diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19699 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9979428571428572,
+  "global_step": 3279,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0202020202020205e-07,
+      "loss": 1.3414,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.040404040404041e-07,
+      "loss": 1.581,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 1.2284,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.080808080808082e-07,
+      "loss": 1.087,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.01010101010101e-06,
+      "loss": 1.166,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 1.1451,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4141414141414143e-06,
+      "loss": 1.1736,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6161616161616164e-06,
+      "loss": 1.1449,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 1.0964,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.02020202020202e-06,
+      "loss": 1.169,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 1.3164,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 1.1253,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6262626262626267e-06,
+      "loss": 1.0958,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8282828282828286e-06,
+      "loss": 1.1517,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 1.0489,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.232323232323233e-06,
+      "loss": 1.0708,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4343434343434347e-06,
+      "loss": 1.0553,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 1.0495,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8383838383838385e-06,
+      "loss": 0.9567,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.04040404040404e-06,
+      "loss": 1.0277,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 1.0435,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.0486,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.646464646464647e-06,
+      "loss": 0.9802,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 1.0626,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0505050505050515e-06,
+      "loss": 0.9966,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.252525252525253e-06,
+      "loss": 0.955,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.0238,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.656565656565657e-06,
+      "loss": 0.9536,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.858585858585859e-06,
+      "loss": 0.9811,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.8889,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.262626262626264e-06,
+      "loss": 1.0214,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.464646464646466e-06,
+      "loss": 0.9981,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.9014,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.868686868686869e-06,
+      "loss": 0.8736,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.070707070707071e-06,
+      "loss": 1.0456,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.0409,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.474747474747476e-06,
+      "loss": 1.0405,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.676767676767677e-06,
+      "loss": 0.8925,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 0.9798,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.08080808080808e-06,
+      "loss": 1.0048,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.282828282828283e-06,
+      "loss": 0.9878,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.0403,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.686868686868687e-06,
+      "loss": 0.9321,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.9593,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.9827,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.292929292929294e-06,
+      "loss": 0.9805,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.494949494949497e-06,
+      "loss": 1.0418,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 0.9612,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.8989898989899e-06,
+      "loss": 0.9013,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0101010101010103e-05,
+      "loss": 0.8888,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0303030303030304e-05,
+      "loss": 0.8924,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0505050505050507e-05,
+      "loss": 0.9957,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0707070707070708e-05,
+      "loss": 0.9887,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0909090909090909e-05,
+      "loss": 0.9941,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 1.0991,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1313131313131314e-05,
+      "loss": 1.0076,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1515151515151517e-05,
+      "loss": 1.0065,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1717171717171718e-05,
+      "loss": 0.9868,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.191919191919192e-05,
+      "loss": 0.9156,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.8382,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2323232323232323e-05,
+      "loss": 0.9507,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2525252525252527e-05,
+      "loss": 0.9707,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2727272727272728e-05,
+      "loss": 0.9343,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2929292929292931e-05,
+      "loss": 0.9677,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3131313131313132e-05,
+      "loss": 0.9293,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.0126,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3535353535353538e-05,
+      "loss": 0.9187,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3737373737373739e-05,
+      "loss": 0.9066,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3939393939393942e-05,
+      "loss": 0.9361,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4141414141414143e-05,
+      "loss": 0.9218,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4343434343434344e-05,
+      "loss": 0.9278,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4545454545454546e-05,
+      "loss": 0.9535,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4747474747474747e-05,
+      "loss": 1.0225,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4949494949494952e-05,
+      "loss": 1.0373,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.9291,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5353535353535354e-05,
+      "loss": 0.978,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 1.0299,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.575757575757576e-05,
+      "loss": 1.0262,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.595959595959596e-05,
+      "loss": 0.8195,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.616161616161616e-05,
+      "loss": 0.9034,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6363636363636366e-05,
+      "loss": 0.8745,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6565656565656567e-05,
+      "loss": 0.9635,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6767676767676768e-05,
+      "loss": 0.9052,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6969696969696972e-05,
+      "loss": 0.9106,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7171717171717173e-05,
+      "loss": 0.9609,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7373737373737375e-05,
+      "loss": 0.919,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7575757575757576e-05,
+      "loss": 0.8958,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.975,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.797979797979798e-05,
+      "loss": 0.9397,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.968,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8383838383838387e-05,
+      "loss": 0.9304,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8585858585858588e-05,
+      "loss": 0.9911,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8787878787878792e-05,
+      "loss": 1.0626,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8989898989898993e-05,
+      "loss": 0.9815,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9191919191919194e-05,
+      "loss": 0.979,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9393939393939395e-05,
+      "loss": 0.9367,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9595959595959596e-05,
+      "loss": 0.8822,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97979797979798e-05,
+      "loss": 0.9143,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.9985,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999995120048853e-05,
+      "loss": 0.9839,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999980480200173e-05,
+      "loss": 0.955,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999956080468243e-05,
+      "loss": 1.0357,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999921920876882e-05,
+      "loss": 0.9288,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999878001459436e-05,
+      "loss": 1.0697,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999824322258754e-05,
+      "loss": 0.9964,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999976088332724e-05,
+      "loss": 0.8158,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999687684726803e-05,
+      "loss": 0.8999,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999960472652889e-05,
+      "loss": 0.8613,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951200881446e-05,
+      "loss": 0.9826,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999409531674007e-05,
+      "loss": 1.0505,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999929729520755e-05,
+      "loss": 0.9588,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999175299524628e-05,
+      "loss": 0.9167,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999904354474431e-05,
+      "loss": 1.0284,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999890203099519e-05,
+      "loss": 0.919,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999875075841538e-05,
+      "loss": 0.8996,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999858972715252e-05,
+      "loss": 1.0541,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999841893736377e-05,
+      "loss": 0.9601,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998238389215832e-05,
+      "loss": 1.0069,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999804808288491e-05,
+      "loss": 0.939,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997848018556745e-05,
+      "loss": 0.98,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997638196426595e-05,
+      "loss": 0.8682,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997418616699246e-05,
+      "loss": 1.0236,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997189279589003e-05,
+      "loss": 0.9945,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99969501853197e-05,
+      "loss": 0.834,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996701334124693e-05,
+      "loss": 0.9289,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999644272624685e-05,
+      "loss": 0.9051,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999617436193858e-05,
+      "loss": 0.8805,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995896241461797e-05,
+      "loss": 0.9646,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995608365087945e-05,
+      "loss": 1.0058,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999531073309799e-05,
+      "loss": 0.8868,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995003345782416e-05,
+      "loss": 0.9385,
+      "step": 131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994686203441232e-05,
+      "loss": 0.9179,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994359306383964e-05,
+      "loss": 0.9673,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994022654929665e-05,
+      "loss": 0.974,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993676249406895e-05,
+      "loss": 0.9309,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999332009015375e-05,
+      "loss": 0.9767,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992954177517835e-05,
+      "loss": 0.8175,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992578511856276e-05,
+      "loss": 0.9322,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999219309353572e-05,
+      "loss": 0.908,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991797922932333e-05,
+      "loss": 0.8858,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991393000431798e-05,
+      "loss": 0.8947,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990978326429312e-05,
+      "loss": 1.0159,
