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528124885358393e-06,
+      "loss": 1.1792,
+      "step": 4336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.52372798175721e-06,
+      "loss": 1.2969,
+      "step": 4337
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.519332160124215e-06,
+      "loss": 1.29,
+      "step": 4338
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.514937421521925e-06,
+      "loss": 1.2573,
+      "step": 4339
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510543767012611e-06,
+      "loss": 1.3569,
+      "step": 4340
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.506151197658254e-06,
+      "loss": 1.2568,
+      "step": 4341
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.501759714520602e-06,
+      "loss": 1.2705,
+      "step": 4342
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.497369318661125e-06,
+      "loss": 1.3887,
+      "step": 4343
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.492980011141039e-06,
+      "loss": 1.1357,
+      "step": 4344
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.488591793021287e-06,
+      "loss": 1.417,
+      "step": 4345
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4842046653625515e-06,
+      "loss": 1.1328,
+      "step": 4346
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.47981862922526e-06,
+      "loss": 1.3442,
+      "step": 4347
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475433685669562e-06,
+      "loss": 1.124,
+      "step": 4348
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471049835755364e-06,
+      "loss": 1.2197,
+      "step": 4349
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466667080542276e-06,
+      "loss": 1.1335,
+      "step": 4350
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462285421089675e-06,
+      "loss": 1.3721,
+      "step": 4351
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457904858456653e-06,
+      "loss": 1.0671,
+      "step": 4352
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453525393702053e-06,
+      "loss": 1.2988,
+      "step": 4353
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.449147027884436e-06,
+      "loss": 1.1963,
+      "step": 4354
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4447697620621055e-06,
+      "loss": 1.2354,
+      "step": 4355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.440393597293103e-06,
+      "loss": 1.3789,
+      "step": 4356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436018534635195e-06,
+      "loss": 1.3125,
+      "step": 4357
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.431644575145895e-06,
+      "loss": 1.3408,
+      "step": 4358
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4272717198824255e-06,
+      "loss": 1.3076,
+      "step": 4359
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.422899969901773e-06,
+      "loss": 1.3828,
+      "step": 4360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41852932626063e-06,
+      "loss": 1.2539,
+      "step": 4361
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.41415979001544e-06,
+      "loss": 1.4019,
+      "step": 4362
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.409791362222374e-06,
+      "loss": 1.3623,
+      "step": 4363
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.405424043937322e-06,
+      "loss": 1.124,
+      "step": 4364
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.401057836215928e-06,
+      "loss": 1.2173,
+      "step": 4365
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.39669274011355e-06,
+      "loss": 1.1865,
+      "step": 4366
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.392328756685293e-06,
+      "loss": 1.1584,
+      "step": 4367
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387965886985967e-06,
+      "loss": 1.4023,
+      "step": 4368
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3836041320701425e-06,
+      "loss": 1.417,
+      "step": 4369
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.379243492992101e-06,
+      "loss": 1.3369,
+      "step": 4370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374883970805867e-06,
+      "loss": 1.249,
+      "step": 4371
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.370525566565186e-06,
+      "loss": 1.2188,
+      "step": 4372
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.366168281323533e-06,
+      "loss": 1.293,
+      "step": 4373
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361812116134122e-06,
+      "loss": 1.2798,
+      "step": 4374
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357457072049881e-06,
+      "loss": 1.335,
+      "step": 4375
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35310315012349e-06,
+      "loss": 1.2524,
+      "step": 4376
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348750351407326e-06,
+      "loss": 1.2212,
+      "step": 4377
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344398676953526e-06,
+      "loss": 1.3623,
+      "step": 4378
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.340048127813933e-06,
+      "loss": 1.2324,
+      "step": 4379
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.335698705040134e-06,
+      "loss": 1.2568,
+      "step": 4380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.331350409683433e-06,
+      "loss": 1.0942,
+      "step": 4381
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32700324279486e-06,
+      "loss": 1.4082,
+      "step": 4382
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.322657205425184e-06,
+      "loss": 1.2637,
+      "step": 4383
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.318312298624887e-06,
+      "loss": 1.2656,
+      "step": 4384
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.313968523444197e-06,
+      "loss": 1.1328,
+      "step": 4385
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.309625880933039e-06,
+      "loss": 1.1924,
+      "step": 4386
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305284372141095e-06,
+      "loss": 1.1367,
+      "step": 4387
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30094399811775e-06,
+      "loss": 1.2676,
+      "step": 4388
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.296604759912133e-06,
+      "loss": 1.3662,
+      "step": 4389
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.292266658573084e-06,
+      "loss": 1.3032,
+      "step": 4390
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287929695149171e-06,
+      "loss": 1.2109,
+      "step": 4391
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283593870688697e-06,
+      "loss": 1.3125,
+      "step": 4392
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.279259186239674e-06,
+      "loss": 1.0586,
+      "step": 4393
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274925642849859e-06,
+      "loss": 1.2188,
+      "step": 4394
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.270593241566709e-06,
+      "loss": 1.1909,
+      "step": 4395
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.266261983437421e-06,
+      "loss": 1.1611,
+      "step": 4396
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261931869508909e-06,
+      "loss": 1.3872,
+      "step": 4397
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257602900827821e-06,
+      "loss": 1.2822,
+      "step": 4398
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.253275078440514e-06,
+      "loss": 1.1172,
+      "step": 4399
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.24894840339307e-06,
+      "loss": 1.0088,
+      "step": 4400
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.244622876731309e-06,
+      "loss": 1.147,
+      "step": 4401
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240298499500751e-06,
+      "loss": 1.2466,
+      "step": 4402
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.235975272746664e-06,
+      "loss": 1.1038,
+      "step": 4403
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2316531975140065e-06,
+      "loss": 1.1987,
+      "step": 4404
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.227332274847485e-06,
+      "loss": 1.333,
+      "step": 4405
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.223012505791513e-06,
+      "loss": 1.2988,
+      "step": 4406
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218693891390232e-06,
+      "loss": 1.1226,
+      "step": 4407
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2143764326875126e-06,
+      "loss": 1.2646,
+      "step": 4408
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.210060130726918e-06,
+      "loss": 1.2456,
+      "step": 4409
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205744986551763e-06,
+      "loss": 1.2627,
+      "step": 4410
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20143100120506e-06,
+      "loss": 1.1284,
+      "step": 4411
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.197118175729563e-06,
+      "loss": 1.2598,
+      "step": 4412
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1928065111677185e-06,
+      "loss": 1.3223,
+      "step": 4413
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1884960085617144e-06,
+      "loss": 1.3174,
+      "step": 4414
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.184186668953454e-06,
+      "loss": 1.1062,
+      "step": 4415
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.179878493384547e-06,
+      "loss": 1.2529,
+      "step": 4416
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.175571482896344e-06,
+      "loss": 1.292,
+      "step": 4417
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.171265638529885e-06,
+      "loss": 1.1782,
+      "step": 4418
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.166960961325955e-06,
+      "loss": 1.2588,
+      "step": 4419
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.162657452325039e-06,
+      "loss": 1.3394,
+      "step": 4420
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158355112567351e-06,
+      "loss": 1.2891,
+      "step": 4421
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.154053943092817e-06,
+      "loss": 1.2842,
+      "step": 4422
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.149753944941076e-06,
+      "loss": 1.312,
+      "step": 4423
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1454551191514965e-06,
+      "loss": 1.1626,
+      "step": 4424
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.141157466763147e-06,
+      "loss": 1.1494,
+      "step": 4425
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.136860988814833e-06,
+      "loss": 1.2598,
+      "step": 4426
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1325656863450525e-06,
+      "loss": 1.2783,
+      "step": 4427
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128271560392037e-06,
+      "loss": 1.4316,
+      "step": 4428
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123978611993723e-06,
+      "loss": 1.3584,
+      "step": 4429
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.119686842187777e-06,
+      "loss": 1.2695,
+      "step": 4430
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115396252011562e-06,
+      "loss": 1.1506,
+      "step": 4431
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1111068425021644e-06,
+      "loss": 1.1968,
+      "step": 4432
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106818614696394e-06,
+      "loss": 1.061,
+      "step": 4433
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102531569630755e-06,
+      "loss": 1.2876,
+      "step": 4434
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0982457083414914e-06,
+      "loss": 1.3799,
+      "step": 4435
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0939610318645314e-06,
+      "loss": 1.2261,
+      "step": 4436
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.089677541235544e-06,
+      "loss": 1.2378,
+      "step": 4437
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08539523748989e-06,
+      "loss": 1.1655,
+      "step": 4438
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081114121662663e-06,
+      "loss": 1.3271,
+      "step": 4439
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.076834194788656e-06,
+      "loss": 1.0737,
+      "step": 4440
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0725554579023705e-06,
+      "loss": 1.2627,
+      "step": 4441
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068277912038042e-06,
+      "loss": 1.3027,
+      "step": 4442
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.064001558229591e-06,
+      "loss": 1.1816,
+      "step": 4443
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.059726397510678e-06,
+      "loss": 1.3652,
+      "step": 4444
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.055452430914643e-06,
+      "loss": 1.251,
+      "step": 4445
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.051179659474568e-06,
+      "loss": 1.2031,
+      "step": 4446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.046908084223221e-06,
+      "loss": 1.459,
+      "step": 4447
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.042637706193106e-06,
+      "loss": 1.2529,
+      "step": 4448
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.038368526416415e-06,
+      "loss": 1.3223,
+      "step": 4449
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0341005459250595e-06,
+      "loss": 1.3408,
+      "step": 4450
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029833765750667e-06,
+      "loss": 1.2056,
+      "step": 4451
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.025568186924561e-06,
+      "loss": 1.293,
+      "step": 4452
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.021303810477796e-06,
+      "loss": 1.2402,
+      "step": 4453
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017040637441104e-06,
+      "loss": 1.2363,
+      "step": 4454
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.01277866884496e-06,
+      "loss": 1.3486,
+      "step": 4455
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.008517905719521e-06,
+      "loss": 1.1357,
+      "step": 4456
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.004258349094675e-06,
+      "loss": 1.1079,
+      "step": 4457
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 1.3267,
+      "step": 4458
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995742859464792e-06,
+      "loss": 1.1704,
+      "step": 4459
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.991486928518056e-06,
+      "loss": 1.3218,
+      "step": 4460
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987232208188491e-06,
+      "loss": 1.1577,
+      "step": 4461
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9829786995045296e-06,
+      "loss": 1.3545,
+      "step": 4462
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978726403494277e-06,
+      "loss": 1.3076,
+      "step": 4463
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974475321185578e-06,
+      "loss": 1.2793,
+      "step": 4464
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97022545360596e-06,
+      "loss": 1.2295,
+      "step": 4465
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.965976801782675e-06,
+      "loss": 1.1128,
+      "step": 4466
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.961729366742668e-06,
+      "loss": 1.1885,
+      "step": 4467
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.957483149512592e-06,
+      "loss": 1.2129,
+      "step": 4468
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9532381511188144e-06,
+      "loss": 1.333,
+      "step": 4469
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.948994372587396e-06,
+      "loss": 1.1958,
+      "step": 4470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.944751814944118e-06,
+      "loss": 1.3135,
+      "step": 4471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940510479214443e-06,
+      "loss": 1.0518,
+      "step": 4472
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936270366423563e-06,
+      "loss": 1.2427,
+      "step": 4473
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.932031477596359e-06,
+      "loss": 1.2588,
+      "step": 4474
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927793813757424e-06,
+      "loss": 1.2476,
+      "step": 4475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92355737593105e-06,
+      "loss": 1.2695,
+      "step": 4476
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.91932216514123e-06,
+      "loss": 1.2495,
+      "step": 4477
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915088182411674e-06,
+      "loss": 1.3398,
+      "step": 4478
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9108554287657775e-06,
+      "loss": 1.2017,
+      "step": 4479
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906623905226659e-06,
+      "loss": 1.3135,
+      "step": 4480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9023936128171115e-06,
+      "loss": 1.3389,
+      "step": 4481
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89816455255966e-06,
+      "loss": 1.2554,
+      "step": 4482
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.89393672547651e-06,
+      "loss": 1.3584,
+      "step": 4483
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889710132589586e-06,
+      "loss": 1.2998,
+      "step": 4484
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.885484774920501e-06,
+      "loss": 1.2417,
+      "step": 4485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881260653490572e-06,
+      "loss": 1.3721,
+      "step": 4486
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.877037769320826e-06,
+      "loss": 1.2466,
+      "step": 4487
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872816123431977e-06,
+      "loss": 1.1934,
+      "step": 4488
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868595716844453e-06,
+      "loss": 1.2153,
+      "step": 4489
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8643765505783755e-06,
+      "loss": 1.3159,
+      "step": 4490
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8601586256535645e-06,
+      "loss": 1.3081,
+      "step": 4491
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.855941943089541e-06,
+      "loss": 1.4414,
+      "step": 4492
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.851726503905534e-06,
+      "loss": 1.2251,
+      "step": 4493
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.847512309120461e-06,
+      "loss": 1.2837,
+      "step": 4494
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.84329935975294e-06,
+      "loss": 1.3423,
+      "step": 4495
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8390876568212985e-06,
+      "loss": 1.2305,
+      "step": 4496
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.834877201343546e-06,
+      "loss": 1.0894,
+      "step": 4497
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.83066799433741e-06,
+      "loss": 1.1787,
+      "step": 4498
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8264600368202995e-06,
+      "loss": 1.1289,
+      "step": 4499
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82225332980933e-06,
+      "loss": 1.0486,
+      "step": 4500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.818047874321308e-06,
+      "loss": 1.2441,
+      "step": 4501
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.813843671372749e-06,
+      "loss": 1.1489,
+      "step": 4502
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.809640721979856e-06,
+      "loss": 1.1934,
+      "step": 4503
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805439027158527e-06,
+      "loss": 1.1758,
+      "step": 4504
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801238587924371e-06,
+      "loss": 1.2554,
+      "step": 4505
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.797039405292674e-06,
+      "loss": 1.0801,
+      "step": 4506
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.792841480278438e-06,
+      "loss": 1.0879,
+      "step": 4507
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.788644813896347e-06,
+      "loss": 1.0884,
+      "step": 4508
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.784449407160786e-06,
+      "loss": 1.3545,
+      "step": 4509
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780255261085829e-06,
+      "loss": 1.3066,
+      "step": 4510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.776062376685261e-06,
+      "loss": 1.1777,
+      "step": 4511
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.771870754972545e-06,
+      "loss": 1.2588,
+      "step": 4512
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.767680396960845e-06,
+      "loss": 1.2363,
+      "step": 4513
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7634913036630274e-06,
+      "loss": 1.3301,
+      "step": 4514
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.759303476091637e-06,
+      "loss": 1.3428,
+      "step": 4515
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75511691525893e-06,
+      "loss": 1.4316,
+      "step": 4516
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.750931622176844e-06,
+      "loss": 1.3135,
+      "step": 4517
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.746747597857014e-06,
+      "loss": 1.2534,
+      "step": 4518
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742564843310766e-06,
+      "loss": 1.3467,
+      "step": 4519
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738383359549127e-06,
+      "loss": 1.2197,
+      "step": 4520
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.734203147582809e-06,
+      "loss": 1.0835,
+      "step": 4521
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.730024208422214e-06,
+      "loss": 1.2598,
+      "step": 4522
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.725846543077449e-06,
+      "loss": 1.2476,
+      "step": 4523
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7216701525583e-06,
+      "loss": 1.2437,
+      "step": 4524
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717495037874255e-06,
+      "loss": 1.2866,
+      "step": 4525
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713321200034486e-06,
+      "loss": 1.2656,
+      "step": 4526
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7091486400478605e-06,
+      "loss": 1.2251,
+      "step": 4527
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7049773589229306e-06,
+      "loss": 1.2212,
+      "step": 4528
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700807357667953e-06,
+      "loss": 1.2266,
+      "step": 4529
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696638637290861e-06,
+      "loss": 1.2363,
+      "step": 4530
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6924711987992835e-06,
+      "loss": 1.2549,
+      "step": 4531
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688305043200544e-06,
+      "loss": 1.1392,
+      "step": 4532
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.684140171501647e-06,
+      "loss": 1.1855,
+      "step": 4533
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.679976584709297e-06,
+      "loss": 1.1807,
+      "step": 4534
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.67581428382988e-06,
+      "loss": 1.1021,
+      "step": 4535
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.671653269869474e-06,
+      "loss": 1.2549,
+      "step": 4536
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.66749354383384e-06,
+      "loss": 1.2031,
+      "step": 4537
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663335106728442e-06,
+      "loss": 1.2627,
+      "step": 4538
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6591779595584204e-06,
+      "loss": 1.2344,
+      "step": 4539
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655022103328603e-06,
+      "loss": 1.2852,
+      "step": 4540
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.650867539043515e-06,
+      "loss": 1.1836,
+      "step": 4541
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.646714267707361e-06,
+      "loss": 1.2725,
+      "step": 4542
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642562290324039e-06,
+      "loss": 1.1875,
+      "step": 4543
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63841160789713e-06,
+      "loss": 1.2993,
+      "step": 4544
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.634262221429903e-06,
+      "loss": 1.1367,
+      "step": 4545
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63011413192531e-06,
+      "loss": 1.4414,
+      "step": 4546
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.625967340386001e-06,
+      "loss": 1.4424,
+      "step": 4547
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621821847814298e-06,
+      "loss": 1.25,
+      "step": 4548
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.617677655212224e-06,
+      "loss": 1.2539,
+      "step": 4549
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.613534763581473e-06,
+      "loss": 1.4072,
+      "step": 4550
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60939317392343e-06,
+      "loss": 1.1221,
+      "step": 4551
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.605252887239173e-06,
+      "loss": 1.124,
+      "step": 4552
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.601113904529456e-06,
+      "loss": 1.3721,
+      "step": 4553
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.596976226794718e-06,
+      "loss": 1.1797,
+      "step": 4554
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.592839855035084e-06,
+      "loss": 1.0422,
+      "step": 4555
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.588704790250366e-06,
+      "loss": 1.1782,
+      "step": 4556
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.584571033440062e-06,
+      "loss": 1.2095,
+      "step": 4557
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580438585603347e-06,
+      "loss": 1.0859,
+      "step": 4558
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.576307447739084e-06,
+      "loss": 1.1763,
+      "step": 4559
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.572177620845811e-06,
+      "loss": 1.1826,
+      "step": 4560
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.568049105921768e-06,
+      "loss": 1.1782,
+      "step": 4561
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5639219039648594e-06,
+      "loss": 1.3228,
+      "step": 4562
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559796015972677e-06,
+      "loss": 1.2373,
+      "step": 4563
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.555671442942504e-06,
+      "loss": 1.2817,
+      "step": 4564
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5515481858712904e-06,
+      "loss": 1.3491,
+      "step": 4565
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547426245755685e-06,
+      "loss": 1.1255,
+      "step": 4566
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.543305623592005e-06,
+      "loss": 1.2383,
+      "step": 4567
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5391863203762554e-06,
+      "loss": 1.2676,
+      "step": 4568
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5350683371041146e-06,
+      "loss": 1.2183,
+      "step": 4569
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.530951674770958e-06,
+      "loss": 1.3008,
+      "step": 4570
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.526836334371827e-06,
+      "loss": 1.374,
+      "step": 4571
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5227223169014455e-06,
+      "loss": 1.2163,
+      "step": 4572
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.518609623354227e-06,
+      "loss": 1.2881,
+      "step": 4573
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.51449825472425e-06,
+      "loss": 1.4219,
+      "step": 4574
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.51038821200529e-06,
+      "loss": 1.3228,
+      "step": 4575
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.50627949619079e-06,
+      "loss": 1.4067,
+      "step": 4576
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5021721082738744e-06,
+      "loss": 1.3481,
+      "step": 4577
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.498066049247344e-06,
+      "loss": 1.2363,
+      "step": 4578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49396132010369e-06,
+      "loss": 1.2246,
+      "step": 4579
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.489857921835071e-06,
+      "loss": 1.2939,
+      "step": 4580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4857558554333225e-06,
+      "loss": 1.0532,
+      "step": 4581
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.481655121889971e-06,
+      "loss": 1.2065,
+      "step": 4582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.477555722196204e-06,
+      "loss": 1.1094,
+      "step": 4583
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473457657342903e-06,
+      "loss": 1.2925,
+      "step": 4584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469360928320618e-06,
+      "loss": 1.2773,
+      "step": 4585
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.465265536119572e-06,
+      "loss": 1.4229,
+      "step": 4586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4611714817296705e-06,
+      "loss": 1.3076,
+      "step": 4587
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4570787661405e-06,
+      "loss": 1.3516,
+      "step": 4588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.452987390341316e-06,
+      "loss": 1.1543,
+      "step": 4589
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.448897355321049e-06,
+      "loss": 1.168,
+      "step": 4590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.444808662068316e-06,
+      "loss": 1.0713,
+      "step": 4591
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440721311571396e-06,
+      "loss": 1.2324,
+      "step": 4592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4366353048182586e-06,
+      "loss": 1.2539,
+      "step": 4593
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.432550642796534e-06,
+      "loss": 1.2295,
+      "step": 4594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.428467326493535e-06,
+      "loss": 1.3145,
+      "step": 4595
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424385356896245e-06,
+      "loss": 1.2222,
+      "step": 4596
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4203047349913295e-06,
+      "loss": 1.0537,
+      "step": 4597
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.416225461765122e-06,
+      "loss": 1.2261,
+      "step": 4598
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412147538203626e-06,
+      "loss": 1.1646,
+      "step": 4599
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408070965292534e-06,
+      "loss": 1.1333,
+      "step": 4600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403995744017191e-06,
+      "loss": 1.1841,
+      "step": 4601
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399921875362636e-06,
+      "loss": 1.0571,
+      "step": 4602
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.395849360313569e-06,
+      "loss": 1.2461,
+      "step": 4603
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3917781998543626e-06,
+      "loss": 1.1572,
+      "step": 4604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.387708394969063e-06,
+      "loss": 1.1187,
+      "step": 4605
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.383639946641397e-06,
+      "loss": 1.231,
+      "step": 4606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37957285585475e-06,
+      "loss": 1.1631,
+      "step": 4607
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375507123592194e-06,
+      "loss": 1.1733,
+      "step": 4608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37144275083646e-06,
+      "loss": 1.1548,
+      "step": 4609
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.367379738569952e-06,
+      "loss": 1.1592,
+      "step": 4610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363318087774756e-06,
+      "loss": 1.1362,
+      "step": 4611
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.359257799432617e-06,
+      "loss": 1.1587,
+      "step": 4612
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3551988745249565e-06,
+      "loss": 1.397,
+      "step": 4613
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.351141314032858e-06,
+      "loss": 1.2139,
+      "step": 4614
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3470851189370944e-06,
+      "loss": 1.2134,
+      "step": 4615
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.343030290218084e-06,
+      "loss": 1.3936,
+      "step": 4616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.338976828855938e-06,
+      "loss": 1.0881,
+      "step": 4617
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.334924735830421e-06,