+      "step": 142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999055390132959e-05,
+      "loss": 0.9793,
+      "step": 143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999011972554688e-05,
+      "loss": 0.9845,
+      "step": 144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989675799504923e-05,
+      "loss": 0.9152,
+      "step": 145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989222123636986e-05,
+      "loss": 0.8651,
+      "step": 146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988758698385854e-05,
+      "loss": 0.9883,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988285524203828e-05,
+      "loss": 0.8313,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987802601552717e-05,
+      "loss": 0.8789,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987309930903855e-05,
+      "loss": 0.8917,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986807512738075e-05,
+      "loss": 1.1032,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986295347545738e-05,
+      "loss": 0.8723,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985773435826716e-05,
+      "loss": 0.9284,
+      "step": 153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998524177809038e-05,
+      "loss": 0.9412,
+      "step": 154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998470037485563e-05,
+      "loss": 1.0094,
+      "step": 155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998414922665087e-05,
+      "loss": 0.9569,
+      "step": 156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998358833401401e-05,
+      "loss": 0.9792,
+      "step": 157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998301769749248e-05,
+      "loss": 0.9994,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982437317643218e-05,
+      "loss": 0.9995,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998184719503266e-05,
+      "loss": 0.9938,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981247330236767e-05,
+      "loss": 0.867,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980637723841005e-05,
+      "loss": 0.9597,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998001837644033e-05,
+      "loss": 0.9129,
+      "step": 163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979389288639237e-05,
+      "loss": 0.8977,
+      "step": 164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978750461051698e-05,
+      "loss": 0.9129,
+      "step": 165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99781018943012e-05,
+      "loss": 0.8885,
+      "step": 166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997744358902075e-05,
+      "loss": 0.8942,
+      "step": 167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976775545852836e-05,
+      "loss": 0.9809,
+      "step": 168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976097765449463e-05,
+      "loss": 0.9909,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975410248472147e-05,
+      "loss": 0.958,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974712995591887e-05,
+      "loss": 0.944,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974006007489203e-05,
+      "loss": 0.9756,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973289284854104e-05,
+      "loss": 0.9864,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99725628283861e-05,
+      "loss": 0.9228,
+      "step": 174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997182663879422e-05,
+      "loss": 1.0225,
+      "step": 175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971080716796964e-05,
+      "loss": 0.952,
+      "step": 176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970325063122348e-05,
+      "loss": 0.9391,
+      "step": 177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996955967850789e-05,
+      "loss": 0.8499,
+      "step": 178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968784563700586e-05,
+      "loss": 0.9585,
+      "step": 179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967999719456944e-05,
+      "loss": 0.9802,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967205146542973e-05,
+      "loss": 0.8679,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966400845734157e-05,
+      "loss": 0.9034,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965586817815494e-05,
+      "loss": 0.78,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964763063581465e-05,
+      "loss": 0.8784,
+      "step": 184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963929583836043e-05,
+      "loss": 0.9087,
+      "step": 185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963086379392698e-05,
+      "loss": 0.9302,
+      "step": 186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996223345107439e-05,
+      "loss": 0.9158,
+      "step": 187
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996137079971357e-05,
+      "loss": 1.0904,
+      "step": 188
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996049842615217e-05,
+      "loss": 0.8892,
+      "step": 189
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959616331241628e-05,
+      "loss": 0.8382,
+      "step": 190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958724515842856e-05,
+      "loss": 0.8461,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957822980826257e-05,
+      "loss": 0.9942,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995691172707172e-05,
+      "loss": 0.8786,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955990755468614e-05,
+      "loss": 0.9105,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995506006691581e-05,
+      "loss": 0.7972,
+      "step": 195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954119662321643e-05,
+      "loss": 0.8615,
+      "step": 196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953169542603942e-05,
+      "loss": 0.9485,
+      "step": 197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995220970869001e-05,
+      "loss": 1.06,
+      "step": 198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951240161516643e-05,
+      "loss": 0.9071,
+      "step": 199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950260902030104e-05,
+      "loss": 0.9927,
+      "step": 200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994927193118614e-05,
+      "loss": 0.9858,
+      "step": 201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994827324994998e-05,
+      "loss": 1.0367,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994726485929633e-05,
+      "loss": 0.9767,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994624676020936e-05,
+      "loss": 0.9419,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 1.0116,
+      "step": 205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944181440719577e-05,
+      "loss": 0.9853,
+      "step": 206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943134222332493e-05,
+      "loss": 0.9729,
+      "step": 207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942077299543553e-05,
+      "loss": 0.8523,
+      "step": 208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941010673384307e-05,
+      "loss": 1.0664,
+      "step": 209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993993434489577e-05,
+      "loss": 0.8913,
+      "step": 210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993884831512843e-05,
+      "loss": 0.9395,
+      "step": 211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993775258514224e-05,
+      "loss": 0.9643,
+      "step": 212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936647156006623e-05,
+      "loss": 0.8591,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935532028800465e-05,
+      "loss": 0.9196,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934407204612124e-05,
+      "loss": 0.8508,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993327268453941e-05,
+      "loss": 0.9152,
+      "step": 216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993212846968961e-05,
+      "loss": 0.8998,
+      "step": 217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993097456117946e-05,
+      "loss": 0.8472,
+      "step": 218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992981096013517e-05,
+      "loss": 0.8898,
+      "step": 219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99286376676924e-05,
+      "loss": 0.833,
+      "step": 220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927454684996276e-05,
+      "loss": 1.0059,
+      "step": 221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992626201320137e-05,
+      "loss": 0.8284,
+      "step": 222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992505965347172e-05,
+      "loss": 0.9868,
+      "step": 223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923847606980824e-05,
+      "loss": 1.0244,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922625874911624e-05,
+      "loss": 0.8332,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921394458456514e-05,
+      "loss": 0.9581,
+      "step": 226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992015335881735e-05,
+      "loss": 0.9231,
+      "step": 227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991890257720543e-05,
+      "loss": 0.8909,
+      "step": 228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917642114841505e-05,
+      "loss": 0.8423,
+      "step": 229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916371972955778e-05,
+      "loss": 0.9186,
+      "step": 230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915092152787888e-05,
+      "loss": 0.959,
+      "step": 231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991380265558693e-05,
+      "loss": 1.0065,
+      "step": 232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912503482611446e-05,
+      "loss": 0.9962,
+      "step": 233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991119463512941e-05,
+      "loss": 0.9263,
+      "step": 234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909876114418242e-05,
+      "loss": 1.0371,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908547921764814e-05,
+      "loss": 0.9661,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990721005846542e-05,
+      "loss": 1.0151,
+      "step": 237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905862525825814e-05,
+      "loss": 0.8864,
+      "step": 238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990450532516116e-05,
+      "loss": 0.8544,
+      "step": 239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990313845779608e-05,
+      "loss": 0.9322,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990176192506462e-05,
+      "loss": 0.9988,
+      "step": 241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900375728310267e-05,
+      "loss": 0.9571,
+      "step": 242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898979868885933e-05,
+      "loss": 0.887,
+      "step": 243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897574348153963e-05,
+      "loss": 0.9616,
+      "step": 244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896159167486134e-05,
+      "loss": 0.8804,
+      "step": 245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894734328263645e-05,
+      "loss": 0.9911,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893299831877124e-05,
+      "loss": 0.9893,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891855679726632e-05,
+      "loss": 0.9558,
+      "step": 248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890401873221642e-05,
+      "loss": 0.8807,
+      "step": 249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888938413781056e-05,
+      "loss": 1.0169,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887465302833194e-05,
+      "loss": 0.9123,
+      "step": 251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885982541815805e-05,
+      "loss": 0.8542,
+      "step": 252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988449013217604e-05,
+      "loss": 0.9046,
+      "step": 253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988298807537049e-05,
+      "loss": 0.9273,
+      "step": 254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988147637286513e-05,
+      "loss": 0.9856,
+      "step": 255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879955026135377e-05,
+      "loss": 0.881,
+      "step": 256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.987842403666605e-05,
+      "loss": 1.0774,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.8468,
+      "step": 258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875333135495e-05,
+      "loss": 0.8823,
+      "step": 259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873773226809974e-05,
+      "loss": 0.8475,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872203681418745e-05,
+      "loss": 0.8981,
+      "step": 261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987062450085318e-05,
+      "loss": 0.9019,
+      "step": 262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869035686654538e-05,
+      "loss": 0.8574,
+      "step": 263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986743724037349e-05,
+      "loss": 0.9591,
+      "step": 264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865829163570104e-05,
+      "loss": 0.8535,
+      "step": 265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986421145781385e-05,
+      "loss": 0.9552,
+      "step": 266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862584124683587e-05,
+      "loss": 1.01,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986094716576758e-05,
+      "loss": 0.982,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985930058266348e-05,
+      "loss": 0.9461,
+      "step": 269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857644376978347e-05,
+      "loss": 1.0262,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985597855032861e-05,
+      "loss": 0.9215,
+      "step": 271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854303104340106e-05,
+      "loss": 0.9044,
+      "step": 272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985261804064805e-05,
+      "loss": 0.8892,
+      "step": 273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850923360897055e-05,
+      "loss": 0.95,
+      "step": 274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849219066741102e-05,
+      "loss": 0.897,
+      "step": 275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847505159843574e-05,
+      "loss": 0.8869,
+      "step": 276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845781641877217e-05,
+      "loss": 0.9752,
+      "step": 277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984404851452418e-05,
+      "loss": 0.8943,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984230577947597e-05,
+      "loss": 0.9793,
+      "step": 279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840553438433482e-05,
+      "loss": 0.8199,
+      "step": 280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838791493106985e-05,
+      "loss": 0.9541,
+      "step": 281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983701994521612e-05,
+      "loss": 0.883,
+      "step": 282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98352387964899e-05,
+      "loss": 0.8463,
+      "step": 283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98334480486667e-05,
+      "loss": 1.0044,
+      "step": 284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831647703494287e-05,
+      "loss": 0.9429,
+      "step": 285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829837762729776e-05,
+      "loss": 1.0667,
+      "step": 286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828018228139647e-05,
+      "loss": 0.8432,
+      "step": 287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982618910149975e-05,
+      "loss": 0.8957,
+      "step": 288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824350384595295e-05,
+      "loss": 1.0524,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822502079220854e-05,
+      "loss": 0.9636,
+      "step": 290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820644187180354e-05,
+      "loss": 0.9498,
+      "step": 291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818776710287072e-05,
+      "loss": 1.0486,
+      "step": 292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816899650363658e-05,
+      "loss": 0.9157,
+      "step": 293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815013009242103e-05,
+      "loss": 0.923,
+      "step": 294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813116788763744e-05,
+      "loss": 0.9505,
+      "step": 295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811210990779275e-05,
+      "loss": 0.9437,
+      "step": 296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980929561714874e-05,
+      "loss": 0.8678,
+      "step": 297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807370669741522e-05,
+      "loss": 0.9705,
+      "step": 298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805436150436352e-05,
+      "loss": 0.8886,
+      "step": 299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98034920611213e-05,
+      "loss": 0.7751,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801538403693782e-05,
+      "loss": 1.0173,
+      "step": 301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799575180060544e-05,
+      "loss": 0.9569,
+      "step": 302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797602392137678e-05,
+      "loss": 0.8524,
+      "step": 303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795620041850602e-05,
+      "loss": 1.0323,
+      "step": 304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793628131134073e-05,
+      "loss": 1.0158,
+      "step": 305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791626661932172e-05,
+      "loss": 0.9054,
+      "step": 306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978961563619832e-05,
+      "loss": 0.9269,
+      "step": 307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787595055895253e-05,
+      "loss": 0.9574,
+      "step": 308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785564922995042e-05,
+      "loss": 0.9999,
+      "step": 309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978352523947907e-05,
+      "loss": 0.9429,
+      "step": 310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781476007338058e-05,
+      "loss": 0.9068,
+      "step": 311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977941722857203e-05,
+      "loss": 0.924,
+      "step": 312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777348905190334e-05,
+      "loss": 0.8962,
+      "step": 313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775271039211633e-05,
+      "loss": 0.9294,
+      "step": 314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773183632663907e-05,
+      "loss": 0.9526,
+      "step": 315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771086687584443e-05,
+      "loss": 0.9028,
+      "step": 316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768980206019838e-05,
+      "loss": 0.8341,
+      "step": 317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766864190025997e-05,
+      "loss": 0.9338,
+      "step": 318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764738641668137e-05,
+      "loss": 0.8725,
+      "step": 319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762603563020765e-05,
+      "loss": 0.9987,
+      "step": 320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760458956167698e-05,
+      "loss": 0.8647,
+      "step": 321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758304823202056e-05,
+      "loss": 0.8858,
+      "step": 322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9756141166226246e-05,
+      "loss": 0.9212,
+      "step": 323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975396798735198e-05,
+      "loss": 0.9524,
+      "step": 324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751785288700255e-05,
+      "loss": 1.0198,
+      "step": 325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974959307240137e-05,
+      "loss": 0.9041,
+      "step": 326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97473913405949e-05,
+      "loss": 1.1178,
+      "step": 327
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974518009542972e-05,
+      "loss": 1.0656,
+      "step": 328
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742959339063977e-05,
+      "loss": 0.9484,
+      "step": 329
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740729073665115e-05,
+      "loss": 0.932,
+      "step": 330
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738489301409848e-05,
+      "loss": 0.9525,
+      "step": 331
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736240024484167e-05,
+      "loss": 0.8499,
+      "step": 332
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733981245083355e-05,
+      "loss": 0.9412,
+      "step": 333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731712965411947e-05,
+      "loss": 0.778,
+      "step": 334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972943518768377e-05,
+      "loss": 1.1284,
+      "step": 335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727147914121905e-05,
+      "loss": 0.9607,
+      "step": 336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724851146958713e-05,
+      "loss": 0.8447,
+      "step": 337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722544888435823e-05,
+      "loss": 0.9549,
+      "step": 338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972022914080411e-05,
+      "loss": 0.8569,
+      "step": 339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717903906323727e-05,
+      "loss": 0.79,
+      "step": 340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715569187264077e-05,
+      "loss": 0.8571,
+      "step": 341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713224985903828e-05,
+      "loss": 0.8961,
+      "step": 342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971087130453089e-05,
+      "loss": 0.8269,
+      "step": 343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708508145442443e-05,
+      "loss": 0.7575,
+      "step": 344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9706135510944894e-05,
+      "loss": 0.9035,
+      "step": 345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703753403353926e-05,
+      "loss": 0.9816,
+      "step": 346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970136182499444e-05,
+      "loss": 0.923,
+      "step": 347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698960778200603e-05,
+      "loss": 0.92,
+      "step": 348
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696550265315805e-05,
+      "loss": 0.9841,
+      "step": 349
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694130288692688e-05,
+      "loss": 0.8766,
+      "step": 350
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691700850693126e-05,
+      "loss": 0.8762,
+      "step": 351
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689261953688224e-05,
+      "loss": 0.8369,
+      "step": 352
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968681360005832e-05,
+      "loss": 0.813,
+      "step": 353
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968435579219299e-05,
+      "loss": 0.8423,
+      "step": 354
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681888532491022e-05,
+      "loss": 0.868,
+      "step": 355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679411823360443e-05,
+      "loss": 0.9451,
+      "step": 356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676925667218498e-05,
+      "loss": 0.9529,
+      "step": 357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674430066491644e-05,
+      "loss": 0.8636,
+      "step": 358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671925023615572e-05,
+      "loss": 0.9766,
+      "step": 359
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669410541035176e-05,
+      "loss": 0.8872,
+      "step": 360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666886621204563e-05,
+      "loss": 0.9897,
+      "step": 361
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664353266587055e-05,
+      "loss": 0.8516,
+      "step": 362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661810479655184e-05,
+      "loss": 0.9416,
+      "step": 363
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 0.9641,
+      "step": 364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656696618784494e-05,
+      "loss": 1.0015,
+      "step": 365
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654125549836754e-05,
+      "loss": 0.97,
+      "step": 366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651545058556803e-05,
+      "loss": 0.9077,
+      "step": 367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9648955147463172e-05,
+      "loss": 0.9849,
+      "step": 368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964635581908359e-05,
+      "loss": 0.9465,
+      "step": 369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643747075954974e-05,
+      "loss": 0.9638,
+      "step": 370
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641128920623438e-05,
+      "loss": 0.8555,
+      "step": 371
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638501355644274e-05,
+      "loss": 1.0046,
+      "step": 372
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635864383581954e-05,
+      "loss": 0.9257,
+      "step": 373
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633218007010146e-05,
+      "loss": 0.8664,
+      "step": 374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630562228511682e-05,
+      "loss": 0.9133,
+      "step": 375