+      "loss": 1.1133,
+      "step": 4618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3308740121209695e-06,
+      "loss": 1.1929,
+      "step": 4619
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326824658706697e-06,
+      "loss": 1.1313,
+      "step": 4620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322776676566378e-06,
+      "loss": 1.2476,
+      "step": 4621
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318730066678458e-06,
+      "loss": 1.1465,
+      "step": 4622
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.314684830021047e-06,
+      "loss": 1.0767,
+      "step": 4623
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.310640967571933e-06,
+      "loss": 1.27,
+      "step": 4624
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306598480308559e-06,
+      "loss": 1.2959,
+      "step": 4625
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3025573692080516e-06,
+      "loss": 1.2881,
+      "step": 4626
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.298517635247188e-06,
+      "loss": 1.3594,
+      "step": 4627
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29447927940242e-06,
+      "loss": 1.1509,
+      "step": 4628
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290442302649871e-06,
+      "loss": 1.1709,
+      "step": 4629
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.286406705965324e-06,
+      "loss": 1.2998,
+      "step": 4630
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.282372490324229e-06,
+      "loss": 1.2363,
+      "step": 4631
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2783396567017025e-06,
+      "loss": 1.0334,
+      "step": 4632
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.274308206072535e-06,
+      "loss": 1.124,
+      "step": 4633
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.270278139411167e-06,
+      "loss": 1.4219,
+      "step": 4634
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266249457691724e-06,
+      "loss": 1.1875,
+      "step": 4635
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.262222161887979e-06,
+      "loss": 1.271,
+      "step": 4636
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258196252973375e-06,
+      "loss": 1.2393,
+      "step": 4637
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.254171731921032e-06,
+      "loss": 1.0742,
+      "step": 4638
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250148599703718e-06,
+      "loss": 1.1792,
+      "step": 4639
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.246126857293873e-06,
+      "loss": 1.3545,
+      "step": 4640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.242106505663594e-06,
+      "loss": 1.3086,
+      "step": 4641
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238087545784659e-06,
+      "loss": 1.4268,
+      "step": 4642
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.23406997862849e-06,
+      "loss": 1.2178,
+      "step": 4643
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230053805166186e-06,
+      "loss": 1.3613,
+      "step": 4644
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2260390263685014e-06,
+      "loss": 1.2373,
+      "step": 4645
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.222025643205854e-06,
+      "loss": 1.3154,
+      "step": 4646
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.21801365664833e-06,
+      "loss": 1.1221,
+      "step": 4647
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.214003067665674e-06,
+      "loss": 1.2646,
+      "step": 4648
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.209993877227291e-06,
+      "loss": 1.3682,
+      "step": 4649
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.205986086302247e-06,
+      "loss": 1.2085,
+      "step": 4650
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2019796958592805e-06,
+      "loss": 1.2554,
+      "step": 4651
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197974706866775e-06,
+      "loss": 1.2012,
+      "step": 4652
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.193971120292793e-06,
+      "loss": 1.3408,
+      "step": 4653
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189968937105046e-06,
+      "loss": 1.0908,
+      "step": 4654
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1859681582709025e-06,
+      "loss": 1.4492,
+      "step": 4655
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.181968784757408e-06,
+      "loss": 1.1465,
+      "step": 4656
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177970817531254e-06,
+      "loss": 1.4482,
+      "step": 4657
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173974257558797e-06,
+      "loss": 1.2158,
+      "step": 4658
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.16997910580605e-06,
+      "loss": 1.1685,
+      "step": 4659
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1659853632386955e-06,
+      "loss": 1.1255,
+      "step": 4660
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161993030822061e-06,
+      "loss": 1.2109,
+      "step": 4661
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158002109521149e-06,
+      "loss": 1.0127,
+      "step": 4662
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154012600300607e-06,
+      "loss": 1.1733,
+      "step": 4663
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.150024504124746e-06,
+      "loss": 1.1406,
+      "step": 4664
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146037821957542e-06,
+      "loss": 1.2549,
+      "step": 4665
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.142052554762615e-06,
+      "loss": 1.2837,
+      "step": 4666
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.138068703503264e-06,
+      "loss": 1.2065,
+      "step": 4667
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1340862691424184e-06,
+      "loss": 1.2622,
+      "step": 4668
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.130105252642692e-06,
+      "loss": 1.1973,
+      "step": 4669
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1261256549663344e-06,
+      "loss": 1.2559,
+      "step": 4670
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 1.335,
+      "step": 4671
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.118170719931067e-06,
+      "loss": 1.3813,
+      "step": 4672
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114195384494954e-06,
+      "loss": 1.1797,
+      "step": 4673
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110221471727822e-06,
+      "loss": 1.2803,
+      "step": 4674
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1062489825902055e-06,
+      "loss": 1.2188,
+      "step": 4675
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.102277918042315e-06,
+      "loss": 1.2524,
+      "step": 4676
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098308279043987e-06,
+      "loss": 1.1479,
+      "step": 4677
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.094340066554742e-06,
+      "loss": 1.1982,
+      "step": 4678
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090373281533741e-06,
+      "loss": 1.2705,
+      "step": 4679
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086407924939803e-06,
+      "loss": 1.3105,
+      "step": 4680
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082443997731405e-06,
+      "loss": 1.2437,
+      "step": 4681
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.078481500866668e-06,
+      "loss": 1.2402,
+      "step": 4682
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.074520435303383e-06,
+      "loss": 1.188,
+      "step": 4683
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07056080199898e-06,
+      "loss": 1.2422,
+      "step": 4684
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066602601910558e-06,
+      "loss": 1.2173,
+      "step": 4685
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.062645835994848e-06,
+      "loss": 1.1699,
+      "step": 4686
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05869050520826e-06,
+      "loss": 1.1978,
+      "step": 4687
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054736610506836e-06,
+      "loss": 1.1885,
+      "step": 4688
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0507841528462845e-06,
+      "loss": 1.2212,
+      "step": 4689
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.04683313318196e-06,
+      "loss": 0.9077,
+      "step": 4690
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.042883552468867e-06,
+      "loss": 1.186,
+      "step": 4691
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.038935411661673e-06,
+      "loss": 1.3125,
+      "step": 4692
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034988711714685e-06,
+      "loss": 1.1719,
+      "step": 4693
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031043453581875e-06,
+      "loss": 1.3604,
+      "step": 4694
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.027099638216847e-06,
+      "loss": 1.2256,
+      "step": 4695
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.023157266572879e-06,
+      "loss": 1.2524,
+      "step": 4696
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0192163396028804e-06,
+      "loss": 1.1831,
+      "step": 4697
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.015276858259427e-06,
+      "loss": 1.1821,
+      "step": 4698
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011338823494737e-06,
+      "loss": 1.2144,
+      "step": 4699
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007402236260674e-06,
+      "loss": 1.311,
+      "step": 4700
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.003467097508767e-06,
+      "loss": 1.2954,
+      "step": 4701
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.99953340819018e-06,
+      "loss": 1.1543,
+      "step": 4702
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9956011692557385e-06,
+      "loss": 1.2832,
+      "step": 4703
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.991670381655901e-06,
+      "loss": 1.4375,
+      "step": 4704
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987741046340792e-06,
+      "loss": 1.0417,
+      "step": 4705
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.983813164260182e-06,
+      "loss": 1.3584,
+      "step": 4706
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979886736363482e-06,
+      "loss": 1.2451,
+      "step": 4707
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975961763599759e-06,
+      "loss": 1.1836,
+      "step": 4708
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.972038246917719e-06,
+      "loss": 1.3506,
+      "step": 4709
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968116187265732e-06,
+      "loss": 1.1431,
+      "step": 4710
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.964195585591797e-06,
+      "loss": 1.0864,
+      "step": 4711
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.96027644284358e-06,
+      "loss": 1.0854,
+      "step": 4712
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.95635875996838e-06,
+      "loss": 1.2603,
+      "step": 4713
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.952442537913142e-06,
+      "loss": 1.1851,
+      "step": 4714
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948527777624475e-06,
+      "loss": 1.2812,
+      "step": 4715
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.944614480048614e-06,
+      "loss": 1.3242,
+      "step": 4716
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9407026461314535e-06,
+      "loss": 1.1035,
+      "step": 4717
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.936792276818525e-06,
+      "loss": 1.2505,
+      "step": 4718
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932883373055019e-06,
+      "loss": 1.2837,
+      "step": 4719
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928975935785757e-06,
+      "loss": 1.3101,
+      "step": 4720
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9250699659552195e-06,
+      "loss": 1.2202,
+      "step": 4721
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.921165464507523e-06,
+      "loss": 1.1738,
+      "step": 4722
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917262432386428e-06,
+      "loss": 1.2124,
+      "step": 4723
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913360870535352e-06,
+      "loss": 1.2563,
+      "step": 4724
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90946077989734e-06,
+      "loss": 1.2598,
+      "step": 4725
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.905562161415102e-06,
+      "loss": 1.2515,
+      "step": 4726
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.901665016030965e-06,
+      "loss": 1.2437,
+      "step": 4727
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.897769344686929e-06,
+      "loss": 1.1763,
+      "step": 4728
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893875148324614e-06,
+      "loss": 1.1294,
+      "step": 4729
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.889982427885302e-06,
+      "loss": 1.1602,
+      "step": 4730
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.886091184309906e-06,
+      "loss": 1.2197,
+      "step": 4731
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882201418538982e-06,
+      "loss": 1.2949,
+      "step": 4732
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87831313151274e-06,
+      "loss": 1.1582,
+      "step": 4733
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.874426324171019e-06,
+      "loss": 1.2852,
+      "step": 4734
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870540997453316e-06,
+      "loss": 1.2432,
+      "step": 4735
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.866657152298746e-06,
+      "loss": 1.2744,
+      "step": 4736
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8627747896460935e-06,
+      "loss": 1.085,
+      "step": 4737
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858893910433763e-06,
+      "loss": 1.21,
+      "step": 4738
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855014515599815e-06,
+      "loss": 1.2598,
+      "step": 4739
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.851136606081946e-06,
+      "loss": 1.1357,
+      "step": 4740
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.847260182817485e-06,
+      "loss": 1.2568,
+      "step": 4741
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 1.207,
+      "step": 4742
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.839511798796357e-06,
+      "loss": 1.3779,
+      "step": 4743
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835639839912569e-06,
+      "loss": 1.1099,
+      "step": 4744
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.831769371027941e-06,
+      "loss": 1.3062,
+      "step": 4745
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827900393078021e-06,
+      "loss": 1.2539,
+      "step": 4746
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.82403290699798e-06,
+      "loss": 1.3633,
+      "step": 4747
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820166913722642e-06,
+      "loss": 1.2373,
+      "step": 4748
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.816302414186459e-06,
+      "loss": 1.2266,
+      "step": 4749
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.812439409323525e-06,
+      "loss": 1.3232,
+      "step": 4750
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8085779000675802e-06,
+      "loss": 1.2861,
+      "step": 4751
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.804717887351991e-06,
+      "loss": 1.1782,
+      "step": 4752
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8008593721097775e-06,
+      "loss": 1.2842,
+      "step": 4753
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7970023552735755e-06,
+      "loss": 1.1553,
+      "step": 4754
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7931468377756832e-06,
+      "loss": 1.2959,
+      "step": 4755
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7892928205480173e-06,
+      "loss": 1.1831,
+      "step": 4756
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7854403045221466e-06,
+      "loss": 1.2358,
+      "step": 4757
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.781589290629266e-06,
+      "loss": 1.0854,
+      "step": 4758
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777739779800208e-06,
+      "loss": 1.1973,
+      "step": 4759
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7738917729654534e-06,
+      "loss": 1.147,
+      "step": 4760
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7700452710551028e-06,
+      "loss": 1.4277,
+      "step": 4761
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7662002749989123e-06,
+      "loss": 1.4038,
+      "step": 4762
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.76235678572625e-06,
+      "loss": 1.3169,
+      "step": 4763
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.758514804166142e-06,
+      "loss": 1.2529,
+      "step": 4764
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7546743312472366e-06,
+      "loss": 1.1431,
+      "step": 4765
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.750835367897825e-06,
+      "loss": 1.4492,
+      "step": 4766
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.74699791504583e-06,
+      "loss": 1.2222,
+      "step": 4767
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.743161973618804e-06,
+      "loss": 1.335,
+      "step": 4768
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.739327544543948e-06,
+      "loss": 1.2559,
+      "step": 4769
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.735494628748082e-06,
+      "loss": 1.2026,
+      "step": 4770
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7316632271576768e-06,
+      "loss": 1.2959,
+      "step": 4771
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7278333406988144e-06,
+      "loss": 1.1602,
+      "step": 4772
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7240049702972347e-06,
+      "loss": 1.2266,
+      "step": 4773
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7201781168782925e-06,
+      "loss": 1.2349,
+      "step": 4774
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.716352781366992e-06,
+      "loss": 1.2441,
+      "step": 4775
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.712528964687958e-06,
+      "loss": 1.1763,
+      "step": 4776
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.708706667765449e-06,
+      "loss": 1.1904,
+      "step": 4777
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7048858915233665e-06,
+      "loss": 1.1406,
+      "step": 4778
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.701066636885231e-06,
+      "loss": 1.1738,
+      "step": 4779
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.697248904774212e-06,
+      "loss": 1.2778,
+      "step": 4780
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6934326961130875e-06,
+      "loss": 1.4141,
+      "step": 4781
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6896180118242897e-06,
+      "loss": 1.1631,
+      "step": 4782
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.685804852829867e-06,
+      "loss": 1.2842,
+      "step": 4783
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.681993220051512e-06,
+      "loss": 1.1279,
+      "step": 4784
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6781831144105384e-06,
+      "loss": 1.1499,
+      "step": 4785
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6743745368278903e-06,
+      "loss": 1.3613,
+      "step": 4786
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6705674882241527e-06,
+      "loss": 1.1926,
+      "step": 4787
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6667619695195287e-06,
+      "loss": 1.3672,
+      "step": 4788
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6629579816338676e-06,
+      "loss": 1.2134,
+      "step": 4789
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.659155525486624e-06,
+      "loss": 1.1226,
+      "step": 4790
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6553546019969075e-06,
+      "loss": 1.3896,
+      "step": 4791
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.651555212083441e-06,
+      "loss": 1.1938,
+      "step": 4792
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6477573566645854e-06,
+      "loss": 1.3369,
+      "step": 4793
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.643961036658328e-06,
+      "loss": 1.3037,
+      "step": 4794
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.640166252982279e-06,
+      "loss": 1.25,
+      "step": 4795
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6363730065536895e-06,
+      "loss": 1.2463,
+      "step": 4796
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6325812982894272e-06,
+      "loss": 1.2319,
+      "step": 4797
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6287911291060017e-06,
+      "loss": 1.2236,
+      "step": 4798
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6250024999195287e-06,
+      "loss": 1.2285,
+      "step": 4799
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6212154116457752e-06,
+      "loss": 1.334,
+      "step": 4800
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6174298652001184e-06,
+      "loss": 1.3301,
+      "step": 4801
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.613645861497578e-06,
+      "loss": 1.2959,
+      "step": 4802
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6098634014527866e-06,
+      "loss": 1.2817,
+      "step": 4803
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.606082485980009e-06,
+      "loss": 1.0889,
+      "step": 4804
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.602303115993144e-06,
+      "loss": 1.1606,
+      "step": 4805
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5985252924057023e-06,
+      "loss": 1.1108,
+      "step": 4806
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5947490161308395e-06,
+      "loss": 1.2031,
+      "step": 4807
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5909742880813125e-06,
+      "loss": 1.2471,
+      "step": 4808
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5872011091695292e-06,
+      "loss": 1.2505,
+      "step": 4809
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.583429480307503e-06,
+      "loss": 1.3496,
+      "step": 4810
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.579659402406891e-06,
+      "loss": 1.2886,
+      "step": 4811
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.575890876378961e-06,
+      "loss": 1.2441,
+      "step": 4812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 1.144,
+      "step": 4813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5683584835843577e-06,
+      "loss": 1.4155,
+      "step": 4814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5645946186383553e-06,
+      "loss": 1.1719,
+      "step": 4815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5608323092063767e-06,
+      "loss": 1.2617,
+      "step": 4816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5570715561978076e-06,
+      "loss": 1.3818,
+      "step": 4817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5533123605216748e-06,
+      "loss": 1.1597,
+      "step": 4818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5495547230866134e-06,
+      "loss": 1.1411,
+      "step": 4819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5457986448008973e-06,
+      "loss": 1.2847,
+      "step": 4820
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5420441265724104e-06,
+      "loss": 1.2095,
+      "step": 4821
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.538291169308662e-06,
+      "loss": 1.1968,
+      "step": 4822
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.534539773916793e-06,
+      "loss": 1.1294,
+      "step": 4823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.530789941303554e-06,
+      "loss": 1.1904,
+      "step": 4824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5270416723753333e-06,
+      "loss": 1.1724,
+      "step": 4825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.523294968038119e-06,
+      "loss": 1.2285,
+      "step": 4826
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.519549829197544e-06,
+      "loss": 1.335,
+      "step": 4827
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5158062567588468e-06,
+      "loss": 1.1387,
+      "step": 4828
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5120642516269e-06,
+      "loss": 1.2568,
+      "step": 4829
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.508323814706187e-06,
+      "loss": 1.3091,
+      "step": 4830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.504584946900813e-06,
+      "loss": 1.2822,
+      "step": 4831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5008476491145137e-06,
+      "loss": 1.21,
+      "step": 4832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49711192225063e-06,
+      "loss": 1.3516,
+      "step": 4833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.493377767212144e-06,
+      "loss": 1.1753,
+      "step": 4834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48964518490163e-06,
+      "loss": 1.2783,
+      "step": 4835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.485914176221308e-06,
+      "loss": 1.2173,
+      "step": 4836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4821847420730013e-06,
+      "loss": 1.2285,
+      "step": 4837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478456883358162e-06,
+      "loss": 1.4424,
+      "step": 4838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4747306009778635e-06,
+      "loss": 1.249,
+      "step": 4839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4710058958327797e-06,
+      "loss": 1.2432,
+      "step": 4840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.467282768823226e-06,
+      "loss": 1.2715,
+      "step": 4841
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4635612208491197e-06,
+      "loss": 1.2476,
+      "step": 4842
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.45984125281001e-06,
+      "loss": 1.27,
+      "step": 4843
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4561228656050548e-06,
+      "loss": 1.1978,
+      "step": 4844
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.452406060133032e-06,
+      "loss": 1.2339,
+      "step": 4845
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4486908372923333e-06,
+      "loss": 1.1191,
+      "step": 4846
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4449771979809775e-06,
+      "loss": 1.2827,
+      "step": 4847
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4412651430966006e-06,
+      "loss": 1.3867,
+      "step": 4848
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4375546735364384e-06,
+      "loss": 1.2173,
+      "step": 4849
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4338457901973633e-06,
+      "loss": 1.2812,
+      "step": 4850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.430138493975852e-06,
+      "loss": 1.1494,
+      "step": 4851
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.426432785768008e-06,
+      "loss": 1.0283,
+      "step": 4852
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4227286664695416e-06,
+      "loss": 1.2612,
+      "step": 4853
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.419026136975779e-06,
+      "loss": 1.2793,
+      "step": 4854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4153251981816736e-06,
+      "loss": 1.2725,
+      "step": 4855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4116258509817777e-06,
+      "loss": 1.147,
+      "step": 4856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4079280962702788e-06,
+      "loss": 1.2275,
+      "step": 4857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4042319349409536e-06,
+      "loss": 1.0903,
+      "step": 4858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.40053736788722e-06,
+      "loss": 1.0591,
+      "step": 4859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3968443960020913e-06,
+      "loss": 1.2949,
+      "step": 4860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.393153020178209e-06,
+      "loss": 1.0576,
+      "step": 4861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.38946324130782e-06,
+      "loss": 1.2612,
+      "step": 4862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3857750602827832e-06,
+      "loss": 1.1689,
+      "step": 4863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3820884779945838e-06,
+      "loss": 1.3369,
+      "step": 4864
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.378403495334306e-06,
+      "loss": 1.252,
+      "step": 4865
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.374720113192662e-06,
+      "loss": 1.2993,
+      "step": 4866
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3710383324599572e-06,
+      "loss": 1.1084,
+      "step": 4867
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3673581540261324e-06,
+      "loss": 1.3149,
+      "step": 4868
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.363679578780723e-06,
+      "loss": 1.248,
+      "step": 4869
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3600026076128913e-06,
+      "loss": 1.2373,
+      "step": 4870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3563272414114e-06,
+      "loss": 1.2622,
+      "step": 4871
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3526534810646284e-06,
+      "loss": 1.1914,
+      "step": 4872
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.348981327460573e-06,
+      "loss": 1.1343,
+      "step": 4873
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.34531078148683e-06,
+      "loss": 1.2964,
+      "step": 4874
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3416418440306254e-06,
+      "loss": 1.115,
+      "step": 4875
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3379745159787703e-06,
+      "loss": 1.334,
+      "step": 4876
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3343087982177124e-06,
+      "loss": 1.1602,
+      "step": 4877
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.330644691633492e-06,
+      "loss": 1.3516,
+      "step": 4878
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3269821971117742e-06,
+      "loss": 1.2764,
+      "step": 4879
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.323321315537825e-06,
+      "loss": 1.1528,
+      "step": 4880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.319662047796519e-06,
+      "loss": 1.2041,
+      "step": 4881
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3160043947723506e-06,
+      "loss": 1.1953,
+      "step": 4882
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.312348357349412e-06,
+      "loss": 1.1748,
+      "step": 4883
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 1.0205,
+      "step": 4884
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.305041132841681e-06,
+      "loss": 1.1411,
+      "step": 4885
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301389947523129e-06,
+      "loss": 1.2002,
+      "step": 4886
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2977403813382923e-06,
+      "loss": 1.1714,
+      "step": 4887
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2940924351693216e-06,
+      "loss": 1.3125,
+      "step": 4888
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2904461098979644e-06,
+      "loss": 1.1411,
+      "step": 4889
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.28680140640558e-06,
+      "loss": 1.2329,
+      "step": 4890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2831583255731403e-06,
+      "loss": 1.3179,
+      "step": 4891
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2795168682812174e-06,
+      "loss": 1.3428,
+      "step": 4892
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.275877035410002e-06,
+      "loss": 1.0693,
+      "step": 4893
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2722388278392734e-06,
+      "loss": 1.1587,
+      "step": 4894
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2686022464484402e-06,
+      "loss": 1.2217,
+      "step": 4895
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2649672921165e-06,
+      "loss": 1.2354,
+      "step": 4896
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.26133396572207e-06,
+      "loss": 1.0942,
+      "step": 4897
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2577022681433666e-06,
+      "loss": 1.252,
+      "step": 4898
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2540722002582102e-06,
+      "loss": 1.2271,
+      "step": 4899
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250443762944039e-06,
+      "loss": 1.0791,
+      "step": 4900
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2468169570778817e-06,
+      "loss": 1.1289,