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627897050678575e-05,
+      "loss": 0.844,
+      "step": 376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962522247611202e-05,
+      "loss": 0.8511,
+      "step": 377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622538507422363e-05,
+      "loss": 0.9692,
+      "step": 378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961984514722914e-05,
+      "loss": 0.8855,
+      "step": 379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961714239816104e-05,
+      "loss": 0.9232,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961443026285593e-05,
+      "loss": 0.9026,
+      "step": 381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611708743960817e-05,
+      "loss": 0.8979,
+      "step": 382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9608977844131875e-05,
+      "loss": 0.9881,
+      "step": 383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606237566034443e-05,
+      "loss": 0.9712,
+      "step": 384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9603487912343003e-05,
+      "loss": 1.0181,
+      "step": 385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600728885741188e-05,
+      "loss": 0.9939,
+      "step": 386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597960488921785e-05,
+      "loss": 0.9163,
+      "step": 387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959518272458672e-05,
+      "loss": 0.9167,
+      "step": 388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592395595447064e-05,
+      "loss": 0.877,
+      "step": 389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589599104223027e-05,
+      "loss": 0.991,
+      "step": 390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958679325364396e-05,
+      "loss": 0.8778,
+      "step": 391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583978046448343e-05,
+      "loss": 0.8047,
+      "step": 392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958115348538379e-05,
+      "loss": 0.8966,
+      "step": 393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578319573207047e-05,
+      "loss": 0.9214,
+      "step": 394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575476312683985e-05,
+      "loss": 0.9173,
+      "step": 395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572623706589598e-05,
+      "loss": 0.9572,
+      "step": 396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569761757708e-05,
+      "loss": 0.8786,
+      "step": 397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566890468832428e-05,
+      "loss": 0.9745,
+      "step": 398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564009842765225e-05,
+      "loss": 0.941,
+      "step": 399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561119882317863e-05,
+      "loss": 0.9889,
+      "step": 400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955822059031091e-05,
+      "loss": 0.9464,
+      "step": 401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955531196957405e-05,
+      "loss": 0.9692,
+      "step": 402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552394022946068e-05,
+      "loss": 0.892,
+      "step": 403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549466753274843e-05,
+      "loss": 0.9064,
+      "step": 404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546530163417374e-05,
+      "loss": 0.8169,
+      "step": 405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543584256239738e-05,
+      "loss": 0.802,
+      "step": 406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540629034617108e-05,
+      "loss": 0.865,
+      "step": 407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953766450143376e-05,
+      "loss": 0.9968,
+      "step": 408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534690659583045e-05,
+      "loss": 0.9214,
+      "step": 409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9531707511967402e-05,
+      "loss": 0.8974,
+      "step": 410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9528715061498355e-05,
+      "loss": 0.876,
+      "step": 411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525713311096506e-05,
+      "loss": 0.8862,
+      "step": 412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952270226369154e-05,
+      "loss": 0.8421,
+      "step": 413
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9519681922222195e-05,
+      "loss": 0.9565,
+      "step": 414
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951665228963631e-05,
+      "loss": 0.9584,
+      "step": 415
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951361336889077e-05,
+      "loss": 0.9306,
+      "step": 416
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9510565162951538e-05,
+      "loss": 0.9785,
+      "step": 417
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507507674793624e-05,
+      "loss": 1.0068,
+      "step": 418
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504440907401113e-05,
+      "loss": 0.9803,
+      "step": 419
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501364863767134e-05,
+      "loss": 0.9044,
+      "step": 420
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498279546893885e-05,
+      "loss": 0.9102,
+      "step": 421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495184959792598e-05,
+      "loss": 0.8586,
+      "step": 422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949208110548356e-05,
+      "loss": 0.8529,
+      "step": 423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488967986996105e-05,
+      "loss": 0.8289,
+      "step": 424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9485845607368606e-05,
+      "loss": 1.0007,
+      "step": 425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948271396964847e-05,
+      "loss": 0.868,
+      "step": 426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479573076892152e-05,
+      "loss": 0.8841,
+      "step": 427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947642293216513e-05,
+      "loss": 0.9878,
+      "step": 428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473263538541916e-05,
+      "loss": 0.8987,
+      "step": 429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947009489910604e-05,
+      "loss": 0.8179,
+      "step": 430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466917016950076e-05,
+      "loss": 1.0086,
+      "step": 431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463729895175594e-05,
+      "loss": 0.9908,
+      "step": 432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460533536893197e-05,
+      "loss": 0.9412,
+      "step": 433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457327945222505e-05,
+      "loss": 0.9154,
+      "step": 434
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9454113123292133e-05,
+      "loss": 0.8769,
+      "step": 435
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450889074239728e-05,
+      "loss": 0.945,
+      "step": 436
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447655801211924e-05,
+      "loss": 0.8924,
+      "step": 437
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444413307364365e-05,
+      "loss": 0.9352,
+      "step": 438
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944116159586169e-05,
+      "loss": 0.9174,
+      "step": 439
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437900669877544e-05,
+      "loss": 0.8824,
+      "step": 440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434630532594555e-05,
+      "loss": 0.9282,
+      "step": 441
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943135118720435e-05,
+      "loss": 0.9299,
+      "step": 442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9428062636907526e-05,
+      "loss": 0.8891,
+      "step": 443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942476488491369e-05,
+      "loss": 0.8615,
+      "step": 444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421457934441407e-05,
+      "loss": 0.8262,
+      "step": 445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941814178871823e-05,
+      "loss": 0.7999,
+      "step": 446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9414816450980686e-05,
+      "loss": 1.0586,
+      "step": 447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411481924474276e-05,
+      "loss": 0.8303,
+      "step": 448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408138212453456e-05,
+      "loss": 0.9892,
+      "step": 449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404785318181662e-05,
+      "loss": 0.9614,
+      "step": 450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940142324493129e-05,
+      "loss": 0.8951,
+      "step": 451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939805199598368e-05,
+      "loss": 0.9504,
+      "step": 452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394671574629147e-05,
+      "loss": 0.8506,
+      "step": 453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391281984166944e-05,
+      "loss": 0.9006,
+      "step": 454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9387883227905285e-05,
+      "loss": 0.9175,
+      "step": 455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384475309161315e-05,
+      "loss": 0.87,
+      "step": 456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381058231261135e-05,
+      "loss": 0.9294,
+      "step": 457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377631997539776e-05,
+      "loss": 0.7557,
+      "step": 458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374196611341212e-05,
+      "loss": 0.8288,
+      "step": 459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370752076018344e-05,
+      "loss": 0.8616,
+      "step": 460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367298394933e-05,
+      "loss": 0.9612,
+      "step": 461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936383557145595e-05,
+      "loss": 0.9082,
+      "step": 462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936036360896687e-05,
+      "loss": 0.9103,
+      "step": 463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935688251085436e-05,
+      "loss": 0.9297,
+      "step": 464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353392280515938e-05,
+      "loss": 0.7854,
+      "step": 465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934989292135804e-05,
+      "loss": 0.8918,
+      "step": 466
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9346384436796e-05,
+      "loss": 0.8384,
+      "step": 467
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342866830254066e-05,
+      "loss": 0.8377,
+      "step": 468
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933934010516539e-05,
+      "loss": 0.87,
+      "step": 469
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9335804264972018e-05,
+      "loss": 0.8854,
+      "step": 470
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93322593131249e-05,
+      "loss": 0.8994,
+      "step": 471
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932870525308387e-05,
+      "loss": 1.0305,
+      "step": 472
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932514208831766e-05,
+      "loss": 1.0101,
+      "step": 473
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932156982230388e-05,
+      "loss": 0.8908,
+      "step": 474
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931798845852903e-05,
+      "loss": 0.9151,
+      "step": 475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314398000488484e-05,
+      "loss": 0.8491,
+      "step": 476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310798451686488e-05,
+      "loss": 0.9944,
+      "step": 477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307189815636184e-05,
+      "loss": 0.9264,
+      "step": 478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9303572095859545e-05,
+      "loss": 0.9209,
+      "step": 479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299945295887444e-05,
+      "loss": 0.9303,
+      "step": 480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.92963094192596e-05,
+      "loss": 0.8468,
+      "step": 481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9292664469524587e-05,
+      "loss": 0.9206,
+      "step": 482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289010450239843e-05,
+      "loss": 0.8915,
+      "step": 483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285347364971662e-05,
+      "loss": 0.8742,
+      "step": 484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928167521729517e-05,
+      "loss": 1.017,
+      "step": 485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927799401079435e-05,
+      "loss": 1.0436,
+      "step": 486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274303749062028e-05,
+      "loss": 0.8583,
+      "step": 487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927060443569986e-05,
+      "loss": 0.9063,
+      "step": 488
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266896074318335e-05,
+      "loss": 0.9404,
+      "step": 489
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263178668536785e-05,
+      "loss": 0.9019,
+      "step": 490
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925945222198336e-05,
+      "loss": 0.9586,
+      "step": 491
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9255716738295035e-05,
+      "loss": 0.8527,
+      "step": 492
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251972221117603e-05,
+      "loss": 0.8532,
+      "step": 493
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248218674105678e-05,
+      "loss": 0.7986,
+      "step": 494
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924445610092269e-05,
+      "loss": 0.8746,
+      "step": 495
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9240684505240862e-05,
+      "loss": 0.8434,
+      "step": 496
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236903890741247e-05,
+      "loss": 0.8904,
+      "step": 497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923311426111368e-05,
+      "loss": 0.8422,
+      "step": 498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9229315620056805e-05,
+      "loss": 0.8426,
+      "step": 499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9225507971278057e-05,
+      "loss": 0.9086,
+      "step": 500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221691318493666e-05,
+      "loss": 0.8575,
+      "step": 501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921786566542865e-05,
+      "loss": 0.7999,
+      "step": 502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214031015816803e-05,
+      "loss": 0.8696,
+      "step": 503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921018737340071e-05,
+      "loss": 1.0075,
+      "step": 504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920633474193173e-05,
+      "loss": 0.9381,
+      "step": 505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202473125169986e-05,
+      "loss": 0.9506,
+      "step": 506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198602526884388e-05,
+      "loss": 0.9163,
+      "step": 507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194722950852594e-05,
+      "loss": 0.9182,
+      "step": 508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190834400861035e-05,
+      "loss": 0.9579,
+      "step": 509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91869368807049e-05,
+      "loss": 0.9198,
+      "step": 510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918303039418813e-05,
+      "loss": 0.8779,
+      "step": 511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179114945123414e-05,
+      "loss": 0.8959,
+      "step": 512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9175190537332198e-05,
+      "loss": 0.7918,
+      "step": 513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171257174644658e-05,
+      "loss": 0.8551,
+      "step": 514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9167314860899724e-05,
+      "loss": 0.8779,
+      "step": 515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9163363599945057e-05,
+      "loss": 0.8678,
+      "step": 516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915940339563704e-05,
+      "loss": 0.7786,
+      "step": 517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91554342518408e-05,
+      "loss": 0.8084,
+      "step": 518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151456172430186e-05,
+      "loss": 1.0102,
+      "step": 519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147469161287756e-05,
+      "loss": 0.9359,
+      "step": 520
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143473222304803e-05,
+      "loss": 0.8538,
+      "step": 521
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913946835938132e-05,
+      "loss": 0.875,
+      "step": 522
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.9364,
+      "step": 523
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131431877356292e-05,
+      "loss": 0.7775,
+      "step": 524
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912740026609828e-05,
+      "loss": 0.9301,
+      "step": 525
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123359746586785e-05,
+      "loss": 0.9047,
+      "step": 526
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119310322765315e-05,
+      "loss": 0.957,
+      "step": 527
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911525199858607e-05,
+      "loss": 0.8311,
+      "step": 528
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111184778009934e-05,
+      "loss": 0.9187,
+      "step": 529
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107108665006472e-05,
+      "loss": 0.7972,
+      "step": 530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910302366355393e-05,
+      "loss": 0.9579,
+      "step": 531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9098929777639238e-05,
+      "loss": 0.8229,
+      "step": 532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094827011257982e-05,
+      "loss": 0.9275,
+      "step": 533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909071536841442e-05,
+      "loss": 0.8628,
+      "step": 534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908659485312148e-05,
+      "loss": 0.9558,
+      "step": 535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908246546940074e-05,
+      "loss": 0.8974,
+      "step": 536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907832722128244e-05,
+      "loss": 0.871,
+      "step": 537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074180112805473e-05,
+      "loss": 0.8091,
+      "step": 538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070024148017375e-05,
+      "loss": 0.7603,
+      "step": 539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906585933097432e-05,
+      "loss": 0.8467,
+      "step": 540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906168566574114e-05,
+      "loss": 0.9481,
+      "step": 541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9057503156391284e-05,
+      "loss": 0.8084,
+      "step": 542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053311807006845e-05,
+      "loss": 0.8133,
+      "step": 543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049111621678534e-05,
+      "loss": 0.9831,
+      "step": 544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044902604505693e-05,
+      "loss": 0.8048,
+      "step": 545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040684759596282e-05,
+      "loss": 0.9359,
+      "step": 546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036458091066875e-05,
+      "loss": 1.0393,
+      "step": 547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903222260304266e-05,
+      "loss": 0.8936,
+      "step": 548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027978299657436e-05,
+      "loss": 0.9743,
+      "step": 549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9023725185053596e-05,
+      "loss": 1.0751,
+      "step": 550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019463263382142e-05,
+      "loss": 0.9372,
+      "step": 551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015192538802664e-05,
+      "loss": 0.8068,
+      "step": 552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901091301548335e-05,
+      "loss": 0.8951,
+      "step": 553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900662469760097e-05,
+      "loss": 1.0006,
+      "step": 554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900232758934089e-05,
+      "loss": 0.8453,
+      "step": 555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998021694897032e-05,
+      "loss": 0.9771,
+      "step": 556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8993707018471918e-05,
+      "loss": 0.8556,
+      "step": 557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8989383564276624e-05,
+      "loss": 0.8707,
+      "step": 558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89850513365308e-05,
+      "loss": 0.8754,
+      "step": 559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980710339462657e-05,
+      "loss": 0.9125,
+      "step": 560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897636057730897e-05,
+      "loss": 0.8121,
+      "step": 561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897200205431506e-05,
+      "loss": 0.7994,
+      "step": 562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967634774734807e-05,
+      "loss": 0.8168,
+      "step": 563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896325874283063e-05,
+      "loss": 0.9251,
+      "step": 564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958873962873493e-05,
+      "loss": 0.9357,
+      "step": 565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8954480439142898e-05,
+      "loss": 0.8923,
+      "step": 566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950078175926886e-05,
+      "loss": 0.8286,
+      "step": 567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945667177522016e-05,
+      "loss": 0.9756,
+      "step": 568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941247448233386e-05,
+      "loss": 0.9335,
+      "step": 569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936818992374606e-05,
+      "loss": 0.8577,
+      "step": 570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932381814267802e-05,
+      "loss": 0.9053,
+      "step": 571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892793591824362e-05,
+      "loss": 0.8044,
+      "step": 572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923481308641216e-05,
+      "loss": 0.8978,
+      "step": 573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8919017989808238e-05,
+      "loss": 0.9049,
+      "step": 574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914545966100843e-05,
+      "loss": 0.879,
+      "step": 575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.8634,
+      "step": 576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.89055758215299e-05,
+      "loss": 0.9076,
+      "step": 577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8901077709421127e-05,
+      "loss": 0.8104,
+      "step": 578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896570909947477e-05,
+      "loss": 0.8782,
+      "step": 579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889205542750754e-05,
+      "loss": 0.9777,
+      "step": 580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887531266508383e-05,
+      "loss": 0.9063,
+      "step": 581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882998431365545e-05,
+      "loss": 0.9108,
+      "step": 582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887845692650303e-05,
+      "loss": 0.8512,
+      "step": 583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88739067563533e-05,
+      "loss": 0.9421,
+      "step": 584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869347925357275e-05,
+      "loss": 0.8686,
+      "step": 585
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864780437964333e-05,
+      "loss": 0.9189,
+      "step": 586
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860204298632294e-05,
+      "loss": 0.7707,
+      "step": 587
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885561951182743e-05,
+      "loss": 0.9605,
+      "step": 588
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885102608202444e-05,
+      "loss": 0.8338,
+      "step": 589
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8846424013706476e-05,
+      "loss": 0.8665,
+      "step": 590
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8841813311365105e-05,
+      "loss": 0.9193,
+      "step": 591
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883719397950033e-05,
+      "loss": 0.9165,
+      "step": 592
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832566022620574e-05,
+      "loss": 0.9296,
+      "step": 593
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827929445242678e-05,
+      "loss": 0.9546,
+      "step": 594
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882328425189189e-05,
+      "loss": 0.8645,
+      "step": 595
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818630447101884e-05,
+      "loss": 0.8792,
+      "step": 596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881396803541472e-05,
+      "loss": 1.0456,
+      "step": 597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880929702138087e-05,
+      "loss": 0.7963,
+      "step": 598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88046174095592e-05,
+      "loss": 0.9594,
+      "step": 599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879992920451696e-05,