+      "step": 4901
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2431917835363867e-06,
+      "loss": 1.2417,
+      "step": 4902
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2395682431957997e-06,
+      "loss": 1.1406,
+      "step": 4903
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.235946336931973e-06,
+      "loss": 1.0156,
+      "step": 4904
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2323260656203614e-06,
+      "loss": 1.2954,
+      "step": 4905
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.228707430136032e-06,
+      "loss": 1.1731,
+      "step": 4906
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2250904313536523e-06,
+      "loss": 1.1191,
+      "step": 4907
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2214750701474875e-06,
+      "loss": 1.2583,
+      "step": 4908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.217861347391421e-06,
+      "loss": 1.0479,
+      "step": 4909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2142492639589274e-06,
+      "loss": 1.3457,
+      "step": 4910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2106388207230955e-06,
+      "loss": 1.1909,
+      "step": 4911
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2070300185566084e-06,
+      "loss": 1.1167,
+      "step": 4912
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.203422858331757e-06,
+      "loss": 1.2607,
+      "step": 4913
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1998173409204326e-06,
+      "loss": 1.1577,
+      "step": 4914
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.196213467194137e-06,
+      "loss": 1.0967,
+      "step": 4915
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1926112380239673e-06,
+      "loss": 1.1816,
+      "step": 4916
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1890106542806197e-06,
+      "loss": 1.3613,
+      "step": 4917
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185411716834407e-06,
+      "loss": 1.2065,
+      "step": 4918
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1818144265552273e-06,
+      "loss": 1.3486,
+      "step": 4919
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1782187843125957e-06,
+      "loss": 1.2207,
+      "step": 4920
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1746247909756176e-06,
+      "loss": 1.2656,
+      "step": 4921
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.171032447413005e-06,
+      "loss": 1.2266,
+      "step": 4922
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.167441754493066e-06,
+      "loss": 1.187,
+      "step": 4923
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1638527130837192e-06,
+      "loss": 1.2363,
+      "step": 4924
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1602653240524784e-06,
+      "loss": 1.0981,
+      "step": 4925
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1566795882664527e-06,
+      "loss": 1.2529,
+      "step": 4926
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.153095506592364e-06,
+      "loss": 1.1987,
+      "step": 4927
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.149513079896521e-06,
+      "loss": 1.1006,
+      "step": 4928
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1459323090448458e-06,
+      "loss": 1.1475,
+      "step": 4929
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1423531949028486e-06,
+      "loss": 1.1743,
+      "step": 4930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1387757383356455e-06,
+      "loss": 1.3037,
+      "step": 4931
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.135199940207947e-06,
+      "loss": 1.1694,
+      "step": 4932
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1316258013840707e-06,
+      "loss": 1.1279,
+      "step": 4933
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1280533227279275e-06,
+      "loss": 1.3091,
+      "step": 4934
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1244825051030236e-06,
+      "loss": 1.2324,
+      "step": 4935
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.120913349372474e-06,
+      "loss": 1.2012,
+      "step": 4936
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.117345856398981e-06,
+      "loss": 1.145,
+      "step": 4937
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.113780027044856e-06,
+      "loss": 1.2021,
+      "step": 4938
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.110215862171999e-06,
+      "loss": 1.3022,
+      "step": 4939
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1066533626419102e-06,
+      "loss": 1.23,
+      "step": 4940
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1030925293156865e-06,
+      "loss": 1.2437,
+      "step": 4941
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099533363054029e-06,
+      "loss": 1.3271,
+      "step": 4942
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.095975864717228e-06,
+      "loss": 1.2422,
+      "step": 4943
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09242003516517e-06,
+      "loss": 1.1333,
+      "step": 4944
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0888658752573466e-06,
+      "loss": 1.376,
+      "step": 4945
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.085313385852834e-06,
+      "loss": 1.2158,
+      "step": 4946
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0817625678103193e-06,
+      "loss": 1.1392,
+      "step": 4947
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.078213421988073e-06,
+      "loss": 1.2227,
+      "step": 4948
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0746659492439667e-06,
+      "loss": 1.2537,
+      "step": 4949
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0711201504354628e-06,
+      "loss": 1.2451,
+      "step": 4950
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0675760264196287e-06,
+      "loss": 1.0562,
+      "step": 4951
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0640335780531207e-06,
+      "loss": 1.1489,
+      "step": 4952
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0604928061921847e-06,
+      "loss": 1.2207,
+      "step": 4953
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.056953711692675e-06,
+      "loss": 1.2461,
+      "step": 4954
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0534162954100264e-06,
+      "loss": 1.334,
+      "step": 4955
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.049880558199281e-06,
+      "loss": 1.1504,
+      "step": 4956
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.046346500915065e-06,
+      "loss": 1.2563,
+      "step": 4957
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.042814124411603e-06,
+      "loss": 1.208,
+      "step": 4958
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0392834295427078e-06,
+      "loss": 1.0923,
+      "step": 4959
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.035754417161798e-06,
+      "loss": 1.3057,
+      "step": 4960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0322270881218706e-06,
+      "loss": 1.2783,
+      "step": 4961
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.028701443275529e-06,
+      "loss": 1.1807,
+      "step": 4962
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0251774834749627e-06,
+      "loss": 1.3408,
+      "step": 4963
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0216552095719474e-06,
+      "loss": 1.252,
+      "step": 4964
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.018134622417869e-06,
+      "loss": 1.3369,
+      "step": 4965
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0146157228636907e-06,
+      "loss": 1.2959,
+      "step": 4966
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.011098511759971e-06,
+      "loss": 1.2354,
+      "step": 4967
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.00758298995686e-06,
+      "loss": 1.2803,
+      "step": 4968
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.004069158304107e-06,
+      "loss": 1.3691,
+      "step": 4969
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0005570176510413e-06,
+      "loss": 1.3325,
+      "step": 4970
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.997046568846593e-06,
+      "loss": 1.124,
+      "step": 4971
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.99353781273928e-06,
+      "loss": 1.2036,
+      "step": 4972
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990030750177204e-06,
+      "loss": 1.2939,
+      "step": 4973
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.986525382008072e-06,
+      "loss": 1.3584,
+      "step": 4974
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.983021709079168e-06,
+      "loss": 1.0669,
+      "step": 4975
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.979519732237374e-06,
+      "loss": 1.2085,
+      "step": 4976
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.976019452329153e-06,
+      "loss": 1.1509,
+      "step": 4977
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9725208702005736e-06,
+      "loss": 1.2202,
+      "step": 4978
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.969023986697278e-06,
+      "loss": 1.1511,
+      "step": 4979
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965528802664509e-06,
+      "loss": 1.252,
+      "step": 4980
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.962035318947093e-06,
+      "loss": 1.1982,
+      "step": 4981
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.958543536389442e-06,
+      "loss": 1.2207,
+      "step": 4982
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.955053455835568e-06,
+      "loss": 1.1836,
+      "step": 4983
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.951565078129063e-06,
+      "loss": 1.2783,
+      "step": 4984
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.948078404113107e-06,
+      "loss": 1.0732,
+      "step": 4985
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9445934346304706e-06,
+      "loss": 1.1924,
+      "step": 4986
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9411101705235125e-06,
+      "loss": 1.2427,
+      "step": 4987
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9376286126341847e-06,
+      "loss": 1.1631,
+      "step": 4988
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9341487618040153e-06,
+      "loss": 1.3682,
+      "step": 4989
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9306706188741275e-06,
+      "loss": 1.2681,
+      "step": 4990
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.927194184685227e-06,
+      "loss": 1.1699,
+      "step": 4991
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9237194600776144e-06,
+      "loss": 1.2979,
+      "step": 4992
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9202464458911695e-06,
+      "loss": 1.2549,
+      "step": 4993
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9167751429653613e-06,
+      "loss": 1.1748,
+      "step": 4994
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9133055521392407e-06,
+      "loss": 1.1758,
+      "step": 4995
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9098376742514533e-06,
+      "loss": 1.2314,
+      "step": 4996
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.906371510140229e-06,
+      "loss": 1.1997,
+      "step": 4997
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9029070606433787e-06,
+      "loss": 1.292,
+      "step": 4998
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8994443265983e-06,
+      "loss": 1.2505,
+      "step": 4999
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8959833088419744e-06,
+      "loss": 1.1919,
+      "step": 5000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.892524008210976e-06,
+      "loss": 1.2983,
+      "step": 5001
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889066425541458e-06,
+      "loss": 1.2344,
+      "step": 5002
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8856105616691552e-06,
+      "loss": 1.3184,
+      "step": 5003
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.882156417429396e-06,
+      "loss": 1.3052,
+      "step": 5004
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8787039936570837e-06,
+      "loss": 1.2324,
+      "step": 5005
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8752532911867147e-06,
+      "loss": 1.269,
+      "step": 5006
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.871804310852363e-06,
+      "loss": 1.3271,
+      "step": 5007
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868357053487688e-06,
+      "loss": 1.3271,
+      "step": 5008
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.86491151992593e-06,
+      "loss": 1.1304,
+      "step": 5009
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.861467710999921e-06,
+      "loss": 1.2588,
+      "step": 5010
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.858025627542067e-06,
+      "loss": 1.1802,
+      "step": 5011
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.854585270384359e-06,
+      "loss": 1.1841,
+      "step": 5012
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8511466403583766e-06,
+      "loss": 1.1782,
+      "step": 5013
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.847709738295272e-06,
+      "loss": 1.1948,
+      "step": 5014
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.844274565025793e-06,
+      "loss": 1.3306,
+      "step": 5015
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8408411213802568e-06,
+      "loss": 1.23,
+      "step": 5016
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8374094081885686e-06,
+      "loss": 1.2314,
+      "step": 5017
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.83397942628021e-06,
+      "loss": 1.291,
+      "step": 5018
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.830551176484255e-06,
+      "loss": 1.3574,
+      "step": 5019
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.827124659629351e-06,
+      "loss": 1.2764,
+      "step": 5020
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.823699876543724e-06,
+      "loss": 1.2358,
+      "step": 5021
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8202768280551894e-06,
+      "loss": 1.2842,
+      "step": 5022
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8168555149911334e-06,
+      "loss": 1.019,
+      "step": 5023
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.813435938178536e-06,
+      "loss": 1.3477,
+      "step": 5024
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.810018098443944e-06,
+      "loss": 1.1191,
+      "step": 5025
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8066019966134907e-06,
+      "loss": 1.1753,
+      "step": 5026
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.803187633512886e-06,
+      "loss": 1.187,
+      "step": 5027
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7997750099674282e-06,
+      "loss": 1.2441,
+      "step": 5028
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7963641268019815e-06,
+      "loss": 1.4229,
+      "step": 5029
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7929549848410054e-06,
+      "loss": 1.1934,
+      "step": 5030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7895475849085254e-06,
+      "loss": 1.2793,
+      "step": 5031
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7861419278281467e-06,
+      "loss": 1.1348,
+      "step": 5032
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7827380144230655e-06,
+      "loss": 1.2061,
+      "step": 5033
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7793358455160435e-06,
+      "loss": 1.231,
+      "step": 5034
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7759354219294256e-06,
+      "loss": 1.1792,
+      "step": 5035
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7725367444851326e-06,
+      "loss": 1.3354,
+      "step": 5036
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.76913981400467e-06,
+      "loss": 1.3428,
+      "step": 5037
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7657446313091105e-06,
+      "loss": 1.0503,
+      "step": 5038
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7623511972191176e-06,
+      "loss": 1.2256,
+      "step": 5039
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7589595125549195e-06,
+      "loss": 1.1499,
+      "step": 5040
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7555695781363256e-06,
+      "loss": 1.125,
+      "step": 5041
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.752181394782728e-06,
+      "loss": 1.3223,
+      "step": 5042
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7487949633130882e-06,
+      "loss": 1.2788,
+      "step": 5043
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7454102845459475e-06,
+      "loss": 1.1021,
+      "step": 5044
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7420273592994206e-06,
+      "loss": 1.3047,
+      "step": 5045
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.738646188391204e-06,
+      "loss": 1.2168,
+      "step": 5046
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.735266772638564e-06,
+      "loss": 1.1763,
+      "step": 5047
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7318891128583514e-06,
+      "loss": 1.1479,
+      "step": 5048
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7285132098669807e-06,
+      "loss": 1.1372,
+      "step": 5049
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7251390644804475e-06,
+      "loss": 1.3623,
+      "step": 5050
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.721766677514329e-06,
+      "loss": 1.2188,
+      "step": 5051
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.718396049783767e-06,
+      "loss": 1.2109,
+      "step": 5052
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7150271821034824e-06,
+      "loss": 1.1211,
+      "step": 5053
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7116600752877688e-06,
+      "loss": 1.2764,
+      "step": 5054
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7082947301505e-06,
+      "loss": 1.2036,
+      "step": 5055
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7049311475051155e-06,
+      "loss": 1.1348,
+      "step": 5056
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7015693281646393e-06,
+      "loss": 1.0986,
+      "step": 5057
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.698209272941659e-06,
+      "loss": 1.168,
+      "step": 5058
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.694850982648338e-06,
+      "loss": 1.2334,
+      "step": 5059
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69149445809642e-06,
+      "loss": 1.2681,
+      "step": 5060
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688139700097214e-06,
+      "loss": 1.2598,
+      "step": 5061
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6847867094616063e-06,
+      "loss": 1.1719,
+      "step": 5062
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6814354870000503e-06,
+      "loss": 1.3638,
+      "step": 5063
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.678086033522583e-06,
+      "loss": 1.3633,
+      "step": 5064
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6747383498388003e-06,
+      "loss": 1.2959,
+      "step": 5065
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6713924367578836e-06,
+      "loss": 1.0693,
+      "step": 5066
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6680482950885777e-06,
+      "loss": 1.3047,
+      "step": 5067
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.664705925639197e-06,
+      "loss": 1.2363,
+      "step": 5068
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6613653292176387e-06,
+      "loss": 1.2974,
+      "step": 5069
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6580265066313614e-06,
+      "loss": 0.9995,
+      "step": 5070
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6546894586873993e-06,
+      "loss": 1.1675,
+      "step": 5071
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.651354186192352e-06,
+      "loss": 1.3867,
+      "step": 5072
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6480206899524008e-06,
+      "loss": 1.3018,
+      "step": 5073
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.644688970773287e-06,
+      "loss": 1.25,
+      "step": 5074
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.641359029460331e-06,
+      "loss": 1.1802,
+      "step": 5075
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6380308668184163e-06,
+      "loss": 1.1538,
+      "step": 5076
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6347044836519975e-06,
+      "loss": 1.3008,
+      "step": 5077
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.631379880765107e-06,
+      "loss": 1.3213,
+      "step": 5078
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6280570589613364e-06,
+      "loss": 1.2061,
+      "step": 5079
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.624736019043852e-06,
+      "loss": 1.2773,
+      "step": 5080
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6214167618153862e-06,
+      "loss": 1.2617,
+      "step": 5081
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6180992880782476e-06,
+      "loss": 1.2178,
+      "step": 5082
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6147835986343052e-06,
+      "loss": 1.1064,
+      "step": 5083
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6114696942850036e-06,
+      "loss": 1.3887,
+      "step": 5084
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6081575758313525e-06,
+      "loss": 1.2329,
+      "step": 5085
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.604847244073927e-06,
+      "loss": 1.2202,
+      "step": 5086
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6015386998128788e-06,
+      "loss": 1.2715,
+      "step": 5087
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5982319438479163e-06,
+      "loss": 1.1289,
+      "step": 5088
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5949269769783315e-06,
+      "loss": 1.1489,
+      "step": 5089
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5916238000029615e-06,
+      "loss": 1.3179,
+      "step": 5090
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5883224137202335e-06,
+      "loss": 1.3184,
+      "step": 5091
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5850228189281247e-06,
+      "loss": 1.2632,
+      "step": 5092
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5817250164241923e-06,
+      "loss": 1.3252,
+      "step": 5093
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.578429007005552e-06,
+      "loss": 1.2427,
+      "step": 5094
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.575134791468885e-06,
+      "loss": 1.0801,
+      "step": 5095
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.57184237061045e-06,
+      "loss": 1.1743,
+      "step": 5096
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 1.3037,
+      "step": 5097
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5652629161110977e-06,
+      "loss": 1.3062,
+      "step": 5098
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.561975884060509e-06,
+      "loss": 1.1421,
+      "step": 5099
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.558690649868817e-06,
+      "loss": 0.9121,
+      "step": 5100
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5554072143300932e-06,
+      "loss": 1.1821,
+      "step": 5101
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5521255782379908e-06,
+      "loss": 1.1533,
+      "step": 5102
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548845742385717e-06,
+      "loss": 1.374,
+      "step": 5103
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5455677075660434e-06,
+      "loss": 1.2993,
+      "step": 5104
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5422914745713157e-06,
+      "loss": 1.209,
+      "step": 5105
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5390170441934337e-06,
+      "loss": 1.2832,
+      "step": 5106
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5357444172238753e-06,
+      "loss": 1.1733,
+      "step": 5107
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5324735944536594e-06,
+      "loss": 1.2505,
+      "step": 5108
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.529204576673393e-06,
+      "loss": 1.2021,
+      "step": 5109
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525937364673231e-06,
+      "loss": 1.4072,
+      "step": 5110
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5226719592429037e-06,
+      "loss": 1.2705,
+      "step": 5111
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5194083611716937e-06,
+      "loss": 1.0742,
+      "step": 5112
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.516146571248449e-06,
+      "loss": 1.3594,
+      "step": 5113
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.512886590261591e-06,
+      "loss": 1.1265,
+      "step": 5114
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5096284189990873e-06,
+      "loss": 1.2759,
+      "step": 5115
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5063720582484863e-06,
+      "loss": 1.2666,
+      "step": 5116
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5031175087968774e-06,
+      "loss": 1.3984,
+      "step": 5117
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.499864771430934e-06,
+      "loss": 1.2109,
+      "step": 5118
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496613846936874e-06,
+      "loss": 1.2061,
+      "step": 5119
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.493364736100491e-06,
+      "loss": 0.9937,
+      "step": 5120
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4901174397071306e-06,
+      "loss": 1.2168,
+      "step": 5121
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4868719585417e-06,
+      "loss": 1.2744,
+      "step": 5122
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4836282933886757e-06,
+      "loss": 1.1001,
+      "step": 5123
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4803864450320867e-06,
+      "loss": 1.2148,
+      "step": 5124
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4771464142555325e-06,
+      "loss": 1.1416,
+      "step": 5125
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4739082018421567e-06,
+      "loss": 1.3643,
+      "step": 5126
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4706718085746828e-06,
+      "loss": 1.2852,
+      "step": 5127
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.467437235235378e-06,
+      "loss": 1.2207,
+      "step": 5128
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.464204482606085e-06,
+      "loss": 1.0557,
+      "step": 5129
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.460973551468194e-06,
+      "loss": 1.2456,
+      "step": 5130
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4577444426026574e-06,
+      "loss": 1.23,
+      "step": 5131
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4545171567899962e-06,
+      "loss": 1.0044,
+      "step": 5132
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.451291694810275e-06,
+      "loss": 1.0703,
+      "step": 5133
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.448068057443138e-06,
+      "loss": 1.3169,
+      "step": 5134
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4448462454677636e-06,
+      "loss": 1.1958,
+      "step": 5135
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.441626259662907e-06,
+      "loss": 1.2046,
+      "step": 5136
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4384081008068827e-06,
+      "loss": 1.0889,
+      "step": 5137
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4351917696775528e-06,
+      "loss": 1.0986,
+      "step": 5138
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4319772670523432e-06,
+      "loss": 1.1235,
+      "step": 5139
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.428764593708235e-06,
+      "loss": 1.2598,
+      "step": 5140
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.425553750421774e-06,
+      "loss": 1.2158,
+      "step": 5141
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4223447379690557e-06,
+      "loss": 1.3184,
+      "step": 5142
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4191375571257438e-06,
+      "loss": 1.2065,
+      "step": 5143
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.415932208667041e-06,
+      "loss": 1.2197,
+      "step": 5144
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.412728693367723e-06,
+      "loss": 1.145,
+      "step": 5145
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409527012002122e-06,
+      "loss": 1.2104,
+      "step": 5146
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4063271653441147e-06,
+      "loss": 1.3608,
+      "step": 5147
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.403129154167153e-06,
+      "loss": 0.9956,
+      "step": 5148
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3999329792442203e-06,
+      "loss": 1.3257,
+      "step": 5149
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.396738641347881e-06,
+      "loss": 1.1865,
+      "step": 5150
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.393546141250238e-06,
+      "loss": 1.3359,
+      "step": 5151
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3903554797229612e-06,
+      "loss": 1.4463,
+      "step": 5152
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.38716665753727e-06,
+      "loss": 1.3076,
+      "step": 5153
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3839796754639365e-06,
+      "loss": 1.4072,
+      "step": 5154
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3807945342733e-06,
+      "loss": 1.3008,
+      "step": 5155
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3776112347352388e-06,
+      "loss": 1.2261,
+      "step": 5156
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.374429777619205e-06,
+      "loss": 1.3125,
+      "step": 5157
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.371250163694181e-06,
+      "loss": 1.2676,
+      "step": 5158
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3680723937287274e-06,
+      "loss": 1.2959,
+      "step": 5159
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3648964684909427e-06,
+      "loss": 1.3022,
+      "step": 5160
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3617223887484918e-06,
+      "loss": 1.3271,
+      "step": 5161
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358550155268585e-06,
+      "loss": 1.2729,
+      "step": 5162
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.355379768817986e-06,
+      "loss": 1.2524,
+      "step": 5163
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3522112301630195e-06,
+      "loss": 1.2695,
+      "step": 5164
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3490445400695538e-06,
+      "loss": 1.2061,
+      "step": 5165
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.345879699303025e-06,
+      "loss": 1.2314,
+      "step": 5166
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3427167086284007e-06,
+      "loss": 1.2439,
+      "step": 5167
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 1.3213,
+      "step": 5168
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.336396280612566e-06,
+      "loss": 1.3037,
+      "step": 5169
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3332388447990805e-06,
+      "loss": 1.3574,
+      "step": 5170
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3300832621329494e-06,
+      "loss": 1.293,
+      "step": 5171
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3269295333769114e-06,
+      "loss": 1.0913,
+      "step": 5172
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3237776592932683e-06,
+      "loss": 1.2373,
+      "step": 5173
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.320627640643858e-06,
+      "loss": 1.2529,
+      "step": 5174
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3174794781900858e-06,
+      "loss": 1.0659,
+      "step": 5175
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3143331726928895e-06,
+      "loss": 1.252,
+      "step": 5176
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3111887249127773e-06,
+      "loss": 1.2739,
+      "step": 5177
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3080461356097938e-06,
+      "loss": 1.2832,
+      "step": 5178
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.304905405543545e-06,
+      "loss": 1.2002,
+      "step": 5179
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.301766535473181e-06,
+      "loss": 1.3525,
+      "step": 5180
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2986295261574e-06,
+      "loss": 1.1323,
+      "step": 5181
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2954943783544613e-06,
+      "loss": 1.1147,
+      "step": 5182
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.292361092822161e-06,
+      "loss": 1.1396,