+      "loss": 0.8759,
+      "step": 600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795232410829794e-05,
+      "loss": 0.8908,
+      "step": 601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879052703308173e-05,
+      "loss": 0.8762,
+      "step": 602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785813075865164e-05,
+      "loss": 0.8357,
+      "step": 603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8781090543780876e-05,
+      "loss": 0.8705,
+      "step": 604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776359441438017e-05,
+      "loss": 1.043,
+      "step": 605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877161977345409e-05,
+      "loss": 0.8323,
+      "step": 606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8766871544454963e-05,
+      "loss": 0.8874,
+      "step": 607
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762114759074863e-05,
+      "loss": 0.8693,
+      "step": 608
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875734942195637e-05,
+      "loss": 0.9408,
+      "step": 609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8752575537750402e-05,
+      "loss": 0.8655,
+      "step": 610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747793111116226e-05,
+      "loss": 0.8939,
+      "step": 611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874300214672144e-05,
+      "loss": 0.8315,
+      "step": 612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738202649241984e-05,
+      "loss": 0.786,
+      "step": 613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8733394623362116e-05,
+      "loss": 0.9311,
+      "step": 614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8728578073774427e-05,
+      "loss": 0.9655,
+      "step": 615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723753005179816e-05,
+      "loss": 0.9812,
+      "step": 616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8718919422287503e-05,
+      "loss": 0.9047,
+      "step": 617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714077329815027e-05,
+      "loss": 0.8981,
+      "step": 618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8709226732488216e-05,
+      "loss": 0.9014,
+      "step": 619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8704367635041204e-05,
+      "loss": 0.8224,
+      "step": 620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699500042216423e-05,
+      "loss": 0.9056,
+      "step": 621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86946239587646e-05,
+      "loss": 0.9984,
+      "step": 622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689739389444744e-05,
+      "loss": 0.886,
+      "step": 623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8684846339024145e-05,
+      "loss": 0.9171,
+      "step": 624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867994481227837e-05,
+      "loss": 0.9209,
+      "step": 625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675034813991266e-05,
+      "loss": 0.9431,
+      "step": 626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670116348954945e-05,
+      "loss": 0.8099,
+      "step": 627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8665189421969773e-05,
+      "loss": 0.8616,
+      "step": 628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.7549,
+      "step": 629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655310201395678e-05,
+      "loss": 0.8659,
+      "step": 630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8650357917448774e-05,
+      "loss": 0.9435,
+      "step": 631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645397190837063e-05,
+      "loss": 0.8345,
+      "step": 632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640428026402158e-05,
+      "loss": 0.9687,
+      "step": 633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863545042899392e-05,
+      "loss": 0.8464,
+      "step": 634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8630464403470435e-05,
+      "loss": 0.9199,
+      "step": 635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.862546995469802e-05,
+      "loss": 1.0104,
+      "step": 636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620467087551194e-05,
+      "loss": 0.846,
+      "step": 637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615455806912722e-05,
+      "loss": 0.8973,
+      "step": 638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610436117673557e-05,
+      "loss": 0.9474,
+      "step": 639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8605408024732866e-05,
+      "loss": 0.7645,
+      "step": 640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600371532998026e-05,
+      "loss": 0.86,
+      "step": 641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595326647384594e-05,
+      "loss": 0.8601,
+      "step": 642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859027337281633e-05,
+      "loss": 0.9962,
+      "step": 643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858521171422519e-05,
+      "loss": 0.902,
+      "step": 644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580141676551298e-05,
+      "loss": 0.8539,
+      "step": 645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575063264742956e-05,
+      "loss": 0.9807,
+      "step": 646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856997648375665e-05,
+      "loss": 1.0244,
+      "step": 647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564881338557033e-05,
+      "loss": 0.8049,
+      "step": 648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559777834116906e-05,
+      "loss": 0.991,
+      "step": 649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554665975417245e-05,
+      "loss": 0.951,
+      "step": 650
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549545767447174e-05,
+      "loss": 0.8113,
+      "step": 651
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8544417215203968e-05,
+      "loss": 1.0278,
+      "step": 652
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853928032369304e-05,
+      "loss": 0.9467,
+      "step": 653
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853413509792795e-05,
+      "loss": 0.871,
+      "step": 654
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8528981542930382e-05,
+      "loss": 0.9055,
+      "step": 655
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852381966373016e-05,
+      "loss": 0.8592,
+      "step": 656
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851864946536523e-05,
+      "loss": 0.7703,
+      "step": 657
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851347095288165e-05,
+      "loss": 0.8322,
+      "step": 658
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508284131333604e-05,
+      "loss": 0.9205,
+      "step": 659
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850308900578337e-05,
+      "loss": 0.9923,
+      "step": 660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497885581301344e-05,
+      "loss": 0.8297,
+      "step": 661
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492673862966016e-05,
+      "loss": 0.9456,
+      "step": 662
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848745385586398e-05,
+      "loss": 0.8967,
+      "step": 663
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8482225565089905e-05,
+      "loss": 0.9302,
+      "step": 664
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847698899574655e-05,
+      "loss": 0.8815,
+      "step": 665
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8471744152944757e-05,
+      "loss": 0.883,
+      "step": 666
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466491041803446e-05,
+      "loss": 0.8347,
+      "step": 667
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461229667449597e-05,
+      "loss": 0.748,
+      "step": 668
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845596003501826e-05,
+      "loss": 0.8917,
+      "step": 669
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845068214965255e-05,
+      "loss": 0.8755,
+      "step": 670
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445396016503628e-05,
+      "loss": 0.8555,
+      "step": 671
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8440101640730706e-05,
+      "loss": 0.8969,
+      "step": 672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434799027501048e-05,
+      "loss": 0.8844,
+      "step": 673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8429488181989946e-05,
+      "loss": 0.8788,
+      "step": 674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842416910938074e-05,
+      "loss": 0.9327,
+      "step": 675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841884181486479e-05,
+      "loss": 0.9309,
+      "step": 676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413506303641483e-05,
+      "loss": 0.8184,
+      "step": 677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408162580918228e-05,
+      "loss": 0.9796,
+      "step": 678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8402810651910444e-05,
+      "loss": 0.9362,
+      "step": 679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8397450521841564e-05,
+      "loss": 0.8776,
+      "step": 680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8392082195943017e-05,
+      "loss": 0.7915,
+      "step": 681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386705679454243e-05,
+      "loss": 0.9249,
+      "step": 682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8381320977622664e-05,
+      "loss": 1.0211,
+      "step": 683
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8375928095703703e-05,
+      "loss": 0.8227,
+      "step": 684
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837052703896075e-05,
+      "loss": 0.8878,
+      "step": 685
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83651178126652e-05,
+      "loss": 0.8655,
+      "step": 686
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8359700422096385e-05,
+      "loss": 0.957,
+      "step": 687
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835427487254164e-05,
+      "loss": 1.0276,
+      "step": 688
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8348841169296247e-05,
+      "loss": 0.8946,
+      "step": 689
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834339931766344e-05,
+      "loss": 0.8157,
+      "step": 690
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833794932295441e-05,
+      "loss": 0.8958,
+      "step": 691
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332491190488314e-05,
+      "loss": 0.7804,
+      "step": 692
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327024925592226e-05,
+      "loss": 0.8765,
+      "step": 693
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321550533601166e-05,
+      "loss": 0.7798,
+      "step": 694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8316068019858093e-05,
+      "loss": 0.891,
+      "step": 695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310577389713883e-05,
+      "loss": 0.9336,
+      "step": 696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305078648527336e-05,
+      "loss": 0.9073,
+      "step": 697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299571801665172e-05,
+      "loss": 0.7837,
+      "step": 698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829405685450202e-05,
+      "loss": 0.8249,
+      "step": 699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8288533812420413e-05,
+      "loss": 0.9916,
+      "step": 700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283002680810788e-05,
+      "loss": 1.0137,
+      "step": 701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827746346507147e-05,
+      "loss": 0.8911,
+      "step": 702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827191617060869e-05,
+      "loss": 0.8725,
+      "step": 703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266360802836542e-05,
+      "loss": 0.9255,
+      "step": 704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826079736717702e-05,
+      "loss": 0.9491,
+      "step": 705
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255225869059977e-05,
+      "loss": 0.9159,
+      "step": 706
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824964631392314e-05,
+      "loss": 0.9249,
+      "step": 707
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8244058707212106e-05,
+      "loss": 0.813,
+      "step": 708
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823846305438032e-05,
+      "loss": 0.9453,
+      "step": 709
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823285936088908e-05,
+      "loss": 0.8449,
+      "step": 710
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822724763220755e-05,
+      "loss": 0.7867,
+      "step": 711
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822162787381271e-05,
+      "loss": 0.8664,
+      "step": 712
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821600009118939e-05,
+      "loss": 0.9163,
+      "step": 713
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821036428983026e-05,
+      "loss": 0.8726,
+      "step": 714
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82047204752358e-05,
+      "loss": 0.8891,
+      "step": 715
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199068652914316e-05,
+      "loss": 0.8894,
+      "step": 716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819340882838194e-05,
+      "loss": 0.83,
+      "step": 717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8187741007162593e-05,
+      "loss": 0.8706,
+      "step": 718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8182065194788024e-05,
+      "loss": 0.8549,
+      "step": 719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817638139679777e-05,
+      "loss": 0.7793,
+      "step": 720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170689618739152e-05,
+      "loss": 0.8238,
+      "step": 721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81649898661673e-05,
+      "loss": 0.9374,
+      "step": 722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815928214464511e-05,
+      "loss": 0.8358,
+      "step": 723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153566459743272e-05,
+      "loss": 0.8931,
+      "step": 724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814784281704023e-05,
+      "loss": 0.8591,
+      "step": 725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142111222122204e-05,
+      "loss": 0.8741,
+      "step": 726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136371680583176e-05,
+      "loss": 0.7907,
+      "step": 727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8130624198024883e-05,
+      "loss": 0.8068,
+      "step": 728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124868780056814e-05,
+      "loss": 0.9721,
+      "step": 729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119105432296198e-05,
+      "loss": 0.805,
+      "step": 730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113334160368007e-05,
+      "loss": 0.756,
+      "step": 731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107554969904943e-05,
+      "loss": 0.8849,
+      "step": 732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101767866547444e-05,
+      "loss": 0.7398,
+      "step": 733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8095972855943664e-05,
+      "loss": 0.8876,
+      "step": 734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.8933,
+      "step": 735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084359135628468e-05,
+      "loss": 0.8275,
+      "step": 736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807854043725193e-05,
+      "loss": 0.9048,
+      "step": 737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072713854298856e-05,
+      "loss": 0.8743,
+      "step": 738
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8066879392455932e-05,
+      "loss": 0.9398,
+      "step": 739
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061037057417538e-05,
+      "loss": 0.9082,
+      "step": 740
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055186854885733e-05,
+      "loss": 0.8148,
+      "step": 741
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8049328790570263e-05,
+      "loss": 0.8468,
+      "step": 742
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8043462870188535e-05,
+      "loss": 0.8752,
+      "step": 743
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037589099465637e-05,
+      "loss": 0.8508,
+      "step": 744
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031707484134303e-05,
+      "loss": 0.8812,
+      "step": 745
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802581802993494e-05,
+      "loss": 0.8807,
+      "step": 746
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8019920742615596e-05,
+      "loss": 1.0211,
+      "step": 747
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8014015627931963e-05,
+      "loss": 1.0104,
+      "step": 748
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.800810269164738e-05,
+      "loss": 0.9206,
+      "step": 749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8002181939532804e-05,
+      "loss": 0.8972,
+      "step": 750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996253377366846e-05,
+      "loss": 0.8758,
+      "step": 751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990317010935716e-05,
+      "loss": 0.8722,
+      "step": 752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984372846033252e-05,
+      "loss": 0.8879,
+      "step": 753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79784208884609e-05,
+      "loss": 0.7669,
+      "step": 754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797246114402771e-05,
+      "loss": 0.845,
+      "step": 755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796649361855034e-05,
+      "loss": 0.8812,
+      "step": 756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796051831785303e-05,
+      "loss": 0.8911,
+      "step": 757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954535247767625e-05,
+      "loss": 0.9418,
+      "step": 758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948544414133534e-05,
+      "loss": 0.7427,
+      "step": 759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7942545822797755e-05,
+      "loss": 0.8318,
+      "step": 760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936539479614853e-05,
+      "loss": 0.894,
+      "step": 761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930525390446965e-05,
+      "loss": 0.7704,
+      "step": 762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924503561163775e-05,
+      "loss": 0.9062,
+      "step": 763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7918473997642535e-05,
+      "loss": 0.781,
+      "step": 764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912436705768045e-05,
+      "loss": 0.9465,
+      "step": 765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906391691432633e-05,
+      "loss": 0.8299,
+      "step": 766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900338960536178e-05,
+      "loss": 0.7877,
+      "step": 767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894278518986088e-05,
+      "loss": 0.8579,
+      "step": 768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888210372697292e-05,
+      "loss": 0.8825,
+      "step": 769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882134527592244e-05,
+      "loss": 0.9597,
+      "step": 770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7876050989600908e-05,
+      "loss": 0.8879,
+      "step": 771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786995976466076e-05,
+      "loss": 0.9313,
+      "step": 772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7863860858716768e-05,
+      "loss": 0.9307,
+      "step": 773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857754277721417e-05,
+      "loss": 0.9399,
+      "step": 774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785164002763466e-05,
+      "loss": 0.9204,
+      "step": 775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7845518114423953e-05,
+      "loss": 0.9879,
+      "step": 776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839388544064215e-05,
+      "loss": 0.8222,
+      "step": 777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7833251322537854e-05,
+      "loss": 0.7994,
+      "step": 778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782710645583473e-05,
+      "loss": 0.8919,
+      "step": 779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782095394995219e-05,
+      "loss": 0.8437,
+      "step": 780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814793810895e-05,
+      "loss": 0.9787,
+      "step": 781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7808626044675406e-05,
+      "loss": 0.8452,
+      "step": 782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802450657313086e-05,
+      "loss": 0.8276,
+      "step": 783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779626765483516e-05,
+      "loss": 0.9492,
+      "step": 784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790077043276168e-05,
+      "loss": 0.9135,
+      "step": 785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783878828678096e-05,
+      "loss": 0.8665,
+      "step": 786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777673017090344e-05,
+      "loss": 0.9424,
+      "step": 787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777145961456971e-05,
+      "loss": 0.8589,
+      "step": 788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765238627180424e-05,
+      "loss": 0.8818,
+      "step": 789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759010060994112e-05,
+      "loss": 1.0291,
+      "step": 790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7752773922089784e-05,
+      "loss": 0.8794,
+      "step": 791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7746530216553855e-05,
+      "loss": 0.9243,
+      "step": 792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740278950480122e-05,
+      "loss": 0.8896,
+      "step": 793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773402012996976e-05,
+      "loss": 0.8916,
+      "step": 794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727753761131312e-05,
+      "loss": 0.8554,
+      "step": 795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77214798500807e-05,
+      "loss": 0.8296,
+      "step": 796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7715198402941194e-05,
+      "loss": 0.8858,
+      "step": 797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708909425843426e-05,
+      "loss": 0.8445,
+      "step": 798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7702612924925377e-05,
+      "loss": 0.9593,
+      "step": 799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696308906332373e-05,
+      "loss": 0.8002,
+      "step": 800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768999737621707e-05,
+      "loss": 0.7852,
+      "step": 801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683678340739462e-05,
+      "loss": 0.8962,
+      "step": 802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677351806066863e-05,
+      "loss": 0.7847,
+      "step": 803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7671017778373914e-05,
+      "loss": 0.8376,
+      "step": 804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664676263842562e-05,
+      "loss": 0.8934,
+      "step": 805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7658327268662067e-05,
+      "loss": 0.9583,
+      "step": 806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7651970799028976e-05,
+      "loss": 0.9897,
+      "step": 807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645606861147148e-05,
+      "loss": 0.7914,
+      "step": 808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639235461227727e-05,
+      "loss": 1.0021,
+      "step": 809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7632856605489134e-05,
+      "loss": 0.8762,
+      "step": 810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626470300157064e-05,
+      "loss": 0.9121,
+      "step": 811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76200765514645e-05,
+      "loss": 0.9368,
+      "step": 812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761367536565167e-05,
+      "loss": 0.8273,
+      "step": 813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760726674896607e-05,
+      "loss": 0.9534,
+      "step": 814
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600850707662454e-05,
+      "loss": 0.7531,
+      "step": 815
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594427248002806e-05,
+      "loss": 0.8238,
+      "step": 816
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587996376256366e-05,
+      "loss": 0.9681,
+      "step": 817
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75815580986996e-05,
+      "loss": 0.9058,
+      "step": 818
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575112421616203e-05,
+      "loss": 0.9168,
+      "step": 819
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568659351297094e-05,
+      "loss": 0.8189,
+      "step": 820
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7562198894040407e-05,
+      "loss": 0.7672,
+      "step": 821
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555731056151485e-05,
+      "loss": 0.8878,
+      "step": 822
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7549255843942875e-05,
+      "loss": 0.9433,
+      "step": 823
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542773263734317e-05,
+      "loss": 0.9013,