+      "step": 5183
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2892296703178596e-06,
+      "loss": 1.228,
+      "step": 5184
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.286100111598449e-06,
+      "loss": 1.3271,
+      "step": 5185
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2829724174203894e-06,
+      "loss": 1.2183,
+      "step": 5186
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.279846588539675e-06,
+      "loss": 1.2905,
+      "step": 5187
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2767226257118613e-06,
+      "loss": 1.1841,
+      "step": 5188
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.273600529692045e-06,
+      "loss": 1.1577,
+      "step": 5189
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2704803012348718e-06,
+      "loss": 1.293,
+      "step": 5190
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2673619410945424e-06,
+      "loss": 1.2217,
+      "step": 5191
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.264245450024796e-06,
+      "loss": 1.1191,
+      "step": 5192
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2611308287789347e-06,
+      "loss": 1.2891,
+      "step": 5193
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2580180781097873e-06,
+      "loss": 1.2231,
+      "step": 5194
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2549071987697525e-06,
+      "loss": 1.2085,
+      "step": 5195
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.25179819151076e-06,
+      "loss": 1.0942,
+      "step": 5196
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2486910570843003e-06,
+      "loss": 1.2339,
+      "step": 5197
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.245585796241401e-06,
+      "loss": 1.3672,
+      "step": 5198
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242482409732638e-06,
+      "loss": 1.3682,
+      "step": 5199
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.239380898308142e-06,
+      "loss": 1.1494,
+      "step": 5200
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.236281262717581e-06,
+      "loss": 1.1694,
+      "step": 5201
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2331835037101825e-06,
+      "loss": 1.2695,
+      "step": 5202
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2300876220346977e-06,
+      "loss": 1.0918,
+      "step": 5203
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226993618439449e-06,
+      "loss": 1.2808,
+      "step": 5204
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223901493672287e-06,
+      "loss": 1.2476,
+      "step": 5205
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.220811248480621e-06,
+      "loss": 1.2671,
+      "step": 5206
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2177228836113973e-06,
+      "loss": 1.1919,
+      "step": 5207
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2146363998111077e-06,
+      "loss": 1.0791,
+      "step": 5208
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211551797825797e-06,
+      "loss": 1.229,
+      "step": 5209
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2084690784010455e-06,
+      "loss": 1.1538,
+      "step": 5210
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2053882422819904e-06,
+      "loss": 1.1665,
+      "step": 5211
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.202309290213297e-06,
+      "loss": 1.1479,
+      "step": 5212
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1992322229391904e-06,
+      "loss": 1.1387,
+      "step": 5213
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1961570412034316e-06,
+      "loss": 1.2334,
+      "step": 5214
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1930837457493324e-06,
+      "loss": 1.1191,
+      "step": 5215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1900123373197435e-06,
+      "loss": 1.0811,
+      "step": 5216
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1869428166570562e-06,
+      "loss": 1.1494,
+      "step": 5217
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.183875184503217e-06,
+      "loss": 1.2334,
+      "step": 5218
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.180809441599703e-06,
+      "loss": 1.1558,
+      "step": 5219
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.17774558868755e-06,
+      "loss": 1.335,
+      "step": 5220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1746836265073145e-06,
+      "loss": 1.1709,
+      "step": 5221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1716235557991194e-06,
+      "loss": 1.2627,
+      "step": 5222
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1685653773026148e-06,
+      "loss": 1.2622,
+      "step": 5223
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.165509091757003e-06,
+      "loss": 1.2119,
+      "step": 5224
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.162454699901022e-06,
+      "loss": 1.1548,
+      "step": 5225
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1594022024729533e-06,
+      "loss": 1.2085,
+      "step": 5226
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1563516002106253e-06,
+      "loss": 1.3218,
+      "step": 5227
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153302893851401e-06,
+      "loss": 1.2451,
+      "step": 5228
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1502560841321963e-06,
+      "loss": 1.1489,
+      "step": 5229
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1472111717894516e-06,
+      "loss": 1.1904,
+      "step": 5230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1441681575591643e-06,
+      "loss": 1.1758,
+      "step": 5231
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1411270421768636e-06,
+      "loss": 1.3203,
+      "step": 5232
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1380878263776285e-06,
+      "loss": 1.2812,
+      "step": 5233
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1350505108960706e-06,
+      "loss": 1.1831,
+      "step": 5234
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1320150964663422e-06,
+      "loss": 1.2837,
+      "step": 5235
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.128981583822145e-06,
+      "loss": 1.3262,
+      "step": 5236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1259499736967093e-06,
+      "loss": 1.2119,
+      "step": 5237
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1229202668228197e-06,
+      "loss": 1.3633,
+      "step": 5238
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.119892463932781e-06,
+      "loss": 1.1724,
+      "step": 5239
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1168665657584585e-06,
+      "loss": 1.3232,
+      "step": 5240
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1138425730312404e-06,
+      "loss": 1.2588,
+      "step": 5241
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.110820486482069e-06,
+      "loss": 1.1738,
+      "step": 5242
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1078003068414154e-06,
+      "loss": 1.1182,
+      "step": 5243
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.10478203483929e-06,
+      "loss": 1.2266,
+      "step": 5244
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1017656712052505e-06,
+      "loss": 1.3281,
+      "step": 5245
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.098751216668381e-06,
+      "loss": 1.2554,
+      "step": 5246
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0957386719573225e-06,
+      "loss": 1.2441,
+      "step": 5247
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.092728037800229e-06,
+      "loss": 1.1914,
+      "step": 5248
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.089719314924815e-06,
+      "loss": 1.314,
+      "step": 5249
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0867125040583215e-06,
+      "loss": 1.3672,
+      "step": 5250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.083707605927533e-06,
+      "loss": 1.2329,
+      "step": 5251
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.080704621258769e-06,
+      "loss": 1.2324,
+      "step": 5252
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.077703550777882e-06,
+      "loss": 1.3281,
+      "step": 5253
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.074704395210272e-06,
+      "loss": 1.1982,
+      "step": 5254
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0717071552808666e-06,
+      "loss": 1.2744,
+      "step": 5255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0687118317141407e-06,
+      "loss": 1.2046,
+      "step": 5256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.06571842523409e-06,
+      "loss": 1.1465,
+      "step": 5257
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.062726936564264e-06,
+      "loss": 1.2402,
+      "step": 5258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0597373664277366e-06,
+      "loss": 1.1309,
+      "step": 5259
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0567497155471263e-06,
+      "loss": 1.3223,
+      "step": 5260
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0537639846445823e-06,
+      "loss": 1.3818,
+      "step": 5261
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0507801744417876e-06,
+      "loss": 1.3975,
+      "step": 5262
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.047798285659971e-06,
+      "loss": 1.2344,
+      "step": 5263
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0448183190198857e-06,
+      "loss": 1.3076,
+      "step": 5264
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0418402752418285e-06,
+      "loss": 1.1729,
+      "step": 5265
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038864155045627e-06,
+      "loss": 1.2646,
+      "step": 5266
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.035889959150643e-06,
+      "loss": 1.2456,
+      "step": 5267
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.032917688275774e-06,
+      "loss": 0.9233,
+      "step": 5268
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0299473431394578e-06,
+      "loss": 1.2402,
+      "step": 5269
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0269789244596607e-06,
+      "loss": 1.3965,
+      "step": 5270
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.02401243295388e-06,
+      "loss": 1.2163,
+      "step": 5271
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.02104786933916e-06,
+      "loss": 1.1577,
+      "step": 5272
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0180852343320634e-06,
+      "loss": 1.2769,
+      "step": 5273
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0151245286486998e-06,
+      "loss": 1.3013,
+      "step": 5274
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0121657530047054e-06,
+      "loss": 1.2065,
+      "step": 5275
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.009208908115251e-06,
+      "loss": 1.0972,
+      "step": 5276
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0062539946950388e-06,
+      "loss": 1.1733,
+      "step": 5277
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0033010134583085e-06,
+      "loss": 1.3804,
+      "step": 5278
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.000349965118836e-06,
+      "loss": 1.1719,
+      "step": 5279
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9974008503899156e-06,
+      "loss": 1.2334,
+      "step": 5280
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9944536699843896e-06,
+      "loss": 1.1802,
+      "step": 5281
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9915084246146223e-06,
+      "loss": 1.207,
+      "step": 5282
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.988565114992519e-06,
+      "loss": 1.2578,
+      "step": 5283
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985623741829511e-06,
+      "loss": 1.2275,
+      "step": 5284
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.982684305836561e-06,
+      "loss": 1.2764,
+      "step": 5285
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9797468077241686e-06,
+      "loss": 1.0356,
+      "step": 5286
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9768112482023596e-06,
+      "loss": 1.1714,
+      "step": 5287
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.973877627980699e-06,
+      "loss": 1.1948,
+      "step": 5288
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.970945947768268e-06,
+      "loss": 1.0171,
+      "step": 5289
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9680162082736975e-06,
+      "loss": 1.2988,
+      "step": 5290
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.965088410205135e-06,
+      "loss": 1.3223,
+      "step": 5291
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962162554270267e-06,
+      "loss": 1.103,
+      "step": 5292
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9592386411763087e-06,
+      "loss": 1.3262,
+      "step": 5293
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9563166716299988e-06,
+      "loss": 1.3262,
+      "step": 5294
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.95339664633762e-06,
+      "loss": 1.3252,
+      "step": 5295
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9504785660049697e-06,
+      "loss": 1.1685,
+      "step": 5296
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.947562431337391e-06,
+      "loss": 1.3164,
+      "step": 5297
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9446482430397383e-06,
+      "loss": 1.1924,
+      "step": 5298
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9417360018164144e-06,
+      "loss": 1.269,
+      "step": 5299
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938825708371336e-06,
+      "loss": 1.1558,
+      "step": 5300
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.935917363407961e-06,
+      "loss": 1.2051,
+      "step": 5301
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9330109676292684e-06,
+      "loss": 1.27,
+      "step": 5302
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9301065217377655e-06,
+      "loss": 1.1201,
+      "step": 5303
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9272040264354985e-06,
+      "loss": 1.1875,
+      "step": 5304
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924303482424028e-06,
+      "loss": 1.1846,
+      "step": 5305
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921404890404459e-06,
+      "loss": 1.1626,
+      "step": 5306
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9185082510774043e-06,
+      "loss": 1.29,
+      "step": 5307
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9156135651430252e-06,
+      "loss": 1.2031,
+      "step": 5308
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.912720833300994e-06,
+      "loss": 1.1768,
+      "step": 5309
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 1.3457,
+      "step": 5310
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9069412346903539e-06,
+      "loss": 1.1787,
+      "step": 5311
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9040543693187352e-06,
+      "loss": 1.2578,
+      "step": 5312
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9011694608334674e-06,
+      "loss": 1.2983,
+      "step": 5313
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8982865099318603e-06,
+      "loss": 1.1462,
+      "step": 5314
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8954055173107665e-06,
+      "loss": 1.2095,
+      "step": 5315
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8925264836665448e-06,
+      "loss": 1.1504,
+      "step": 5316
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.889649409695099e-06,
+      "loss": 1.0732,
+      "step": 5317
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8867742960918489e-06,
+      "loss": 1.2002,
+      "step": 5318
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8839011435517474e-06,
+      "loss": 1.2207,
+      "step": 5319
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8810299527692678e-06,
+      "loss": 1.1968,
+      "step": 5320
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.878160724438408e-06,
+      "loss": 1.1235,
+      "step": 5321
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8752934592527006e-06,
+      "loss": 1.2119,
+      "step": 5322
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8724281579051907e-06,
+      "loss": 1.1694,
+      "step": 5323
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.869564821088462e-06,
+      "loss": 1.2373,
+      "step": 5324
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8667034494946158e-06,
+      "loss": 1.2173,
+      "step": 5325
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.863844043815277e-06,
+      "loss": 1.2104,
+      "step": 5326
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.860986604741598e-06,
+      "loss": 1.1543,
+      "step": 5327
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8581311329642592e-06,
+      "loss": 1.1987,
+      "step": 5328
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8552776291734598e-06,
+      "loss": 1.1587,
+      "step": 5329
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8524260940589235e-06,
+      "loss": 1.2515,
+      "step": 5330
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8495765283099044e-06,
+      "loss": 1.3457,
+      "step": 5331
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8467289326151727e-06,
+      "loss": 1.1885,
+      "step": 5332
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8438833076630313e-06,
+      "loss": 1.167,
+      "step": 5333
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.841039654141298e-06,
+      "loss": 1.2554,
+      "step": 5334
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8381979727373166e-06,
+      "loss": 1.3535,
+      "step": 5335
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8353582641379552e-06,
+      "loss": 1.1323,
+      "step": 5336
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.83252052902961e-06,
+      "loss": 1.2432,
+      "step": 5337
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.829684768098191e-06,
+      "loss": 1.2178,
+      "step": 5338
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8268509820291336e-06,
+      "loss": 1.0996,
+      "step": 5339
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824019171507404e-06,
+      "loss": 1.2944,
+      "step": 5340
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8211893372174772e-06,
+      "loss": 1.0435,
+      "step": 5341
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8183614798433646e-06,
+      "loss": 1.2358,
+      "step": 5342
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8155356000685898e-06,
+      "loss": 1.1909,
+      "step": 5343
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8127116985762006e-06,
+      "loss": 1.1875,
+      "step": 5344
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8098897760487666e-06,
+      "loss": 1.2417,
+      "step": 5345
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8070698331683844e-06,
+      "loss": 1.144,
+      "step": 5346
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8042518706166655e-06,
+      "loss": 1.2505,
+      "step": 5347
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.801435889074743e-06,
+      "loss": 1.188,
+      "step": 5348
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7986218892232765e-06,
+      "loss": 1.1284,
+      "step": 5349
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7958098717424387e-06,
+      "loss": 1.2812,
+      "step": 5350
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7929998373119328e-06,
+      "loss": 1.1816,
+      "step": 5351
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7901917866109764e-06,
+      "loss": 1.248,
+      "step": 5352
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7873857203183077e-06,
+      "loss": 1.0986,
+      "step": 5353
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7845816391121828e-06,
+      "loss": 1.2588,
+      "step": 5354
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7817795436703877e-06,
+      "loss": 1.1968,
+      "step": 5355
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7789794346702205e-06,
+      "loss": 1.2563,
+      "step": 5356
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7761813127884964e-06,
+      "loss": 1.1479,
+      "step": 5357
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.773385178701561e-06,
+      "loss": 1.2739,
+      "step": 5358
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7705910330852672e-06,
+      "loss": 1.1206,
+      "step": 5359
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7677988766149988e-06,
+      "loss": 1.4111,
+      "step": 5360
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.765008709965651e-06,
+      "loss": 1.2373,
+      "step": 5361
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7622205338116404e-06,
+      "loss": 1.291,
+      "step": 5362
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7594343488268983e-06,
+      "loss": 1.2954,
+      "step": 5363
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7566501556848858e-06,
+      "loss": 1.2314,
+      "step": 5364
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7538679550585725e-06,
+      "loss": 1.3105,
+      "step": 5365
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7510877476204457e-06,
+      "loss": 1.2363,
+      "step": 5366
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7483095340425205e-06,
+      "loss": 1.1069,
+      "step": 5367
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7455333149963193e-06,
+      "loss": 1.3213,
+      "step": 5368
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7427590911528924e-06,
+      "loss": 1.2676,
+      "step": 5369
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7399868631827988e-06,
+      "loss": 1.1606,
+      "step": 5370
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.73721663175612e-06,
+      "loss": 1.2812,
+      "step": 5371
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.734448397542452e-06,
+      "loss": 1.2671,
+      "step": 5372
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7316821612109136e-06,
+      "loss": 1.1455,
+      "step": 5373
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7289179234301356e-06,
+      "loss": 1.3164,
+      "step": 5374
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7261556848682647e-06,
+      "loss": 1.0273,
+      "step": 5375
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7233954461929713e-06,
+      "loss": 1.3164,
+      "step": 5376
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7206372080714328e-06,
+      "loss": 1.2368,
+      "step": 5377
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7178809711703525e-06,
+      "loss": 1.2334,
+      "step": 5378
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7151267361559454e-06,
+      "loss": 1.2017,
+      "step": 5379
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7123745036939398e-06,
+      "loss": 1.334,
+      "step": 5380
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.709624274449584e-06,
+      "loss": 1.1826,
+      "step": 5381
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7068760490876425e-06,
+      "loss": 1.2231,
+      "step": 5382
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7041298282723917e-06,
+      "loss": 1.1338,
+      "step": 5383
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7013856126676297e-06,
+      "loss": 1.1616,
+      "step": 5384
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.698643402936665e-06,
+      "loss": 1.3721,
+      "step": 5385
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.695903199742317e-06,
+      "loss": 1.2197,
+      "step": 5386
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6931650037469327e-06,
+      "loss": 1.3477,
+      "step": 5387
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.690428815612364e-06,
+      "loss": 1.3184,
+      "step": 5388
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687694635999979e-06,
+      "loss": 1.1357,
+      "step": 5389
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6849624655706597e-06,
+      "loss": 1.2456,
+      "step": 5390
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6822323049848088e-06,
+      "loss": 1.2217,
+      "step": 5391
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6795041549023338e-06,
+      "loss": 1.3623,
+      "step": 5392
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6767780159826651e-06,
+      "loss": 1.2339,
+      "step": 5393
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6740538888847424e-06,
+      "loss": 1.2358,
+      "step": 5394
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6713317742670166e-06,
+      "loss": 1.248,
+      "step": 5395
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.668611672787459e-06,
+      "loss": 1.1768,
+      "step": 5396
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6658935851035484e-06,
+      "loss": 1.2812,
+      "step": 5397
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6631775118722782e-06,
+      "loss": 1.2822,
+      "step": 5398
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6604634537501552e-06,
+      "loss": 1.3145,
+      "step": 5399
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6577514113932036e-06,
+      "loss": 1.2319,
+      "step": 5400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6550413854569503e-06,
+      "loss": 1.1992,
+      "step": 5401
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6523333765964467e-06,
+      "loss": 1.1567,
+      "step": 5402
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6496273854662482e-06,
+      "loss": 1.2666,
+      "step": 5403
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6469234127204226e-06,
+      "loss": 1.199,
+      "step": 5404
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.644221459012556e-06,
+      "loss": 1.293,
+      "step": 5405
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6415215249957427e-06,
+      "loss": 1.2876,
+      "step": 5406
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6388236113225864e-06,
+      "loss": 1.29,
+      "step": 5407
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6361277186452041e-06,
+      "loss": 1.2061,
+      "step": 5408
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.633433847615229e-06,
+      "loss": 1.1819,
+      "step": 5409
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.630741998883798e-06,
+      "loss": 1.2109,
+      "step": 5410
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6280521731015675e-06,
+      "loss": 1.332,
+      "step": 5411
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6253643709186973e-06,
+      "loss": 1.0884,
+      "step": 5412
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6226785929848588e-06,
+      "loss": 1.3877,
+      "step": 5413
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6199948399492416e-06,
+      "loss": 1.2495,
+      "step": 5414
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6173131124605378e-06,
+      "loss": 1.2793,
+      "step": 5415
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6146334111669548e-06,
+      "loss": 1.2246,
+      "step": 5416
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6119557367162031e-06,
+      "loss": 1.2656,
+      "step": 5417
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.609280089755515e-06,
+      "loss": 1.2808,
+      "step": 5418
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.606606470931621e-06,
+      "loss": 1.2871,
+      "step": 5419
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6039348808907707e-06,
+      "loss": 1.1074,
+      "step": 5420
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6012653202787187e-06,
+      "loss": 1.2217,
+      "step": 5421
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5985977897407245e-06,
+      "loss": 1.1689,
+      "step": 5422
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.595932289921569e-06,
+      "loss": 1.3311,
+      "step": 5423
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5932688214655311e-06,
+      "loss": 1.2363,
+      "step": 5424
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.590607385016405e-06,
+      "loss": 1.0737,
+      "step": 5425
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5879479812174869e-06,
+      "loss": 1.1943,
+      "step": 5426
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5852906107115895e-06,
+      "loss": 1.2651,
+      "step": 5427
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5826352741410333e-06,
+      "loss": 1.1479,
+      "step": 5428
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5799819721476429e-06,
+      "loss": 1.3145,
+      "step": 5429
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5773307053727515e-06,
+      "loss": 1.1523,
+      "step": 5430
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5746814744572013e-06,
+      "loss": 1.1992,
+      "step": 5431
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5720342800413458e-06,
+      "loss": 1.2871,
+      "step": 5432
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5693891227650426e-06,
+      "loss": 1.1514,
+      "step": 5433
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.566746003267655e-06,
+      "loss": 1.3955,
+      "step": 5434
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.56410492218806e-06,
+      "loss": 1.0908,
+      "step": 5435
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5614658801646354e-06,
+      "loss": 1.3223,
+      "step": 5436
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5588288778352724e-06,
+      "loss": 1.2402,
+      "step": 5437
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5561939158373629e-06,
+      "loss": 1.062,
+      "step": 5438
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.55356099480781e-06,
+      "loss": 1.4258,
+      "step": 5439
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5509301153830191e-06,
+      "loss": 1.2559,
+      "step": 5440
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5483012781989105e-06,
+      "loss": 1.0532,
+      "step": 5441
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5456744838908988e-06,
+      "loss": 1.0449,
+      "step": 5442
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5430497330939175e-06,
+      "loss": 1.2832,
+      "step": 5443
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5404270264423982e-06,
+      "loss": 1.1846,
+      "step": 5444
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5378063645702767e-06,
+      "loss": 1.2231,
+      "step": 5445
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5351877481110045e-06,
+      "loss": 1.1211,
+      "step": 5446
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5325711776975283e-06,
+      "loss": 1.0732,
+      "step": 5447
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5299566539623046e-06,
+      "loss": 1.1313,
+      "step": 5448
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5273441775372931e-06,
+      "loss": 1.1577,
+      "step": 5449
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5247337490539648e-06,
+      "loss": 1.1885,
+      "step": 5450
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5221253691432869e-06,
+      "loss": 1.2812,
+      "step": 5451
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5195190384357405e-06,
+      "loss": 1.3066,
+      "step": 5452
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.516914757561304e-06,
+      "loss": 1.2197,
+      "step": 5453
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5143125271494607e-06,
+      "loss": 1.1387,
+      "step": 5454
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5117123478292062e-06,
+      "loss": 1.2207,
+      "step": 5455
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.509114220229031e-06,
+      "loss": 1.2095,
+      "step": 5456
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.506518144976935e-06,
+      "loss": 1.3022,
+      "step": 5457
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5039241227004164e-06,
+      "loss": 0.9666,
+      "step": 5458
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.501332154026488e-06,
+      "loss": 1.4043,
+      "step": 5459
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.498742239581652e-06,
+      "loss": 1.1914,
+      "step": 5460
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4961543799919298e-06,
+      "loss": 1.2695,
+      "step": 5461
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4935685758828322e-06,
+      "loss": 1.1318,
+      "step": 5462
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.490984827879378e-06,
+      "loss": 1.1689,
+      "step": 5463
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4884031366060948e-06,
+      "loss": 1.2324,
+      "step": 5464
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4858235026870049e-06,
+      "loss": 1.2764,
+      "step": 5465
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4832459267456367e-06,
+      "loss": 1.1841,
+      "step": 5466
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4806704094050196e-06,
+      "loss": 1.1782,
+      "step": 5467
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4780969512876898e-06,
+      "loss": 1.2437,
+      "step": 5468
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4755255530156798e-06,
+      "loss": 1.2432,
+      "step": 5469
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4729562152105303e-06,
+      "loss": 1.4014,
+      "step": 5470
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.470388938493279e-06,
+      "loss": 1.2222,
+      "step": 5471
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.467823723484465e-06,
+      "loss": 0.9707,
+      "step": 5472
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4652605708041357e-06,
+      "loss": 1.2534,
+      "step": 5473
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4626994810718342e-06,
+      "loss": 1.0596,
+      "step": 5474
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4601404549066046e-06,
+      "loss": 1.2905,
+      "step": 5475
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4575834929269938e-06,
+      "loss": 1.1938,
+      "step": 5476
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4550285957510524e-06,
+      "loss": 1.1216,
+      "step": 5477
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.452475763996326e-06,
+      "loss": 1.4375,
+      "step": 5478
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4499249982798691e-06,
+      "loss": 1.2764,
+      "step": 5479
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4473762992182294e-06,
+      "loss": 1.1465,
+      "step": 5480
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4448296674274565e-06,
+      "loss": 1.3613,
+      "step": 5481
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4422851035231045e-06,
+      "loss": 1.2695,
+      "step": 5482
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4397426081202225e-06,
+      "loss": 1.0093,
+      "step": 5483
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4372021818333637e-06,
+      "loss": 1.2402,
+      "step": 5484
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4346638252765755e-06,
+      "loss": 1.2598,
+      "step": 5485
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4321275390634149e-06,
+      "loss": 1.1387,
+      "step": 5486
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.429593323806926e-06,
+      "loss": 1.2671,
+      "step": 5487
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4270611801196642e-06,
+      "loss": 1.1372,
+      "step": 5488
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4245311086136771e-06,
+      "loss": 1.2188,
+      "step": 5489
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4220031099005094e-06,
+      "loss": 1.1587,
+      "step": 5490
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4194771845912148e-06,
+      "loss": 1.2314,
+      "step": 5491
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4169533332963349e-06,
+      "loss": 1.333,
+      "step": 5492
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.414431556625917e-06,
+      "loss": 1.2656,
+      "step": 5493
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4119118551895006e-06,
+      "loss": 1.2554,
+      "step": 5494
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4093942295961326e-06,
+      "loss": 1.3135,
+      "step": 5495
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4068786804543489e-06,
+      "loss": 1.2637,
+      "step": 5496
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4043652083721914e-06,
+      "loss": 1.3862,
+      "step": 5497
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4018538139571946e-06,
+      "loss": 1.333,
+      "step": 5498
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3993444978163905e-06,
+      "loss": 1.2173,
+      "step": 5499
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.396837260556314e-06,
+      "loss": 1.0923,
+      "step": 5500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3943321027829893e-06,
+      "loss": 1.2974,
+      "step": 5501
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3918290251019517e-06,
+      "loss": 1.2397,
+      "step": 5502
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3893280281182143e-06,
+      "loss": 1.1982,
+      "step": 5503
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3868291124363043e-06,
+      "loss": 1.3291,
+      "step": 5504
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3843322786602353e-06,
+      "loss": 1.0854,
+      "step": 5505
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.381837527393527e-06,
+      "loss": 1.229,
+      "step": 5506
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3793448592391878e-06,
+      "loss": 1.1831,
+      "step": 5507
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3768542747997215e-06,
+      "loss": 1.2695,
+      "step": 5508
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3743657746771387e-06,
+      "loss": 1.2178,
+      "step": 5509
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.371879359472934e-06,
+      "loss": 1.1582,
+      "step": 5510
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3693950297881098e-06,
+      "loss": 1.3398,
+      "step": 5511
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3669127862231502e-06,
+      "loss": 1.1963,
+      "step": 5512
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3644326293780485e-06,
+      "loss": 1.1816,
+      "step": 5513
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3619545598522844e-06,
+      "loss": 1.2666,
+      "step": 5514
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3594785782448417e-06,
+      "loss": 1.2739,
+      "step": 5515
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3570046851541907e-06,
+      "loss": 1.2158,
+      "step": 5516
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.354532881178301e-06,
+      "loss": 1.1919,
+      "step": 5517
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3520631669146388e-06,
+      "loss": 1.1982,
+      "step": 5518
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3495955429601604e-06,
+      "loss": 1.2549,
+      "step": 5519
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.347130009911327e-06,
+      "loss": 1.1528,
+      "step": 5520
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3446665683640781e-06,
+      "loss": 1.1562,
+      "step": 5521
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3422052189138623e-06,
+      "loss": 1.2739,
+      "step": 5522
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 1.2583,
+      "step": 5523
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3372887986837679e-06,
+      "loss": 1.3008,
+      "step": 5524
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3348337290922486e-06,
+      "loss": 1.2061,
+      "step": 5525
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3323807539744726e-06,
+      "loss": 1.1914,
+      "step": 5526
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.329929873923359e-06,
+      "loss": 1.1475,
+      "step": 5527
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3274810895313083e-06,
+      "loss": 1.1611,
+      "step": 5528
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3250344013902293e-06,
+      "loss": 1.2822,
+      "step": 5529
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3225898100915057e-06,
+      "loss": 1.2412,
+      "step": 5530
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3201473162260303e-06,
+      "loss": 1.1792,
+      "step": 5531
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3177069203841796e-06,
+      "loss": 1.0269,
+      "step": 5532
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.315268623155831e-06,
+      "loss": 1.1836,
+      "step": 5533
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3128324251303458e-06,
+      "loss": 1.3262,
+      "step": 5534
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3103983268965825e-06,
+      "loss": 1.2036,
+      "step": 5535
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3079663290428935e-06,
+      "loss": 1.3003,
+      "step": 5536
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3055364321571196e-06,
+      "loss": 1.333,
+      "step": 5537
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3031086368266e-06,
+      "loss": 1.188,
+      "step": 5538
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3006829436381552e-06,
+      "loss": 1.251,
+      "step": 5539
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2982593531781085e-06,
+      "loss": 1.1021,
+      "step": 5540
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.295837866032268e-06,
+      "loss": 1.2451,
+      "step": 5541
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2934184827859387e-06,
+      "loss": 1.3057,
+      "step": 5542
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.291001204023914e-06,
+      "loss": 1.2178,
+      "step": 5543
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2885860303304743e-06,
+      "loss": 1.2964,
+      "step": 5544
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2861729622894026e-06,
+      "loss": 1.1855,
+      "step": 5545
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2837620004839602e-06,
+      "loss": 1.2222,
+      "step": 5546
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.281353145496912e-06,
+      "loss": 1.2236,
+      "step": 5547
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2789463979104988e-06,
+      "loss": 1.1128,
+      "step": 5548
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2765417583064665e-06,
+      "loss": 1.1875,
+      "step": 5549
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2741392272660413e-06,
+      "loss": 1.2627,
+      "step": 5550
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2717388053699464e-06,
+      "loss": 1.3633,
+      "step": 5551
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2693404931983898e-06,
+      "loss": 1.1772,
+      "step": 5552
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2669442913310725e-06,
+      "loss": 1.2646,
+      "step": 5553
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2645502003471877e-06,
+      "loss": 1.0376,
+      "step": 5554
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2621582208254112e-06,
+      "loss": 1.4258,
+      "step": 5555
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2597683533439198e-06,
+      "loss": 1.1089,
+      "step": 5556
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2573805984803633e-06,
+      "loss": 1.3398,
+      "step": 5557
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.254994956811899e-06,
+      "loss": 1.2876,
+      "step": 5558
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2526114289151581e-06,
+      "loss": 1.2305,
+      "step": 5559
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2502300153662738e-06,
+      "loss": 1.3193,
+      "step": 5560
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2478507167408594e-06,
+      "loss": 1.2466,
+      "step": 5561
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2454735336140167e-06,
+      "loss": 1.1663,
+      "step": 5562
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.243098466560345e-06,
+      "loss": 1.186,
+      "step": 5563
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.24072551615392e-06,
+      "loss": 1.3467,
+      "step": 5564
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2383546829683202e-06,
+      "loss": 1.1973,
+      "step": 5565
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2359859675765952e-06,
+      "loss": 1.158,
+      "step": 5566
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2336193705512978e-06,
+      "loss": 1.1089,
+      "step": 5567
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2312548924644585e-06,
+      "loss": 1.1357,
+      "step": 5568
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2288925338876013e-06,
+      "loss": 1.2466,
+      "step": 5569
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2265322953917413e-06,
+      "loss": 1.2021,
+      "step": 5570
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.224174177547368e-06,
+      "loss": 1.2637,
+      "step": 5571
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2218181809244723e-06,
+      "loss": 1.373,
+      "step": 5572
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2194643060925215e-06,
+      "loss": 1.126,
+      "step": 5573
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2171125536204841e-06,
+      "loss": 1.208,
+      "step": 5574
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2147629240767956e-06,
+      "loss": 1.2876,
+      "step": 5575
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2124154180293946e-06,
+      "loss": 1.0317,
+      "step": 5576
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.210070036045704e-06,
+      "loss": 1.332,
+      "step": 5577
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.207726778692625e-06,
+      "loss": 1.0796,
+      "step": 5578
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2053856465365587e-06,
+      "loss": 1.23,
+      "step": 5579
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2030466401433749e-06,
+      "loss": 1.1782,
+      "step": 5580
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2007097600784457e-06,
+      "loss": 1.2446,
+      "step": 5581
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1983750069066192e-06,
+      "loss": 1.2158,
+      "step": 5582
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1960423811922362e-06,
+      "loss": 1.29,
+      "step": 5583
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.193711883499119e-06,
+      "loss": 1.1968,
+      "step": 5584
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.191383514390574e-06,
+      "loss": 1.1787,
+      "step": 5585
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1890572744294004e-06,
+      "loss": 1.0259,
+      "step": 5586
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1867331641778723e-06,
+      "loss": 1.2173,
+      "step": 5587
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1844111841977635e-06,
+      "loss": 1.1543,
+      "step": 5588
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1820913350503139e-06,
+      "loss": 1.3184,
+      "step": 5589
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1797736172962659e-06,
+      "loss": 1.291,
+      "step": 5590
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.177458031495835e-06,
+      "loss": 1.27,
+      "step": 5591
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.17514457820873e-06,
+      "loss": 1.3232,
+      "step": 5592
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.172833257994138e-06,
+      "loss": 1.1982,
+      "step": 5593
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1705240714107301e-06,
+      "loss": 1.4355,
+      "step": 5594
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1682170190166687e-06,
+      "loss": 1.2295,
+      "step": 5595
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1659121013695917e-06,
+      "loss": 1.1504,
+      "step": 5596
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1636093190266318e-06,
+      "loss": 1.2856,
+      "step": 5597
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.161308672544389e-06,
+      "loss": 1.2612,
+      "step": 5598
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1590101624789652e-06,
+      "loss": 1.0396,
+      "step": 5599
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1567137893859315e-06,
+      "loss": 1.2271,
+      "step": 5600
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.154419553820354e-06,
+      "loss": 1.292,
+      "step": 5601
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1521274563367746e-06,
+      "loss": 1.2764,
+      "step": 5602
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149837497489218e-06,
+      "loss": 1.2627,
+      "step": 5603
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.147549677831199e-06,
+      "loss": 1.3242,
+      "step": 5604
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1452639979157066e-06,
+      "loss": 1.2529,
+      "step": 5605
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.142980458295222e-06,
+      "loss": 1.2998,
+      "step": 5606
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1406990595216971e-06,
+      "loss": 1.2534,
+      "step": 5607
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1384198021465787e-06,
+      "loss": 1.2324,
+      "step": 5608
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.136142686720787e-06,
+      "loss": 1.3574,
+      "step": 5609
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1338677137947307e-06,
+      "loss": 1.2368,
+      "step": 5610
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1315948839182956e-06,
+      "loss": 1.2168,
+      "step": 5611
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.129324197640852e-06,
+      "loss": 1.2168,
+      "step": 5612
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1270556555112534e-06,
+      "loss": 1.1909,
+      "step": 5613
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1247892580778307e-06,
+      "loss": 1.3188,
+      "step": 5614
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1225250058884042e-06,
+      "loss": 1.2686,
+      "step": 5615
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1202628994902632e-06,
+      "loss": 1.1211,
+      "step": 5616
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1180029394301916e-06,
+      "loss": 1.0759,
+      "step": 5617
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.115745126254445e-06,
+      "loss": 1.231,
+      "step": 5618
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1134894605087677e-06,
+      "loss": 1.2109,
+      "step": 5619
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1112359427383779e-06,
+      "loss": 1.1353,
+      "step": 5620
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1089845734879756e-06,
+      "loss": 1.3291,
+      "step": 5621
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.10673535330175e-06,
+      "loss": 1.2734,
+      "step": 5622
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1044882827233573e-06,
+      "loss": 1.2939,
+      "step": 5623
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1022433622959495e-06,
+      "loss": 1.2485,
+      "step": 5624
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1000005925621405e-06,
+      "loss": 1.1743,
+      "step": 5625
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.097759974064042e-06,
+      "loss": 1.1548,
+      "step": 5626
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0955215073432335e-06,
+      "loss": 1.314,
+      "step": 5627
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0932851929407828e-06,
+      "loss": 1.1387,
+      "step": 5628
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0910510313972334e-06,
+      "loss": 1.1924,
+      "step": 5629
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0888190232526042e-06,
+      "loss": 1.1968,
+      "step": 5630
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0865891690464037e-06,
+      "loss": 1.2305,
+      "step": 5631
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0843614693176097e-06,
+      "loss": 1.0903,
+      "step": 5632
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0821359246046893e-06,
+      "loss": 1.2847,
+      "step": 5633
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0799125354455752e-06,
+      "loss": 1.3589,
+      "step": 5634
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0776913023776946e-06,
+      "loss": 1.3047,
+      "step": 5635
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0754722259379402e-06,
+      "loss": 1.2803,
+      "step": 5636
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0732553066626938e-06,
+      "loss": 1.2168,
+      "step": 5637
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710405450878092e-06,
+      "loss": 1.2773,
+      "step": 5638
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0688279417486192e-06,
+      "loss": 1.1689,
+      "step": 5639
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.066617497179938e-06,
+      "loss": 0.9976,
+      "step": 5640
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0644092119160553e-06,
+      "loss": 1.1621,
+      "step": 5641
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.062203086490744e-06,
+      "loss": 1.3701,
+      "step": 5642
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.059999121437244e-06,
+      "loss": 1.2471,
+      "step": 5643
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0577973172882838e-06,
+      "loss": 1.1455,
+      "step": 5644
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0555976745760643e-06,
+      "loss": 1.1812,
+      "step": 5645
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.053400193832269e-06,
+      "loss": 1.105,
+      "step": 5646
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0512048755880511e-06,
+      "loss": 1.2153,
+      "step": 5647
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0490117203740447e-06,
+      "loss": 1.1548,
+      "step": 5648
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0468207287203646e-06,
+      "loss": 1.2236,
+      "step": 5649
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0446319011565964e-06,
+      "loss": 1.165,
+      "step": 5650
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0424452382118122e-06,
+      "loss": 1.1089,
+      "step": 5651
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0402607404145448e-06,
+      "loss": 1.2593,
+      "step": 5652
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0380784082928196e-06,
+      "loss": 1.1191,
+      "step": 5653
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0358982423741303e-06,
+      "loss": 1.1279,
+      "step": 5654
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0337202431854509e-06,
+      "loss": 1.1777,
+      "step": 5655
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.031544411253227e-06,
+      "loss": 1.1216,
+      "step": 5656
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0293707471033843e-06,
+      "loss": 1.272,
+      "step": 5657
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0271992512613239e-06,
+      "loss": 1.1455,
+      "step": 5658
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0250299242519201e-06,
+      "loss": 1.2373,
+      "step": 5659
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0228627665995295e-06,
+      "loss": 1.1611,
+      "step": 5660
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0206977788279737e-06,
+      "loss": 1.1816,
+      "step": 5661
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.018534961460561e-06,
+      "loss": 1.2383,
+      "step": 5662
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0163743150200667e-06,
+      "loss": 1.2539,
+      "step": 5663
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0142158400287473e-06,
+      "loss": 1.1992,
+      "step": 5664
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.012059537008332e-06,
+      "loss": 1.3643,
+      "step": 5665
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0099054064800208e-06,
+      "loss": 1.1343,
+      "step": 5666
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.007753448964498e-06,
+      "loss": 1.1899,
+      "step": 5667
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0056036649819123e-06,
+      "loss": 1.2271,
+      "step": 5668
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.003456055051899e-06,
+      "loss": 1.2046,
+      "step": 5669
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0013106196935528e-06,
+      "loss": 1.3066,
+      "step": 5670
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.99167359425457e-07,
+      "loss": 1.103,
+      "step": 5671
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.970262747656589e-07,
+      "loss": 1.1909,
+      "step": 5672
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.948873662316882e-07,
+      "loss": 1.1729,
+      "step": 5673
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.927506343405412e-07,
+      "loss": 1.2227,
+      "step": 5674
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.906160796086928e-07,
+      "loss": 1.2266,
+      "step": 5675
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.884837025520898e-07,
+      "loss": 1.2554,
+      "step": 5676
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.863535036861526e-07,
+      "loss": 1.2588,
+      "step": 5677
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.842254835257792e-07,
+      "loss": 1.1533,
+      "step": 5678
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.820996425853335e-07,
+      "loss": 1.0884,
+      "step": 5679
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.79975981378658e-07,
+      "loss": 1.1143,
+      "step": 5680
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.778545004190644e-07,
+      "loss": 1.1548,
+      "step": 5681
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.757352002193443e-07,
+      "loss": 1.2275,
+      "step": 5682
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.736180812917551e-07,
+      "loss": 1.2021,
+      "step": 5683
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.715031441480272e-07,
+      "loss": 1.3467,
+      "step": 5684
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.693903892993717e-07,
+      "loss": 1.1016,
+      "step": 5685
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.672798172564623e-07,
+      "loss": 1.1692,
+      "step": 5686
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.651714285294521e-07,
+      "loss": 1.2363,
+      "step": 5687
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.630652236279626e-07,
+      "loss": 1.3447,
+      "step": 5688
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.609612030610882e-07,
+      "loss": 1.2559,
+      "step": 5689
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.588593673373937e-07,
+      "loss": 1.1885,
+      "step": 5690
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.567597169649222e-07,
+      "loss": 1.0981,
+      "step": 5691
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.546622524511816e-07,
+      "loss": 1.2041,
+      "step": 5692
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.525669743031529e-07,
+      "loss": 1.2861,
+      "step": 5693
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.504738830272931e-07,
+      "loss": 1.2144,
+      "step": 5694
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.48382979129523e-07,
+      "loss": 1.3418,
+      "step": 5695
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.462942631152439e-07,
+      "loss": 1.147,
+      "step": 5696
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.442077354893198e-07,
+      "loss": 1.2583,
+      "step": 5697
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.421233967560905e-07,
+      "loss": 1.3013,
+      "step": 5698
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.400412474193643e-07,
+      "loss": 1.1228,
+      "step": 5699
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.379612879824229e-07,
+      "loss": 1.1208,
+      "step": 5700
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.358835189480165e-07,
+      "loss": 1.1265,
+      "step": 5701
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.338079408183654e-07,
+      "loss": 1.2954,
+      "step": 5702
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.31734554095165e-07,
+      "loss": 1.3271,
+      "step": 5703
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.296633592795723e-07,
+      "loss": 1.106,
+      "step": 5704
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.275943568722257e-07,
+      "loss": 1.2993,
+      "step": 5705
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.255275473732239e-07,
+      "loss": 1.2256,
+      "step": 5706
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.234629312821408e-07,
+      "loss": 1.3994,
+      "step": 5707
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.21400509098016e-07,
+      "loss": 1.3105,
+      "step": 5708
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.193402813193652e-07,
+      "loss": 1.3511,
+      "step": 5709
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.17282248444169e-07,
+      "loss": 1.2314,
+      "step": 5710
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.15226410969875e-07,
+      "loss": 1.2749,
+      "step": 5711
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.131727693934089e-07,
+      "loss": 1.0171,
+      "step": 5712
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.111213242111549e-07,
+      "loss": 1.1333,
+      "step": 5713
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.090720759189775e-07,
+      "loss": 1.3975,
+      "step": 5714
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070250250122003e-07,
+      "loss": 1.3232,
+      "step": 5715
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.049801719856211e-07,
+      "loss": 1.2588,
+      "step": 5716
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.029375173335031e-07,
+      "loss": 1.2695,
+      "step": 5717
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.008970615495827e-07,
+      "loss": 1.1436,
+      "step": 5718
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.988588051270641e-07,
+      "loss": 1.3076,
+      "step": 5719
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.968227485586134e-07,
+      "loss": 1.2354,
+      "step": 5720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.947888923363734e-07,
+      "loss": 1.3262,
+      "step": 5721
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.927572369519488e-07,
+      "loss": 1.0928,
+      "step": 5722
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.907277828964168e-07,
+      "loss": 1.1436,
+      "step": 5723
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.887005306603203e-07,
+      "loss": 1.2812,
+      "step": 5724
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.866754807336686e-07,
+      "loss": 1.2471,
+      "step": 5725
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.846526336059425e-07,
+      "loss": 1.3779,
+      "step": 5726
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.826319897660863e-07,
+      "loss": 1.1758,
+      "step": 5727
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.806135497025181e-07,
+      "loss": 1.2847,
+      "step": 5728
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.785973139031112e-07,
+      "loss": 1.1401,
+      "step": 5729
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.765832828552201e-07,
+      "loss": 1.3237,
+      "step": 5730
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.745714570456554e-07,
+      "loss": 1.1587,
+      "step": 5731
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.725618369607047e-07,
+      "loss": 1.0972,
+      "step": 5732
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.705544230861141e-07,
+      "loss": 1.1284,
+      "step": 5733
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.685492159070974e-07,
+      "loss": 1.3208,
+      "step": 5734
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.665462159083415e-07,
+      "loss": 1.3198,
+      "step": 5735
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 1.2329,
+      "step": 5736
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.625468393876646e-07,
+      "loss": 1.3364,
+      "step": 5737
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.605504638324424e-07,
+      "loss": 1.2441,
+      "step": 5738
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.585562973908723e-07,
+      "loss": 1.0723,
+      "step": 5739
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.565643405449664e-07,
+      "loss": 1.1768,
+      "step": 5740
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.545745937762073e-07,
+      "loss": 1.2617,
+      "step": 5741
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.525870575655393e-07,
+      "loss": 1.1987,
+      "step": 5742
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.506017323933713e-07,
+      "loss": 1.3535,
+      "step": 5743
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.48618618739584e-07,
+      "loss": 1.1172,
+      "step": 5744
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.466377170835161e-07,
+      "loss": 1.1992,
+      "step": 5745
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.446590279039768e-07,
+      "loss": 1.2402,
+      "step": 5746
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.426825516792403e-07,
+      "loss": 1.2363,
+      "step": 5747
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.407082888870422e-07,
+      "loss": 1.3887,
+      "step": 5748
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.387362400045828e-07,
+      "loss": 1.2295,
+      "step": 5749
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.367664055085356e-07,
+      "loss": 1.2573,
+      "step": 5750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.347987858750306e-07,
+      "loss": 1.2666,
+      "step": 5751
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.328333815796619e-07,
+      "loss": 1.2114,
+      "step": 5752
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.308701930974949e-07,
+      "loss": 1.2646,
+      "step": 5753
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.289092209030536e-07,
+      "loss": 1.1841,
+      "step": 5754
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.269504654703309e-07,
+      "loss": 1.2979,
+      "step": 5755
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.249939272727781e-07,
+      "loss": 1.146,
+      "step": 5756
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.230396067833157e-07,
+      "loss": 1.3008,
+      "step": 5757
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.210875044743238e-07,
+      "loss": 1.2715,
+      "step": 5758
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.191376208176527e-07,
+      "loss": 1.3438,
+      "step": 5759
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.171899562846097e-07,
+      "loss": 1.3369,
+      "step": 5760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.152445113459673e-07,
+      "loss": 1.2178,
+      "step": 5761
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.133012864719669e-07,
+      "loss": 1.1235,
+      "step": 5762
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.113602821323041e-07,
+      "loss": 1.1074,
+      "step": 5763
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.094214987961468e-07,
+      "loss": 1.1543,
+      "step": 5764
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.074849369321214e-07,
+      "loss": 1.1958,
+      "step": 5765
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.055505970083166e-07,
+      "loss": 1.2124,
+      "step": 5766
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.03618479492283e-07,
+      "loss": 1.2656,
+      "step": 5767
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.016885848510414e-07,
+      "loss": 1.1973,
+      "step": 5768
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.997609135510687e-07,
+      "loss": 1.4268,
+      "step": 5769
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.978354660583032e-07,
+      "loss": 1.0952,
+      "step": 5770
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.959122428381516e-07,
+      "loss": 1.4028,
+      "step": 5771
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.939912443554775e-07,
+      "loss": 1.0991,
+      "step": 5772
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.920724710746119e-07,
+      "loss": 1.2222,
+      "step": 5773
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.901559234593437e-07,
+      "loss": 1.415,
+      "step": 5774
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.882416019729255e-07,
+      "loss": 1.103,
+      "step": 5775
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.863295070780696e-07,
+      "loss": 1.157,
+      "step": 5776
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.844196392369541e-07,
+      "loss": 1.292,
+      "step": 5777
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.825119989112173e-07,
+      "loss": 1.1792,
+      "step": 5778
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.806065865619561e-07,
+      "loss": 1.105,
+      "step": 5779
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.787034026497342e-07,
+      "loss": 0.9563,
+      "step": 5780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.768024476345704e-07,
+      "loss": 1.3706,
+      "step": 5781
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.749037219759514e-07,
+      "loss": 1.2612,
+      "step": 5782
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.730072261328214e-07,
+      "loss": 1.2173,
+      "step": 5783
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.711129605635847e-07,
+      "loss": 1.1438,
+      "step": 5784
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.692209257261074e-07,
+      "loss": 1.2197,
+      "step": 5785
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.67331122077718e-07,
+      "loss": 1.1831,
+      "step": 5786
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.654435500752055e-07,
+      "loss": 1.1538,
+      "step": 5787
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.635582101748141e-07,
+      "loss": 1.2266,
+      "step": 5788
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.616751028322588e-07,
+      "loss": 1.1362,
+      "step": 5789
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.597942285027038e-07,
+      "loss": 1.0837,
+      "step": 5790
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.579155876407828e-07,
+      "loss": 1.2144,
+      "step": 5791
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.560391807005852e-07,
+      "loss": 1.2881,
+      "step": 5792
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.541650081356588e-07,
+      "loss": 1.105,
+      "step": 5793
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.522930703990128e-07,
+      "loss": 1.2651,
+      "step": 5794
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.504233679431194e-07,
+      "loss": 1.3008,
+      "step": 5795
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.485559012199062e-07,
+      "loss": 1.0935,
+      "step": 5796
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.46690670680763e-07,
+      "loss": 1.2017,
+      "step": 5797
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.448276767765394e-07,
+      "loss": 1.0977,
+      "step": 5798
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.429669199575396e-07,
+      "loss": 1.2334,
+      "step": 5799
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.411084006735337e-07,
+      "loss": 1.2832,
+      "step": 5800
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.392521193737479e-07,
+      "loss": 1.2573,
+      "step": 5801
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.373980765068667e-07,
+      "loss": 1.144,
+      "step": 5802
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.355462725210316e-07,
+      "loss": 1.1748,
+      "step": 5803
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.336967078638513e-07,
+      "loss": 1.3027,
+      "step": 5804
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.318493829823814e-07,
+      "loss": 1.1465,
+      "step": 5805
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.300042983231481e-07,
+      "loss": 1.1421,
+      "step": 5806
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.281614543321269e-07,
+      "loss": 1.2056,
+      "step": 5807
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.263208514547548e-07,
+      "loss": 1.1948,
+      "step": 5808
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.244824901359293e-07,
+      "loss": 1.2812,
+      "step": 5809
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.226463708200037e-07,
+      "loss": 1.1812,
+      "step": 5810
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.208124939507899e-07,
+      "loss": 1.3008,
+      "step": 5811
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.189808599715553e-07,
+      "loss": 1.3008,
+      "step": 5812
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.171514693250314e-07,
+      "loss": 1.2979,
+      "step": 5813
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.153243224534001e-07,
+      "loss": 1.2432,
+      "step": 5814
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.13499419798308e-07,
+      "loss": 1.2031,
+      "step": 5815
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.116767618008547e-07,
+      "loss": 1.2959,
+      "step": 5816
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.098563489015952e-07,
+      "loss": 1.1533,
+      "step": 5817
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.080381815405501e-07,
+      "loss": 1.2251,
+      "step": 5818
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.0622226015719e-07,
+      "loss": 1.1724,
+      "step": 5819
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.044085851904425e-07,
+      "loss": 1.1621,
+      "step": 5820
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.02597157078696e-07,
+      "loss": 1.1611,
+      "step": 5821
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.007879762597947e-07,
+      "loss": 1.1782,
+      "step": 5822
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.989810431710375e-07,
+      "loss": 1.3398,
+      "step": 5823
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.971763582491853e-07,
+      "loss": 1.1997,
+      "step": 5824
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.95373921930449e-07,
+      "loss": 1.272,
+      "step": 5825
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.935737346504967e-07,
+      "loss": 1.3496,
+      "step": 5826
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.917757968444605e-07,
+      "loss": 1.2363,
+      "step": 5827
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.899801089469205e-07,
+      "loss": 1.1428,
+      "step": 5828
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.881866713919161e-07,
+      "loss": 1.0562,
+      "step": 5829
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.863954846129394e-07,
+      "loss": 1.2378,
+      "step": 5830
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.846065490429477e-07,
+      "loss": 1.188,
+      "step": 5831
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.828198651143425e-07,
+      "loss": 1.2109,
+      "step": 5832
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.810354332589897e-07,
+      "loss": 1.1221,
+      "step": 5833
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.792532539082075e-07,
+      "loss": 1.1475,
+      "step": 5834
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.774733274927681e-07,
+      "loss": 1.1851,
+      "step": 5835
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.756956544429028e-07,
+      "loss": 1.1763,
+      "step": 5836
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.739202351882957e-07,
+      "loss": 1.1206,
+      "step": 5837
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.721470701580856e-07,
+      "loss": 1.1733,
+      "step": 5838
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.703761597808678e-07,
+      "loss": 1.3164,
+      "step": 5839
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.68607504484694e-07,
+      "loss": 1.228,
+      "step": 5840
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.668411046970679e-07,
+      "loss": 1.1841,
+      "step": 5841
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.650769608449504e-07,
+      "loss": 1.1963,
+      "step": 5842
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.633150733547555e-07,
+      "loss": 1.2979,
+      "step": 5843
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.615554426523507e-07,
+      "loss": 1.189,
+      "step": 5844
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.597980691630635e-07,
+      "loss": 1.334,
+      "step": 5845
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.580429533116694e-07,
+      "loss": 1.0859,
+      "step": 5846
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.562900955224005e-07,
+      "loss": 1.1572,
+      "step": 5847
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.545394962189422e-07,
+      "loss": 1.1484,
+      "step": 5848
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.527911558244393e-07,
+      "loss": 1.2852,
+      "step": 5849
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.510450747614816e-07,
+      "loss": 1.1533,
+      "step": 5850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.493012534521225e-07,
+      "loss": 1.2993,
+      "step": 5851
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.475596923178618e-07,
+      "loss": 1.2241,
+      "step": 5852
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.458203917796546e-07,
+      "loss": 1.1655,
+      "step": 5853
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.440833522579126e-07,
+      "loss": 1.1235,
+      "step": 5854
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.423485741724967e-07,
+      "loss": 1.3809,
+      "step": 5855
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.406160579427268e-07,
+      "loss": 1.2139,
+      "step": 5856
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.388858039873691e-07,
+      "loss": 1.3433,
+      "step": 5857
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.371578127246481e-07,
+      "loss": 1.2515,
+      "step": 5858
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.354320845722395e-07,
+      "loss": 1.3096,
+      "step": 5859
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.33708619947273e-07,
+      "loss": 1.1294,
+      "step": 5860
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.3198741926633e-07,
+      "loss": 1.2671,
+      "step": 5861
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.302684829454441e-07,
+      "loss": 1.3438,
+      "step": 5862
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.285518114001066e-07,
+      "loss": 1.2686,
+      "step": 5863
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.268374050452509e-07,
+      "loss": 1.1562,
+      "step": 5864
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.251252642952776e-07,
+      "loss": 1.3281,
+      "step": 5865
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.234153895640238e-07,
+      "loss": 1.3145,
+      "step": 5866
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.217077812647887e-07,
+      "loss": 1.3945,
+      "step": 5867
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.200024398103255e-07,
+      "loss": 1.249,
+      "step": 5868
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.182993656128323e-07,
+      "loss": 1.3198,
+      "step": 5869
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.165985590839618e-07,
+      "loss": 1.3369,
+      "step": 5870
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.149000206348199e-07,
+      "loss": 1.2979,
+      "step": 5871
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.132037506759636e-07,
+      "loss": 1.2388,
+      "step": 5872
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.115097496174016e-07,
+      "loss": 1.0742,
+      "step": 5873
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.098180178685953e-07,
+      "loss": 1.3052,
+      "step": 5874
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.081285558384553e-07,
+      "loss": 1.2446,
+      "step": 5875
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.064413639353428e-07,
+      "loss": 1.2769,
+      "step": 5876
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.047564425670749e-07,
+      "loss": 1.2534,
+      "step": 5877
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 1.314,
+      "step": 5878
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.013934130635845e-07,
+      "loss": 1.0986,
+      "step": 5879
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.997153057412441e-07,
+      "loss": 1.1543,
+      "step": 5880
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.980394705795167e-07,
+      "loss": 1.3154,
+      "step": 5881
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.963659079834694e-07,
+      "loss": 1.2817,
+      "step": 5882
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.946946183576241e-07,
+      "loss": 1.2305,
+      "step": 5883
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.930256021059522e-07,
+      "loss": 1.3076,
+      "step": 5884
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.913588596318698e-07,
+      "loss": 1.3135,
+      "step": 5885
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.896943913382547e-07,
+      "loss": 1.1689,
+      "step": 5886
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.880321976274251e-07,
+      "loss": 1.3096,
+      "step": 5887
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.863722789011528e-07,
+      "loss": 1.2188,
+      "step": 5888
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.847146355606604e-07,
+      "loss": 1.27,
+      "step": 5889
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.830592680066205e-07,
+      "loss": 1.3511,
+      "step": 5890
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.814061766391543e-07,
+      "loss": 1.2197,
+      "step": 5891
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.797553618578356e-07,
+      "loss": 1.2314,
+      "step": 5892
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.78106824061685e-07,
+      "loss": 1.2676,
+      "step": 5893
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.764605636491715e-07,
+      "loss": 1.3701,
+      "step": 5894
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.748165810182183e-07,
+      "loss": 1.2866,
+      "step": 5895
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.73174876566196e-07,
+      "loss": 1.2607,
+      "step": 5896
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.715354506899228e-07,
+      "loss": 1.0562,
+      "step": 5897
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.698983037856665e-07,
+      "loss": 1.1396,
+      "step": 5898
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.682634362491469e-07,
+      "loss": 1.3291,
+      "step": 5899
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.666308484755289e-07,
+      "loss": 1.3428,
+      "step": 5900
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.650005408594306e-07,
+      "loss": 1.2432,
+      "step": 5901
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.633725137949164e-07,
+      "loss": 1.251,
+      "step": 5902
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.617467676754973e-07,
+      "loss": 1.4541,
+      "step": 5903
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.601233028941389e-07,
+      "loss": 1.189,
+      "step": 5904
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.585021198432505e-07,
+      "loss": 1.375,
+      "step": 5905
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.568832189146911e-07,
+      "loss": 1.3721,
+      "step": 5906
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.552666004997665e-07,
+      "loss": 1.2378,
+      "step": 5907
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.536522649892373e-07,
+      "loss": 1.3184,
+      "step": 5908
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.520402127733015e-07,
+      "loss": 1.2217,
+      "step": 5909
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.504304442416175e-07,
+      "loss": 1.3135,
+      "step": 5910
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.488229597832818e-07,
+      "loss": 1.1372,
+      "step": 5911
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.472177597868422e-07,
+      "loss": 1.1826,
+      "step": 5912
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.456148446402976e-07,
+      "loss": 1.1528,
+      "step": 5913
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.440142147310878e-07,
+      "loss": 1.2456,
+      "step": 5914
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.424158704461091e-07,
+      "loss": 1.1812,
+      "step": 5915
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.408198121716946e-07,
+      "loss": 1.0986,
+      "step": 5916
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.392260402936334e-07,
+      "loss": 1.2053,
+      "step": 5917
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.376345551971585e-07,
+      "loss": 1.2061,
+      "step": 5918
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.360453572669533e-07,
+      "loss": 1.3154,
+      "step": 5919
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.344584468871428e-07,
+      "loss": 1.1108,
+      "step": 5920
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.328738244413023e-07,
+      "loss": 1.1528,
+      "step": 5921
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.312914903124566e-07,
+      "loss": 1.0791,
+      "step": 5922
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.297114448830709e-07,
+      "loss": 1.1665,
+      "step": 5923
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.281336885350652e-07,
+      "loss": 1.0845,
+      "step": 5924
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.265582216497966e-07,
+      "loss": 1.01,
+      "step": 5925
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.249850446080796e-07,
+      "loss": 1.2275,
+      "step": 5926
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.234141577901652e-07,
+      "loss": 1.1484,
+      "step": 5927
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.218455615757601e-07,
+      "loss": 1.1113,
+      "step": 5928
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.202792563440096e-07,
+      "loss": 1.1714,
+      "step": 5929
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.187152424735065e-07,
+      "loss": 1.2192,
+      "step": 5930
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.171535203422961e-07,
+      "loss": 1.2178,
+      "step": 5931
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.155940903278622e-07,
+      "loss": 1.2495,
+      "step": 5932
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.14036952807141e-07,
+      "loss": 1.2114,
+      "step": 5933
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.124821081565057e-07,
+      "loss": 1.3047,
+      "step": 5934
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.109295567517846e-07,
+      "loss": 1.1772,
+      "step": 5935
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.093792989682467e-07,
+      "loss": 1.2578,
+      "step": 5936
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078313351806086e-07,
+      "loss": 1.1523,
+      "step": 5937
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.062856657630311e-07,
+      "loss": 1.2144,
+      "step": 5938
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.047422910891187e-07,
+      "loss": 1.4131,
+      "step": 5939
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.032012115319273e-07,
+      "loss": 1.3535,
+      "step": 5940
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.016624274639503e-07,
+      "loss": 1.3345,
+      "step": 5941
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.001259392571355e-07,
+      "loss": 1.2993,
+      "step": 5942
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.985917472828639e-07,
+      "loss": 1.2988,
+      "step": 5943
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.970598519119718e-07,
+      "loss": 1.2207,
+      "step": 5944
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.955302535147355e-07,
+      "loss": 1.186,
+      "step": 5945
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.94002952460878e-07,
+      "loss": 1.2451,
+      "step": 5946
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.924779491195653e-07,
+      "loss": 1.2998,
+      "step": 5947
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.909552438594079e-07,
+      "loss": 1.1763,
+      "step": 5948
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 1.1865,
+      "step": 5949
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.87916729054232e-07,
+      "loss": 1.3262,
+      "step": 5950
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.864009202436603e-07,
+      "loss": 1.1797,
+      "step": 5951
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.848874109831325e-07,
+      "loss": 1.3286,
+      "step": 5952
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833762016384858e-07,
+      "loss": 1.0776,
+      "step": 5953
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.818672925749946e-07,
+      "loss": 1.2407,
+      "step": 5954
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.803606841573827e-07,
+      "loss": 1.1729,
+      "step": 5955
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.788563767498144e-07,
+      "loss": 1.3169,
+      "step": 5956
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.773543707158979e-07,
+      "loss": 1.3467,
+      "step": 5957
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7585466641868696e-07,
+      "loss": 1.105,
+      "step": 5958
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7435726422067617e-07,
+      "loss": 1.2915,
+      "step": 5959
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.728621644838105e-07,
+      "loss": 1.2881,
+      "step": 5960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.713693675694664e-07,
+      "loss": 0.9971,
+      "step": 5961
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6987887383847517e-07,
+      "loss": 1.3433,
+      "step": 5962
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.683906836511043e-07,
+      "loss": 1.1235,
+      "step": 5963
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6690479736707037e-07,
+      "loss": 1.165,
+      "step": 5964
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.654212153455273e-07,
+      "loss": 1.1543,
+      "step": 5965
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.639399379450737e-07,
+      "loss": 1.3193,
+      "step": 5966
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6246096552375444e-07,
+      "loss": 1.2769,
+      "step": 5967
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6098429843905224e-07,
+      "loss": 1.1313,
+      "step": 5968
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.595099370478995e-07,
+      "loss": 1.2202,
+      "step": 5969
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.580378817066611e-07,
+      "loss": 1.3926,
+      "step": 5970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5656813277115353e-07,
+      "loss": 1.2998,
+      "step": 5971
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5510069059663154e-07,
+      "loss": 1.2593,
+      "step": 5972
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.536355555377947e-07,
+      "loss": 1.1899,
+      "step": 5973
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.521727279487831e-07,
+      "loss": 1.3701,
+      "step": 5974
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.507122081831772e-07,
+      "loss": 1.1504,
+      "step": 5975
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.492539965940057e-07,
+      "loss": 1.231,
+      "step": 5976
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.477980935337334e-07,
+      "loss": 1.3013,
+      "step": 5977
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4634449935427203e-07,
+      "loss": 1.2617,
+      "step": 5978
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4489321440696844e-07,
+      "loss": 1.0344,
+      "step": 5979
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.434442390426186e-07,
+      "loss": 1.2383,
+      "step": 5980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4199757361145455e-07,
+      "loss": 1.1699,
+      "step": 5981
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.405532184631556e-07,
+      "loss": 1.1787,
+      "step": 5982
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3911117394683787e-07,
+      "loss": 1.2637,
+      "step": 5983
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.376714404110594e-07,
+      "loss": 1.165,
+      "step": 5984
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.362340182038238e-07,
+      "loss": 1.2651,
+      "step": 5985
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3479890767257006e-07,
+      "loss": 1.1543,
+      "step": 5986
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.333661091641861e-07,
+      "loss": 1.1606,
+      "step": 5987
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.319356230249905e-07,
+      "loss": 1.2012,
+      "step": 5988
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.305074496007533e-07,
+      "loss": 1.1602,
+      "step": 5989
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2908158923667733e-07,
+      "loss": 1.1982,
+      "step": 5990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2765804227741345e-07,
+      "loss": 1.2139,
+      "step": 5991
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.262368090670488e-07,
+      "loss": 1.2461,
+      "step": 5992
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2481788994910956e-07,
+      "loss": 1.2607,
+      "step": 5993
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.234012852665703e-07,
+      "loss": 1.2832,
+      "step": 5994
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2198699536183716e-07,
+      "loss": 1.0737,
+      "step": 5995
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.205750205767645e-07,
+      "loss": 1.0333,
+      "step": 5996
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.191653612526381e-07,
+      "loss": 1.2188,
+      "step": 5997
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.177580177301954e-07,
+      "loss": 1.1841,
+      "step": 5998
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.163529903496022e-07,
+      "loss": 1.2583,
+      "step": 5999
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.149502794504745e-07,
+      "loss": 1.3281,
+      "step": 6000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1354988537186557e-07,
+      "loss": 1.2173,
+      "step": 6001
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.121518084522613e-07,
+      "loss": 1.1475,
+      "step": 6002
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.107560490295992e-07,
+      "loss": 1.2061,
+      "step": 6003
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0936260744124714e-07,
+      "loss": 1.2798,
+      "step": 6004
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.079714840240201e-07,
+      "loss": 1.3809,
+      "step": 6005
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.065826791141658e-07,
+      "loss": 1.1479,
+      "step": 6006
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.051961930473769e-07,
+      "loss": 1.2856,
+      "step": 6007
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.038120261587819e-07,
+      "loss": 1.1021,
+      "step": 6008
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.02430178782951e-07,
+      "loss": 1.3027,
+      "step": 6009
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0105065125389584e-07,
+      "loss": 1.3623,
+      "step": 6010
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.996734439050609e-07,
+      "loss": 1.3545,
+      "step": 6011
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.982985570693354e-07,
+      "loss": 1.2617,
+      "step": 6012
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.969259910790446e-07,
+      "loss": 1.3135,
+      "step": 6013
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.955557462659576e-07,
+      "loss": 1.0928,
+      "step": 6014
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9418782296127276e-07,
+      "loss": 1.2241,
+      "step": 6015
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.928222214956379e-07,
+      "loss": 1.1528,
+      "step": 6016
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.914589421991344e-07,
+      "loss": 1.1274,
+      "step": 6017
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.900979854012821e-07,
+      "loss": 1.1631,
+      "step": 6018
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8873935143104337e-07,
+      "loss": 1.3105,
+      "step": 6019
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8738304061681107e-07,
+      "loss": 1.0693,
+      "step": 6020
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.860290532864264e-07,
+      "loss": 1.207,
+      "step": 6021
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8467738976716074e-07,
+      "loss": 1.0752,
+      "step": 6022
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.833280503857306e-07,
+      "loss": 1.1396,
+      "step": 6023
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.819810354682851e-07,
+      "loss": 1.3799,
+      "step": 6024
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.806363453404127e-07,
+      "loss": 1.0498,
+      "step": 6025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.792939803271445e-07,
+      "loss": 1.3223,
+      "step": 6026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.77953940752942e-07,
+      "loss": 1.1377,
+      "step": 6027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7661622694171393e-07,
+      "loss": 1.1895,
+      "step": 6028
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7528083921679616e-07,
+      "loss": 1.3721,
+      "step": 6029
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.739477779009704e-07,
+      "loss": 1.2705,
+      "step": 6030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7261704331645334e-07,
+      "loss": 1.1924,
+      "step": 6031
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.712886357848988e-07,
+      "loss": 1.2393,
+      "step": 6032
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.699625556274e-07,
+      "loss": 1.2607,
+      "step": 6033
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6863880316448384e-07,
+      "loss": 1.3276,
+      "step": 6034
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6731737871611993e-07,
+      "loss": 1.3164,
+      "step": 6035
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6599828260171054e-07,
+      "loss": 1.2573,
+      "step": 6036
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.646815151400984e-07,
+      "loss": 1.2983,
+      "step": 6037
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6336707664955895e-07,
+      "loss": 1.0898,
+      "step": 6038
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6205496744781023e-07,
+      "loss": 1.0142,
+      "step": 6039
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6074518785200186e-07,
+      "loss": 1.2998,
+      "step": 6040
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.594377381787273e-07,
+      "loss": 1.2139,
+      "step": 6041
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.581326187440104e-07,
+      "loss": 1.1416,
+      "step": 6042
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5682982986331214e-07,
+      "loss": 1.2412,
+      "step": 6043
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5552937185153625e-07,
+      "loss": 1.1089,
+      "step": 6044
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5423124502301455e-07,
+      "loss": 1.3223,
+      "step": 6045
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5293544969152494e-07,
+      "loss": 1.0732,
+      "step": 6046
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5164198617027024e-07,
+      "loss": 1.2622,
+      "step": 6047
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5035085477190143e-07,
+      "loss": 1.248,
+      "step": 6048
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4906205580849673e-07,
+      "loss": 1.3525,
+      "step": 6049
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4777558959157575e-07,
+      "loss": 1.2085,
+      "step": 6050
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.464914564320931e-07,
+      "loss": 1.3311,
+      "step": 6051
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4520965664043725e-07,
+      "loss": 1.2861,
+      "step": 6052
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4393019052643694e-07,
+      "loss": 0.9707,
+      "step": 6053
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4265305839935257e-07,
+      "loss": 1.2334,
+      "step": 6054
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.413782605678839e-07,
+      "loss": 1.1731,
+      "step": 6055
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.401057973401611e-07,
+      "loss": 1.2778,
+      "step": 6056
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.388356690237582e-07,
+      "loss": 1.1299,
+      "step": 6057
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.375678759256762e-07,
+      "loss": 1.0991,
+      "step": 6058
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.363024183523589e-07,
+      "loss": 1.3208,
+      "step": 6059
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.350392966096816e-07,
+      "loss": 1.2959,
+      "step": 6060
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3377851100295346e-07,
+      "loss": 1.1851,
+      "step": 6061
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3252006183692506e-07,
+      "loss": 1.2036,
+      "step": 6062
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3126394941577655e-07,
+      "loss": 1.2432,
+      "step": 6063
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3001017404312607e-07,
+      "loss": 1.3691,
+      "step": 6064
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2875873602202456e-07,
+      "loss": 1.3486,
+      "step": 6065
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.275096356549612e-07,
+      "loss": 1.2427,
+      "step": 6066
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2626287324385666e-07,
+      "loss": 1.1997,
+      "step": 6067
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2501844909007096e-07,
+      "loss": 1.2256,
+      "step": 6068
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.237763634943936e-07,
+      "loss": 1.2471,
+      "step": 6069
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.225366167570509e-07,
+      "loss": 1.3691,
+      "step": 6070
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2129920917770787e-07,
+      "loss": 1.2476,
+      "step": 6071
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.2006414105545727e-07,
+      "loss": 1.249,
+      "step": 6072
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1883141268883365e-07,
+      "loss": 1.2637,
+      "step": 6073
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1760102437579654e-07,
+      "loss": 1.1206,
+      "step": 6074
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.163729764137502e-07,
+      "loss": 1.1509,
+      "step": 6075
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.15147269099525e-07,
+      "loss": 1.145,
+      "step": 6076
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.139239027293917e-07,
+      "loss": 1.3701,
+      "step": 6077
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.127028775990515e-07,
+      "loss": 1.0955,
+      "step": 6078
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.114841940036395e-07,
+      "loss": 1.1802,
+      "step": 6079
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1026785223772893e-07,
+      "loss": 1.1682,
+      "step": 6080
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.090538525953202e-07,
+      "loss": 1.2603,
+      "step": 6081
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0784219536985625e-07,
+      "loss": 1.2256,
+      "step": 6082
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0663288085420404e-07,
+      "loss": 1.1748,
+      "step": 6083
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.054259093406731e-07,
+      "loss": 1.1243,
+      "step": 6084
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.042212811209999e-07,
+      "loss": 1.1802,
+      "step": 6085
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0301899648635947e-07,
+      "loss": 1.2295,
+      "step": 6086
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0181905572735815e-07,
+      "loss": 1.248,
+      "step": 6087
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0062145913403396e-07,
+      "loss": 1.4229,
+      "step": 6088
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.994262069958631e-07,
+      "loss": 1.2271,
+      "step": 6089
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.98233299601749e-07,
+      "loss": 1.1533,
+      "step": 6090
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.970427372400353e-07,
+      "loss": 1.1782,
+      "step": 6091
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9585452019849304e-07,
+      "loss": 1.2734,
+      "step": 6092
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.94668648764328e-07,
+      "loss": 1.3389,
+      "step": 6093
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9348512322417866e-07,
+      "loss": 1.0444,
+      "step": 6094
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.923039438641195e-07,
+      "loss": 1.3711,
+      "step": 6095
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9112511096965556e-07,
+      "loss": 1.2212,
+      "step": 6096
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.899486248257211e-07,
+      "loss": 1.2842,
+      "step": 6097
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8877448571669186e-07,
+      "loss": 1.2842,
+      "step": 6098
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.876026939263676e-07,
+      "loss": 1.1724,
+      "step": 6099
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8643324973798623e-07,
+      "loss": 1.0894,
+      "step": 6100
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.85266153434216e-07,
+      "loss": 1.2129,
+      "step": 6101
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8410140529715803e-07,
+      "loss": 1.2271,
+      "step": 6102
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.829390056083436e-07,
+      "loss": 1.1309,
+      "step": 6103
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.817789546487415e-07,
+      "loss": 1.2124,
+      "step": 6104
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.806212526987484e-07,
+      "loss": 1.1855,
+      "step": 6105
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7946590003819494e-07,
+      "loss": 1.2754,
+      "step": 6106
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7831289694634447e-07,
+      "loss": 1.1748,
+      "step": 6107
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7716224370188836e-07,
+      "loss": 1.1997,
+      "step": 6108
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.760139405829576e-07,
+      "loss": 1.2666,
+      "step": 6109
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.74867987867109e-07,
+      "loss": 1.1953,
+      "step": 6110
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.737243858313321e-07,
+      "loss": 1.2217,
+      "step": 6111
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.72583134752048e-07,
+      "loss": 1.1191,
+      "step": 6112
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7144423490511384e-07,
+      "loss": 1.2363,
+      "step": 6113
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7030768656581384e-07,
+      "loss": 1.2104,
+      "step": 6114
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6917349000886275e-07,
+      "loss": 1.2373,
+      "step": 6115
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.680416455084145e-07,
+      "loss": 1.292,
+      "step": 6116
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.669121533380448e-07,
+      "loss": 1.123,
+      "step": 6117
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.657850137707674e-07,
+      "loss": 1.2144,
+      "step": 6118
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.646602270790255e-07,
+      "loss": 1.2656,
+      "step": 6119
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6353779353469387e-07,
+      "loss": 1.3198,
+      "step": 6120
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.624177134090755e-07,
+      "loss": 1.147,
+      "step": 6121
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.612999869729094e-07,
+      "loss": 1.3076,
+      "step": 6122
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.601846144963627e-07,
+      "loss": 1.1743,
+      "step": 6123
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.590715962490331e-07,
+      "loss": 1.2261,
+      "step": 6124
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5796093249995215e-07,
+      "loss": 1.085,
+      "step": 6125
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5685262351757944e-07,
+      "loss": 1.2974,
+      "step": 6126
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.557466695698074e-07,
+      "loss": 1.0284,
+      "step": 6127
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.546430709239578e-07,
+      "loss": 0.9685,
+      "step": 6128
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.535418278467838e-07,
+      "loss": 1.2393,
+      "step": 6129
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.524429406044659e-07,
+      "loss": 1.2803,
+      "step": 6130
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.513464094626228e-07,
+      "loss": 1.2075,
+      "step": 6131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5025223468629787e-07,
+      "loss": 1.0435,
+      "step": 6132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4916041653996303e-07,
+      "loss": 1.147,
+      "step": 6133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4807095528752824e-07,
+      "loss": 1.1553,
+      "step": 6134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.46983851192325e-07,
+      "loss": 1.0845,
+      "step": 6135
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.458991045171233e-07,
+      "loss": 1.1338,
+      "step": 6136
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4481671552411655e-07,
+      "loss": 1.2954,
+      "step": 6137
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4373668447493225e-07,
+      "loss": 1.3555,
+      "step": 6138
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4265901163062597e-07,
+      "loss": 1.23,
+      "step": 6139
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.41583697251685e-07,
+      "loss": 1.2778,
+      "step": 6140
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4051074159802566e-07,
+      "loss": 1.1528,
+      "step": 6141
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3944014492899183e-07,
+      "loss": 1.1904,
+      "step": 6142
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3837190750336192e-07,
+      "loss": 1.2632,
+      "step": 6143
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3730602957934057e-07,
+      "loss": 1.333,
+      "step": 6144
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.36242511414565e-07,
+      "loss": 1.3018,
+      "step": 6145
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3518135326609736e-07,
+      "loss": 1.3311,
+      "step": 6146
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.341225553904336e-07,
+      "loss": 1.2197,
+      "step": 6147
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3306611804349788e-07,
+      "loss": 1.2764,
+      "step": 6148
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3201204148064262e-07,
+      "loss": 1.186,
+      "step": 6149
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3096032595665176e-07,
+      "loss": 1.2539,
+      "step": 6150
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.299109717257375e-07,
+      "loss": 1.1726,
+      "step": 6151
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2886397904154144e-07,
+      "loss": 1.3149,
+      "step": 6152
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2781934815713224e-07,
+      "loss": 1.1611,
+      "step": 6153
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2677707932501235e-07,
+      "loss": 1.2842,
+      "step": 6154
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2573717279710916e-07,
+      "loss": 1.3516,
+      "step": 6155
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2469962882478046e-07,
+      "loss": 1.1675,
+      "step": 6156
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2366444765881234e-07,
+      "loss": 1.1777,
+      "step": 6157
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.226316295494202e-07,
+      "loss": 1.2241,
+      "step": 6158
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2160117474625098e-07,
+      "loss": 1.29,
+      "step": 6159
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2057308349837548e-07,
+      "loss": 1.2783,
+      "step": 6160
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.19547356054296e-07,
+      "loss": 1.1523,
+      "step": 6161
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 1.2607,
+      "step": 6162
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1750299356867787e-07,
+      "loss": 1.3232,
+      "step": 6163
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1648435902128506e-07,
+      "loss": 1.2808,
+      "step": 6164
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1546808926598106e-07,
+      "loss": 1.1406,
+      "step": 6165
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.144541845484127e-07,
+      "loss": 1.0527,
+      "step": 6166
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1344264511364953e-07,
+      "loss": 1.2705,
+      "step": 6167
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1243347120619595e-07,
+      "loss": 1.2129,
+      "step": 6168
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1142666306997906e-07,
+      "loss": 1.3711,
+      "step": 6169
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1042222094835752e-07,
+      "loss": 1.1885,
+      "step": 6170
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.09420145084116e-07,
+      "loss": 1.3027,
+      "step": 6171
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0842043571946968e-07,
+      "loss": 1.0205,
+      "step": 6172
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0742309309605858e-07,
+      "loss": 1.1802,
+      "step": 6173
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0642811745495206e-07,
+      "loss": 1.1167,
+      "step": 6174
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0543550903664887e-07,
+      "loss": 1.1235,
+      "step": 6175
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0444526808107378e-07,
+      "loss": 1.2529,
+      "step": 6176
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0345739482757977e-07,
+      "loss": 1.2329,
+      "step": 6177
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.02471889514948e-07,
+      "loss": 1.1724,
+      "step": 6178
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0148875238138687e-07,
+      "loss": 1.188,
+      "step": 6179
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0050798366453073e-07,
+      "loss": 1.1748,
+      "step": 6180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9952958360144436e-07,
+      "loss": 1.2568,
+      "step": 6181
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.985535524286186e-07,
+      "loss": 1.1968,
+      "step": 6182
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9757989038197146e-07,
+      "loss": 1.2607,
+      "step": 6183
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9660859769685015e-07,
+      "loss": 1.1548,
+      "step": 6184
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9563967460802469e-07,
+      "loss": 1.1631,
+      "step": 6185
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9467312134969774e-07,
+      "loss": 1.3955,
+      "step": 6186
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9370893815549573e-07,
+      "loss": 1.1729,
+      "step": 6187
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.927471252584745e-07,
+      "loss": 1.3027,
+      "step": 6188
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.917876828911136e-07,
+      "loss": 1.2505,
+      "step": 6189
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.908306112853231e-07,
+      "loss": 1.0459,
+      "step": 6190
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8987591067243793e-07,
+      "loss": 0.9819,
+      "step": 6191
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.889235812832202e-07,
+      "loss": 1.2246,
+      "step": 6192
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8797362334786129e-07,
+      "loss": 1.1123,
+      "step": 6193
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8702603709597423e-07,
+      "loss": 1.2446,
+      "step": 6194
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8608082275660466e-07,
+      "loss": 0.9893,
+      "step": 6195
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8513798055822206e-07,
+      "loss": 1.1406,
+      "step": 6196
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8419751072872083e-07,
+      "loss": 1.1919,
+      "step": 6197
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8325941349542354e-07,
+      "loss": 1.0869,
+      "step": 6198
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8232368908508101e-07,
+      "loss": 1.3652,
+      "step": 6199
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8139033772386794e-07,
+      "loss": 1.1646,
+      "step": 6200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8045935963738715e-07,
+      "loss": 1.1382,
+      "step": 6201
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7953075505066642e-07,
+      "loss": 1.1763,
+      "step": 6202
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7860452418816178e-07,
+      "loss": 1.2402,
+      "step": 6203
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.77680667273753e-07,
+      "loss": 1.2881,
+      "step": 6204
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7675918453074704e-07,
+      "loss": 1.2842,
+      "step": 6205
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7584007618187905e-07,
+      "loss": 1.1196,
+      "step": 6206
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7492334244930575e-07,
+      "loss": 1.1689,
+      "step": 6207
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.740089835546155e-07,
+      "loss": 1.1177,
+      "step": 6208
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.73096999718817e-07,
+      "loss": 1.0532,
+      "step": 6209
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7218739116235061e-07,
+      "loss": 1.311,
+      "step": 6210
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7128015810507714e-07,
+      "loss": 1.2676,
+      "step": 6211
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.703753007662856e-07,
+      "loss": 1.3945,
+      "step": 6212
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6947281936469218e-07,
+      "loss": 1.1299,
+      "step": 6213
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6857271411843678e-07,
+      "loss": 1.1792,
+      "step": 6214
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.676749852450854e-07,
+      "loss": 1.1699,
+      "step": 6215
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.667796329616289e-07,
+      "loss": 1.2778,
+      "step": 6216
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.658866574844864e-07,
+      "loss": 1.1904,
+      "step": 6217
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6499605902949856e-07,
+      "loss": 1.2837,
+      "step": 6218
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.641078378119365e-07,
+      "loss": 1.27,
+      "step": 6219
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6322199404649076e-07,
+      "loss": 1.1763,
+      "step": 6220
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6233852794728222e-07,
+      "loss": 1.0654,
+      "step": 6221
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6145743972785567e-07,
+      "loss": 1.2578,
+      "step": 6222
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6057872960117848e-07,
+      "loss": 1.2993,
+      "step": 6223
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5970239777964747e-07,
+      "loss": 1.2036,
+      "step": 6224
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5882844447507984e-07,
+      "loss": 1.2627,
+      "step": 6225
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5795686989872328e-07,
+      "loss": 1.25,
+      "step": 6226
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5708767426124594e-07,
+      "loss": 1.1201,
+      "step": 6227
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.562208577727442e-07,
+      "loss": 1.1562,
+      "step": 6228
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5535642064273714e-07,
+      "loss": 1.1963,
+      "step": 6229
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5449436308016762e-07,
+      "loss": 1.1387,
+      "step": 6230
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5363468529340896e-07,
+      "loss": 1.1919,
+      "step": 6231
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5277738749025385e-07,
+      "loss": 1.2051,
+      "step": 6232
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 1.2266,
+      "step": 6233
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.510699326630516e-07,
+      "loss": 1.2349,
+      "step": 6234
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5021977605171879e-07,
+      "loss": 1.0981,
+      "step": 6235
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4937200024941366e-07,
+      "loss": 1.2998,
+      "step": 6236
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4852660546105236e-07,
+      "loss": 1.2871,
+      "step": 6237
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4768359189097915e-07,
+      "loss": 1.248,
+      "step": 6238
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.468429597429588e-07,
+      "loss": 1.2656,
+      "step": 6239
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4600470922018218e-07,
+      "loss": 1.2354,
+      "step": 6240
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4516884052526604e-07,
+      "loss": 1.1812,
+      "step": 6241
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4433535386024765e-07,
+      "loss": 1.3213,
+      "step": 6242
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4350424942659146e-07,
+      "loss": 1.334,
+      "step": 6243
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4267552742518564e-07,
+      "loss": 1.1504,
+      "step": 6244
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4184918805634107e-07,
+      "loss": 1.1924,
+      "step": 6245
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4102523151979574e-07,
+      "loss": 1.2212,
+      "step": 6246
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4020365801470815e-07,
+      "loss": 1.2246,
+      "step": 6247
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3938446773966273e-07,
+      "loss": 1.0781,
+      "step": 6248
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.385676608926678e-07,
+      "loss": 1.1899,
+      "step": 6249
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3775323767115544e-07,
+      "loss": 1.2852,
+      "step": 6250
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.369411982719815e-07,
+      "loss": 1.2285,
+      "step": 6251
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3613154289142337e-07,
+      "loss": 1.3867,
+      "step": 6252
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3532427172518792e-07,
+      "loss": 1.2505,
+      "step": 6253
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.34519384968399e-07,
+      "loss": 1.207,
+      "step": 6254
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.337168828156099e-07,
+      "loss": 1.2427,
+      "step": 6255
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3291676546079436e-07,
+      "loss": 1.3076,
+      "step": 6256
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3211903309734874e-07,
+      "loss": 1.2026,
+      "step": 6257
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3132368591809553e-07,
+      "loss": 1.167,
+      "step": 6258
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3053072411527978e-07,
+      "loss": 1.2827,
+      "step": 6259
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2974014788057042e-07,
+      "loss": 1.0881,
+      "step": 6260
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2895195740505682e-07,
+      "loss": 1.2109,
+      "step": 6261
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.281661528792555e-07,
+      "loss": 1.2793,
+      "step": 6262
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2738273449310445e-07,
+      "loss": 1.1812,
+      "step": 6263
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.266017024359656e-07,
+      "loss": 1.3574,
+      "step": 6264
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2582305689662455e-07,
+      "loss": 1.2949,
+      "step": 6265
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2504679806328633e-07,
+      "loss": 1.1216,
+      "step": 6266
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2427292612358532e-07,
+      "loss": 1.2983,
+      "step": 6267
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2350144126457296e-07,
+      "loss": 1.3291,
+      "step": 6268
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2273234367272902e-07,
+      "loss": 1.0964,
+      "step": 6269
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2196563353395031e-07,
+      "loss": 1.1987,
+      "step": 6270
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2120131103356192e-07,
+      "loss": 1.2695,
+      "step": 6271
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2043937635630832e-07,
+      "loss": 1.166,
+      "step": 6272
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1967982968635994e-07,
+      "loss": 1.2427,
+      "step": 6273
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1892267120730661e-07,
+      "loss": 1.1709,
+      "step": 6274
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1816790110216303e-07,
+      "loss": 1.2119,
+      "step": 6275
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1741551955336661e-07,
+      "loss": 1.1245,
+      "step": 6276
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1666552674277632e-07,
+      "loss": 1.3872,
+      "step": 6277
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1591792285167603e-07,
+      "loss": 1.1909,
+      "step": 6278
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1517270806076786e-07,
+      "loss": 1.1938,
+      "step": 6279
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1442988255018106e-07,
+      "loss": 1.2002,
+      "step": 6280
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1368944649946312e-07,
+      "loss": 1.0864,
+      "step": 6281
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1295140008758864e-07,
+      "loss": 1.1899,
+      "step": 6282
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1221574349295161e-07,
+      "loss": 1.1699,
+      "step": 6283
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1148247689336867e-07,
+      "loss": 1.0972,
+      "step": 6284
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1075160046607914e-07,
+      "loss": 1.3232,
+      "step": 6285
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1002311438774394e-07,
+      "loss": 1.1621,
+      "step": 6286
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0929701883444776e-07,
+      "loss": 1.3428,
+      "step": 6287
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0857331398169579e-07,
+      "loss": 1.1665,
+      "step": 6288
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0785200000441698e-07,
+      "loss": 1.0986,
+      "step": 6289
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0713307707696075e-07,
+      "loss": 1.2046,
+      "step": 6290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0641654537310031e-07,
+      "loss": 1.1299,
+      "step": 6291
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0570240506602825e-07,
+      "loss": 1.311,
+      "step": 6292
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0499065632836202e-07,
+      "loss": 1.0884,
+      "step": 6293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0428129933213849e-07,
+      "loss": 1.1846,
+      "step": 6294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0357433424881935e-07,
+      "loss": 1.2598,
+      "step": 6295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0286976124928683e-07,
+      "loss": 1.2529,
+      "step": 6296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0216758050384245e-07,
+      "loss": 1.2446,
+      "step": 6297
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0146779218221159e-07,
+      "loss": 1.1606,
+      "step": 6298
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0077039645354447e-07,
+      "loss": 1.1584,
+      "step": 6299
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0007539348640738e-07,
+      "loss": 1.2754,
+      "step": 6300
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.938278344879038e-08,
+      "loss": 1.3047,
+      "step": 6301
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.869256650810732e-08,
+      "loss": 1.3271,
+      "step": 6302
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.800474283119143e-08,
+      "loss": 1.2817,
+      "step": 6303
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.731931258429638e-08,
+      "loss": 1.1089,
+      "step": 6304
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.663627593310077e-08,
+      "loss": 1.2954,
+      "step": 6305
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.59556330427025e-08,
+      "loss": 1.0149,
+      "step": 6306
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.527738407762e-08,
+      "loss": 1.1875,
+      "step": 6307
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.460152920179432e-08,
+      "loss": 1.2344,
+      "step": 6308
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.392806857858816e-08,
+      "loss": 1.2217,
+      "step": 6309
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.325700237078462e-08,
+      "loss": 1.3477,
+      "step": 6310
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.258833074058726e-08,
+      "loss": 1.2607,
+      "step": 6311
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.19220538496246e-08,
+      "loss": 1.0483,
+      "step": 6312
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.125817185894115e-08,
+      "loss": 1.2017,
+      "step": 6313
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.059668492900741e-08,
+      "loss": 1.0674,
+      "step": 6314
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.993759321970996e-08,
+      "loss": 1.1533,
+      "step": 6315
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.928089689036134e-08,
+      "loss": 1.2764,
+      "step": 6316
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.862659609969237e-08,
+      "loss": 1.3633,
+      "step": 6317
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.797469100585432e-08,
+      "loss": 1.3564,
+      "step": 6318
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.732518176642113e-08,
+      "loss": 1.2212,
+      "step": 6319
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.667806853838723e-08,
+      "loss": 1.2822,
+      "step": 6320
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.603335147816638e-08,
+      "loss": 1.2148,
+      "step": 6321
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.539103074159616e-08,
+      "loss": 1.145,
+      "step": 6322
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.475110648393126e-08,
+      "loss": 1.1748,
+      "step": 6323
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.411357885985128e-08,
+      "loss": 1.2549,
+      "step": 6324
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.347844802345184e-08,
+      "loss": 1.165,
+      "step": 6325
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.284571412825238e-08,
+      "loss": 1.2124,
+      "step": 6326
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.221537732719276e-08,
+      "loss": 1.2471,
+      "step": 6327
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.158743777263334e-08,
+      "loss": 1.1313,
+      "step": 6328
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.09618956163527e-08,
+      "loss": 1.2607,
+      "step": 6329
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.033875100955323e-08,
+      "loss": 1.2231,
+      "step": 6330
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.971800410285557e-08,
+      "loss": 1.229,
+      "step": 6331
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.909965504630301e-08,
+      "loss": 1.229,
+      "step": 6332
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.848370398935711e-08,
+      "loss": 1.2568,
+      "step": 6333
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.78701510808999e-08,
+      "loss": 1.0532,
+      "step": 6334
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.725899646923607e-08,
+      "loss": 1.2842,
+      "step": 6335
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.665024030208635e-08,
+      "loss": 1.2114,
+      "step": 6336
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.604388272659746e-08,
+      "loss": 1.2656,
+      "step": 6337
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.543992388932997e-08,
+      "loss": 1.1812,
+      "step": 6338
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.483836393627041e-08,
+      "loss": 1.2148,
+      "step": 6339
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.42392030128214e-08,
+      "loss": 1.1943,
+      "step": 6340
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.364244126380926e-08,
+      "loss": 1.0876,
+      "step": 6341
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.304807883347642e-08,
+      "loss": 1.2803,
+      "step": 6342
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.245611586548795e-08,
+      "loss": 1.1178,
+      "step": 6343
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.186655250292939e-08,
+      "loss": 1.2969,
+      "step": 6344
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.127938888830344e-08,
+      "loss": 1.3315,
+      "step": 6345
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.069462516353653e-08,
+      "loss": 1.2412,
+      "step": 6346
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.011226146997008e-08,
+      "loss": 1.2554,
+      "step": 6347
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.953229794837146e-08,
+      "loss": 1.1755,
+      "step": 6348
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.895473473892301e-08,
+      "loss": 1.166,
+      "step": 6349
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.837957198122858e-08,
+      "loss": 1.1323,
+      "step": 6350
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.780680981431364e-08,
+      "loss": 1.1704,
+      "step": 6351
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.723644837661858e-08,
+      "loss": 1.2778,
+      "step": 6352
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.666848780600866e-08,
+      "loss": 1.3184,
+      "step": 6353
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.610292823976628e-08,
+      "loss": 1.2227,
+      "step": 6354
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.553976981459542e-08,
+      "loss": 1.1973,
+      "step": 6355
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.497901266661499e-08,
+      "loss": 1.3691,
+      "step": 6356
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.442065693136879e-08,
+      "loss": 1.0894,
+      "step": 6357
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.386470274381662e-08,
+      "loss": 1.2861,
+      "step": 6358
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.33111502383421e-08,
+      "loss": 1.3389,
+      "step": 6359
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.275999954874268e-08,
+      "loss": 1.1899,
+      "step": 6360
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.221125080823842e-08,
+      "loss": 1.2646,
+      "step": 6361
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.16649041494688e-08,
+      "loss": 1.2329,
+      "step": 6362
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.112095970449261e-08,
+      "loss": 1.2163,
+      "step": 6363
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.057941760478803e-08,
+      "loss": 1.187,
+      "step": 6364
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.004027798125144e-08,
+      "loss": 1.0981,
+      "step": 6365
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.9503540964199746e-08,
+      "loss": 1.1445,
+      "step": 6366
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.896920668336692e-08,
+      "loss": 1.1938,
+      "step": 6367
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8437275267910765e-08,
+      "loss": 1.085,
+      "step": 6368
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.79077468464051e-08,
+      "loss": 1.2251,
+      "step": 6369
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7380621546840875e-08,
+      "loss": 1.2686,
+      "step": 6370
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.685589949663173e-08,
+      "loss": 1.2061,
+      "step": 6371
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.633358082260953e-08,
+      "loss": 1.5264,
+      "step": 6372
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.581366565102553e-08,
+      "loss": 1.3252,
+      "step": 6373
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5296154107548075e-08,
+      "loss": 1.0962,
+      "step": 6374
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 1.3062,
+      "step": 6375
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.426834240468859e-08,
+      "loss": 1.1079,
+      "step": 6376
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.375804249374117e-08,
+      "loss": 1.2358,
+      "step": 6377
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.325014670776951e-08,
+      "loss": 1.2695,
+      "step": 6378
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.274465516953653e-08,
+      "loss": 1.1782,
+      "step": 6379
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.2241568001228926e-08,
+      "loss": 1.1641,
+      "step": 6380
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.17408853244461e-08,
+      "loss": 1.2539,
+      "step": 6381
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.124260726021013e-08,
+      "loss": 1.0737,
+      "step": 6382
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.074673392896023e-08,
+      "loss": 1.1782,
+      "step": 6383
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.025326545055498e-08,
+      "loss": 1.1299,
+      "step": 6384
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9762201944272284e-08,
+      "loss": 1.166,
+      "step": 6385
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.92735435288072e-08,
+      "loss": 1.1831,
+      "step": 6386
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.878729032227414e-08,
+      "loss": 1.0771,
+      "step": 6387
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.830344244220686e-08,
+      "loss": 1.2852,
+      "step": 6388
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7822000005557366e-08,
+      "loss": 1.2803,
+      "step": 6389
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.73429631286948e-08,
+      "loss": 1.291,
+      "step": 6390
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.686633192741097e-08,
+      "loss": 1.2085,
+      "step": 6391
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6392106516909284e-08,
+      "loss": 1.3242,
+      "step": 6392
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5920287011818054e-08,
+      "loss": 1.1875,
+      "step": 6393
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.54508735261816e-08,
+      "loss": 1.2461,
+      "step": 6394
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.498386617346251e-08,
+      "loss": 1.209,
+      "step": 6395
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.451926506654158e-08,
+      "loss": 1.2373,
+      "step": 6396
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4057070317719e-08,
+      "loss": 1.1997,
+      "step": 6397
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.359728203871205e-08,
+      "loss": 1.3525,
+      "step": 6398
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3139900340657404e-08,
+      "loss": 1.1455,
+      "step": 6399
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.268492533411106e-08,
+      "loss": 1.272,
+      "step": 6400
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.223235712904283e-08,
+      "loss": 1.4326,
+      "step": 6401
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.178219583484633e-08,
+      "loss": 1.1816,
+      "step": 6402
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1334441560330065e-08,
+      "loss": 1.2319,
+      "step": 6403
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.088909441372191e-08,
+      "loss": 1.2979,
+      "step": 6404
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0446154502666865e-08,
+      "loss": 1.1919,
+      "step": 6405
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.00056219342293e-08,
+      "loss": 1.3687,
+      "step": 6406
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.956749681489069e-08,
+      "loss": 1.2441,
+      "step": 6407
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.913177925055189e-08,
+      "loss": 1.1543,
+      "step": 6408
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8698469346530874e-08,
+      "loss": 1.3281,
+      "step": 6409
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.826756720756275e-08,
+      "loss": 1.168,
+      "step": 6410
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.783907293780309e-08,
+      "loss": 1.1426,
+      "step": 6411
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.741298664082238e-08,
+      "loss": 1.1865,
+      "step": 6412
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.698930841961268e-08,
+      "loss": 1.1567,
+      "step": 6413
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.656803837658096e-08,
+      "loss": 1.1687,
+      "step": 6414
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.614917661355355e-08,
+      "loss": 1.2632,
+      "step": 6415
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5732723231773904e-08,
+      "loss": 1.332,
+      "step": 6416
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5318678331904835e-08,
+      "loss": 1.1558,
+      "step": 6417
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.490704201402406e-08,
+      "loss": 1.3506,
+      "step": 6418
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.449781437763089e-08,
+      "loss": 1.2427,
+      "step": 6419
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.409099552163842e-08,
+      "loss": 1.1543,
+      "step": 6420
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.368658554438131e-08,
+      "loss": 1.2886,
+      "step": 6421
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.328458454361028e-08,
+      "loss": 1.1387,
+      "step": 6422
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.288499261649314e-08,
+      "loss": 1.2183,
+      "step": 6423
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2487809859615974e-08,
+      "loss": 1.3096,
+      "step": 6424
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.209303636898309e-08,
+      "loss": 1.1514,
+      "step": 6425
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.170067224001483e-08,
+      "loss": 1.3115,
+      "step": 6426
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1310717567552e-08,
+      "loss": 1.2827,
+      "step": 6427
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.092317244584919e-08,
+      "loss": 1.1587,
+      "step": 6428
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0538036968582596e-08,
+      "loss": 1.3027,
+      "step": 6429
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.015531122884219e-08,
+      "loss": 1.2119,
+      "step": 6430
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.977499531913952e-08,
+      "loss": 1.2612,
+      "step": 6431
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9397089331399954e-08,
+      "loss": 1.1465,
+      "step": 6432
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.902159335696819e-08,
+      "loss": 1.3848,
+      "step": 6433
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.864850748660608e-08,
+      "loss": 1.0923,
+      "step": 6434
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8277831810492594e-08,
+      "loss": 1.3057,
+      "step": 6435
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7909566418224953e-08,
+      "loss": 1.2812,
+      "step": 6436
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7543711398817508e-08,
+      "loss": 1.1436,
+      "step": 6437
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.718026684070063e-08,
+      "loss": 1.2188,
+      "step": 6438
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6819232831724052e-08,
+      "loss": 1.2607,
+      "step": 6439
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.646060945915463e-08,
+      "loss": 1.1411,
+      "step": 6440
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6104396809675248e-08,
+      "loss": 1.2876,
+      "step": 6441
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5750594969385923e-08,
+      "loss": 1.2163,
+      "step": 6442
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.539920402380491e-08,
+      "loss": 1.2124,
+      "step": 6443
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5050224057868723e-08,
+      "loss": 1.2207,
+      "step": 6444
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4703655155929872e-08,
+      "loss": 1.1484,
+      "step": 6445
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4359497401758026e-08,
+      "loss": 1.3652,
+      "step": 6446
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.401775087853886e-08,
+      "loss": 1.0991,
+      "step": 6447
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3678415668878518e-08,
+      "loss": 1.124,
+      "step": 6448
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.334149185479584e-08,
+      "loss": 1.0459,
+      "step": 6449
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.300697951773234e-08,
+      "loss": 1.1528,
+      "step": 6450
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2674878738542238e-08,
+      "loss": 1.1816,
+      "step": 6451
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2345189597497984e-08,
+      "loss": 1.2842,
+      "step": 6452
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.201791217428917e-08,
+      "loss": 1.1191,
+      "step": 6453
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1693046548023623e-08,
+      "loss": 1.2817,
+      "step": 6454
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1370592797224087e-08,
+      "loss": 1.2515,
+      "step": 6455
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1050550999832665e-08,
+      "loss": 1.2231,
+      "step": 6456
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.073292123320636e-08,
+      "loss": 1.3486,
+      "step": 6457
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.041770357412043e-08,
+      "loss": 1.2354,
+      "step": 6458
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0104898098766147e-08,
+      "loss": 1.1694,
+      "step": 6459
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.979450488275303e-08,
+      "loss": 1.2612,
+      "step": 6460
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9486524001107732e-08,
+      "loss": 1.2344,
+      "step": 6461
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9180955528270706e-08,
+      "loss": 1.3301,
+      "step": 6462
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.887779953810287e-08,
+      "loss": 1.1465,
+      "step": 6463
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8577056103880054e-08,
+      "loss": 1.0864,
+      "step": 6464
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.827872529829522e-08,
+      "loss": 1.1519,
+      "step": 6465
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7982807193460682e-08,
+      "loss": 1.2207,
+      "step": 6466
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7689301860900343e-08,
+      "loss": 1.3232,
+      "step": 6467
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.739820937155967e-08,
+      "loss": 1.229,
+      "step": 6468
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7109529795799052e-08,
+      "loss": 1.2832,
+      "step": 6469
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.682326320339489e-08,
+      "loss": 1.3501,
+      "step": 6470
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6539409663542948e-08,
+      "loss": 1.2461,
+      "step": 6471
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6257969244851678e-08,
+      "loss": 1.332,
+      "step": 6472
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5978942015351107e-08,
+      "loss": 0.9834,
+      "step": 6473
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.570232804248284e-08,
+      "loss": 1.1953,
+      "step": 6474
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.542812739311006e-08,
+      "loss": 1.0229,
+      "step": 6475
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.515634013350864e-08,
+      "loss": 1.2681,
+      "step": 6476
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.488696632937492e-08,
+      "loss": 1.2407,
+      "step": 6477
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4620006045816814e-08,
+      "loss": 1.2959,
+      "step": 6478
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.435545934736382e-08,
+      "loss": 1.1357,
+      "step": 6479
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4093326297959232e-08,
+      "loss": 1.1885,
+      "step": 6480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3833606960964585e-08,
+      "loss": 1.2661,
+      "step": 6481
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3576301399156332e-08,
+      "loss": 1.1216,
+      "step": 6482
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3321409674728058e-08,
+      "loss": 1.2964,
+      "step": 6483
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3068931849291588e-08,
+      "loss": 0.9451,
+      "step": 6484
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2818867983871442e-08,
+      "loss": 1.064,
+      "step": 6485
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.257121813891371e-08,
+      "loss": 1.0796,
+      "step": 6486
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.232598237427607e-08,
+      "loss": 1.1143,
+      "step": 6487
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2083160749236655e-08,
+      "loss": 1.3096,
+      "step": 6488
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.18427533224863e-08,
+      "loss": 0.896,
+      "step": 6489
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1604760152136296e-08,
+      "loss": 1.4033,
+      "step": 6490
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.136918129571174e-08,
+      "loss": 1.3252,
+      "step": 6491
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.113601681015486e-08,
+      "loss": 1.3174,
+      "step": 6492
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0905266751825017e-08,
+      "loss": 1.167,
+      "step": 6493
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0676931176495375e-08,
+      "loss": 1.2427,
+      "step": 6494
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0451010139358453e-08,
+      "loss": 1.2949,
+      "step": 6495
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.022750369502168e-08,
+      "loss": 1.083,
+      "step": 6496
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0006411897510727e-08,
+      "loss": 1.1172,
+      "step": 6497
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.78773480026396e-09,
+      "loss": 1.21,
+      "step": 6498
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.571472456139097e-09,
+      "loss": 1.292,
+      "step": 6499
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.35762491741099e-09,
+      "loss": 1.0664,
+      "step": 6500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.146192235766072e-09,
+      "loss": 1.2197,
+      "step": 6501
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.937174462312348e-09,
+      "loss": 1.2686,
+      "step": 6502
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.730571647570518e-09,
+      "loss": 1.2002,
+      "step": 6503
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.526383841480634e-09,
+      "loss": 1.2207,
+      "step": 6504
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.324611093395441e-09,
+      "loss": 1.1855,
+      "step": 6505
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.125253452087033e-09,
+      "loss": 1.2935,
+      "step": 6506
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.928310965742426e-09,
+      "loss": 1.1606,
+      "step": 6507
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.733783681964646e-09,
+      "loss": 1.0908,
+      "step": 6508
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.541671647772753e-09,
+      "loss": 1.1841,
+      "step": 6509
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.351974909602932e-09,
+      "loss": 1.2124,
+      "step": 6510
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.164693513308507e-09,
+      "loss": 1.1475,
+      "step": 6511
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.97982750415549e-09,
+      "loss": 1.1624,
+      "step": 6512
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.7973769268292464e-09,
+      "loss": 1.2207,
+      "step": 6513
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.617341825430057e-09,
+      "loss": 1.0967,
+      "step": 6514
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.4397222434753326e-09,
+      "loss": 1.293,
+      "step": 6515
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.264518223896288e-09,
+      "loss": 1.3076,
+      "step": 6516
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.091729809042379e-09,
+      "loss": 1.1631,
+      "step": 6517
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.921357040680198e-09,
+      "loss": 1.187,
+      "step": 6518
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.753399959987915e-09,
+      "loss": 1.1665,
+      "step": 6519
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.587858607565278e-09,
+      "loss": 1.4277,
+      "step": 6520
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.424733023424722e-09,
+      "loss": 1.311,
+      "step": 6521
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.26402324699582e-09,
+      "loss": 1.2559,
+      "step": 6522
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.105729317123054e-09,
+      "loss": 1.2378,
+      "step": 6523
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.949851272069151e-09,
+      "loss": 1.1523,
+      "step": 6524
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.796389149511749e-09,
+      "loss": 1.1792,
+      "step": 6525
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6453429865434e-09,
+      "loss": 1.1973,
+      "step": 6526
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.496712819673788e-09,
+      "loss": 1.1716,
+      "step": 6527
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.3504986848297295e-09,
+      "loss": 1.2754,
+      "step": 6528
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.2067006173529545e-09,
+      "loss": 1.1753,
+      "step": 6529
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0653186520001055e-09,
+      "loss": 1.2114,
+      "step": 6530
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.926352822944957e-09,
+      "loss": 1.2251,
+      "step": 6531
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.789803163778416e-09,
+      "loss": 1.3516,
+      "step": 6532
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6556697075051937e-09,
+      "loss": 1.3428,
+      "step": 6533
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5239524865460227e-09,
+      "loss": 1.1196,
+      "step": 6534
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.394651532740989e-09,
+      "loss": 1.208,
+      "step": 6535
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2677668773406502e-09,
+      "loss": 1.293,
+      "step": 6536
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1432985510171377e-09,
+      "loss": 1.2437,
+      "step": 6537
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.021246583855275e-09,
+      "loss": 1.332,
+      "step": 6538
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9016110053559086e-09,
+      "loss": 1.293,
+      "step": 6539
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.7843918444359075e-09,
+      "loss": 1.2148,
+      "step": 6540
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6695891294292732e-09,
+      "loss": 1.0613,
+      "step": 6541
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.55720288808603e-09,
+      "loss": 1.293,
+      "step": 6542
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.447233147570005e-09,
+      "loss": 1.2314,
+      "step": 6543
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.339679934462158e-09,
+      "loss": 1.2036,
+      "step": 6544
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.234543274760581e-09,
+      "loss": 1.251,
+      "step": 6545
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1318231938760593e-09,
+      "loss": 1.2158,
+      "step": 6546
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.03151971663984e-09,
+      "loss": 1.3237,
+      "step": 6547
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9336328672947544e-09,
+      "loss": 1.3096,
+      "step": 6548
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8381626695007647e-09,
+      "loss": 1.0698,
+      "step": 6549
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.745109146336077e-09,
+      "loss": 1.0942,
+      "step": 6550
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.654472320290479e-09,
+      "loss": 1.1914,
+      "step": 6551
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.566252213274222e-09,
+      "loss": 1.1196,
+      "step": 6552
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4804488466091394e-09,
+      "loss": 1.292,
+      "step": 6553
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3970622410364176e-09,
+      "loss": 1.1372,
+      "step": 6554
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3160924167110457e-09,
+      "loss": 1.2041,
+      "step": 6555
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.237539393205145e-09,
+      "loss": 1.2544,
+      "step": 6556
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1614031895046397e-09,
+      "loss": 1.249,
+      "step": 6557
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0876838240125865e-09,
+      "loss": 1.2021,
+      "step": 6558
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0163813145491751e-09,
+      "loss": 1.1919,
+      "step": 6559
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.474956783472876e-10,
+      "loss": 1.271,
+      "step": 6560
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.810269320591591e-10,
+      "loss": 1.2495,
+      "step": 6561
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.169750917508268e-10,
+      "loss": 1.1519,
+      "step": 6562
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.553401729021303e-10,
+      "loss": 1.106,
+      "step": 6563
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.961221904144833e-10,
+      "loss": 1.1523,
+      "step": 6564
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.393211585986603e-10,
+      "loss": 1.2705,
+      "step": 6565
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.849370911858998e-10,
+      "loss": 1.2329,
+      "step": 6566
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.329700013201323e-10,
+      "loss": 1.1147,
+      "step": 6567
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.83419901563531e-10,
+      "loss": 1.1572,
+      "step": 6568
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.3628680389318223e-10,
+      "loss": 1.0317,
+      "step": 6569
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9157071969997407e-10,
+      "loss": 1.2251,
+      "step": 6570
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.492716597941481e-10,
+      "loss": 1.3232,
+      "step": 6571
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.09389634398638e-10,
+      "loss": 1.2993,
+      "step": 6572
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.7192465315351024e-10,
+      "loss": 1.1313,
+      "step": 6573
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3687672511596426e-10,
+      "loss": 1.1851,
+      "step": 6574
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.0424585875589154e-10,
+      "loss": 1.3203,
+      "step": 6575
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7403206196142664e-10,
+      "loss": 1.3018,
+      "step": 6576
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4623534203561663e-10,
+      "loss": 1.23,
+      "step": 6577
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2085570569642102e-10,
+      "loss": 1.0635,
+      "step": 6578
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.789315907893227e-11,
+      "loss": 1.1978,
+      "step": 6579
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.734770773315525e-11,
+      "loss": 1.0293,
+      "step": 6580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.921935662511757e-11,
+      "loss": 1.1831,
+      "step": 6581
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.350811013797973e-11,
+      "loss": 1.2075,
+      "step": 6582
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.021397206870447e-11,
+      "loss": 1.2246,
+      "step": 6583
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9336945629166992e-11,
+      "loss": 1.2305,
+      "step": 6584
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0877033449485653e-11,
+      "loss": 1.2085,
+      "step": 6585
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.834237575801481e-12,
+      "loss": 1.2871,
+      "step": 6586
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2085594669475342e-12,
+      "loss": 1.2998,
+      "step": 6587
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 1.3896,
+      "step": 6588
+    },
+    {
+      "epoch": 3.0,
+      "step": 6588,
+      "total_flos": 75296487103488.0,
+      "train_loss": 1.328846397006157,
+      "train_runtime": 10477.8095,
+      "train_samples_per_second": 20.121,
+      "train_steps_per_second": 0.629
+    }
+  ],
+  "max_steps": 6588,
+  "num_train_epochs": 3,
+  "total_flos": 75296487103488.0,
+  "trial_name": null,
+  "trial_params": null
+}