+      "step": 824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753628332185275e-05,
+      "loss": 0.8715,
+      "step": 825
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529786024632295e-05,
+      "loss": 0.878,
+      "step": 826
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523281378414246e-05,
+      "loss": 0.8952,
+      "step": 827
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751676938954708e-05,
+      "loss": 1.016,
+      "step": 828
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751025006438643e-05,
+      "loss": 0.8219,
+      "step": 829
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750372340929509e-05,
+      "loss": 0.9787,
+      "step": 830
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7497189430643025e-05,
+      "loss": 0.8433,
+      "step": 831
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7490648134807323e-05,
+      "loss": 0.9025,
+      "step": 832
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7484099528172227e-05,
+      "loss": 0.8549,
+      "step": 833
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747754361712911e-05,
+      "loss": 0.9332,
+      "step": 834
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7470980408076484e-05,
+      "loss": 0.8809,
+      "step": 835
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464409907419976e-05,
+      "loss": 0.7651,
+      "step": 836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457832121572323e-05,
+      "loss": 0.9772,
+      "step": 837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745124705695339e-05,
+      "loss": 0.8834,
+      "step": 838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7444654719990128e-05,
+      "loss": 0.8722,
+      "step": 839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438055117116598e-05,
+      "loss": 0.868,
+      "step": 840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 1.0311,
+      "step": 841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424834139410404e-05,
+      "loss": 0.8481,
+      "step": 842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741821277748128e-05,
+      "loss": 0.7851,
+      "step": 843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741158417544897e-05,
+      "loss": 0.9481,
+      "step": 844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404948339782918e-05,
+      "loss": 0.818,
+      "step": 845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7398305276959634e-05,
+      "loss": 0.8486,
+      "step": 846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391654993462686e-05,
+      "loss": 0.9211,
+      "step": 847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738499749578268e-05,
+      "loss": 0.8176,
+      "step": 848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378332790417275e-05,
+      "loss": 0.8855,
+      "step": 849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371660883871154e-05,
+      "loss": 0.8981,
+      "step": 850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364981782656033e-05,
+      "loss": 0.8262,
+      "step": 851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7358295493290653e-05,
+      "loss": 0.8532,
+      "step": 852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7351602022300762e-05,
+      "loss": 0.9354,
+      "step": 853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344901376219126e-05,
+      "loss": 0.9321,
+      "step": 854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338193561585507e-05,
+      "loss": 0.8703,
+      "step": 855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331478584946674e-05,
+      "loss": 0.8899,
+      "step": 856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324756452856367e-05,
+      "loss": 0.9331,
+      "step": 857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.731802717187533e-05,
+      "loss": 0.896,
+      "step": 858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311290748571273e-05,
+      "loss": 0.9191,
+      "step": 859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304547189518884e-05,
+      "loss": 0.9362,
+      "step": 860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72977965012998e-05,
+      "loss": 0.836,
+      "step": 861
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291038690502632e-05,
+      "loss": 0.8712,
+      "step": 862
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284273763722943e-05,
+      "loss": 0.8716,
+      "step": 863
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277501727563227e-05,
+      "loss": 0.8627,
+      "step": 864
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727072258863293e-05,
+      "loss": 0.8541,
+      "step": 865
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726393635354842e-05,
+      "loss": 0.8674,
+      "step": 866
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257143028933004e-05,
+      "loss": 0.7886,
+      "step": 867
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250342621416897e-05,
+      "loss": 0.9405,
+      "step": 868
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243535137637227e-05,
+      "loss": 0.8212,
+      "step": 869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723672058423804e-05,
+      "loss": 0.8144,
+      "step": 870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722989896787026e-05,
+      "loss": 0.8272,
+      "step": 871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223070295191733e-05,
+      "loss": 0.8082,
+      "step": 872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7216234572867165e-05,
+      "loss": 0.8931,
+      "step": 873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720939180756816e-05,
+      "loss": 0.8886,
+      "step": 874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720254200597319e-05,
+      "loss": 0.9463,
+      "step": 875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7195685174767594e-05,
+      "loss": 0.8081,
+      "step": 876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718882132064357e-05,
+      "loss": 0.8314,
+      "step": 877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181950450300176e-05,
+      "loss": 0.8235,
+      "step": 878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717507257044331e-05,
+      "loss": 0.8653,
+      "step": 879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7168187687785726e-05,
+      "loss": 0.9553,
+      "step": 880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7161295809046986e-05,
+      "loss": 0.8997,
+      "step": 881
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154396940953507e-05,
+      "loss": 0.9386,
+      "step": 882
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7147491090238516e-05,
+      "loss": 0.8503,
+      "step": 883
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714057826364205e-05,
+      "loss": 0.8732,
+      "step": 884
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7133658467910968e-05,
+      "loss": 0.7526,
+      "step": 885
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126731709798924e-05,
+      "loss": 0.9024,
+      "step": 886
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7119797996066355e-05,
+      "loss": 0.9442,
+      "step": 887
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112857333480506e-05,
+      "loss": 0.8915,
+      "step": 888
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7105909728815398e-05,
+      "loss": 0.8478,
+      "step": 889
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709895518885182e-05,
+      "loss": 0.9043,
+      "step": 890
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7091993720377336e-05,
+      "loss": 0.7999,
+      "step": 891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085025330186272e-05,
+      "loss": 0.8695,
+      "step": 892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707805002507971e-05,
+      "loss": 0.8478,
+      "step": 893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7071067811865477e-05,
+      "loss": 0.8762,
+      "step": 894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064078697358147e-05,
+      "loss": 0.8635,
+      "step": 895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057082688379028e-05,
+      "loss": 0.7763,
+      "step": 896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7050079791756157e-05,
+      "loss": 0.8598,
+      "step": 897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043070014324287e-05,
+      "loss": 0.9029,
+      "step": 898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036053362924896e-05,
+      "loss": 0.7953,
+      "step": 899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7029029844406168e-05,
+      "loss": 0.9472,
+      "step": 900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702199946562299e-05,
+      "loss": 0.8431,
+      "step": 901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701496223343694e-05,
+      "loss": 0.8384,
+      "step": 902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7007918154716286e-05,
+      "loss": 0.8355,
+      "step": 903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000867236335987e-05,
+      "loss": 0.9277,
+      "step": 904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6993809485177665e-05,
+      "loss": 0.8082,
+      "step": 905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986744908129618e-05,
+      "loss": 0.7883,
+      "step": 906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69796735120868e-05,
+      "loss": 0.7885,
+      "step": 907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972595303950835e-05,
+      "loss": 0.8616,
+      "step": 908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965510290629973e-05,
+      "loss": 0.8334,
+      "step": 909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958418479039124e-05,
+      "loss": 0.879,
+      "step": 910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6951319876099825e-05,
+      "loss": 0.8494,
+      "step": 911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694421448874024e-05,
+      "loss": 0.8354,
+      "step": 912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6937102323895167e-05,
+      "loss": 0.8525,
+      "step": 913
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6929983388506002e-05,
+      "loss": 0.8843,
+      "step": 914
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692285768952076e-05,
+      "loss": 0.8363,
+      "step": 915
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6915725233894047e-05,
+      "loss": 0.9864,
+      "step": 916
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690858602858708e-05,
+      "loss": 0.8729,
+      "step": 917
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901440080567643e-05,
+      "loss": 0.8305,
+      "step": 918
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689428739681012e-05,
+      "loss": 0.9069,
+      "step": 919
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688712798429546e-05,
+      "loss": 0.9194,
+      "step": 920
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6879961850011174e-05,
+      "loss": 0.8873,
+      "step": 921
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687278900095134e-05,
+      "loss": 0.7702,
+      "step": 922
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865609444116594e-05,
+      "loss": 0.9555,
+      "step": 923
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6858423186514108e-05,
+      "loss": 0.8045,
+      "step": 924
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.68512302351576e-05,
+      "loss": 0.9009,
+      "step": 925
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844030597067323e-05,
+      "loss": 0.9709,
+      "step": 926
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836824279270053e-05,
+      "loss": 0.7517,
+      "step": 927
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682961128879908e-05,
+      "loss": 0.8751,
+      "step": 928
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682239163269422e-05,
+      "loss": 0.7841,
+      "step": 929
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6815165318001785e-05,
+      "loss": 0.8513,
+      "step": 930
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6807932351774585e-05,
+      "loss": 0.7958,
+      "step": 931
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6800692741071926e-05,
+      "loss": 0.8805,
+      "step": 932
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793446492959596e-05,
+      "loss": 0.9034,
+      "step": 933
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6786193614509863e-05,
+      "loss": 0.8157,
+      "step": 934
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6778934112801467e-05,
+      "loss": 0.826,
+      "step": 935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771667994919608e-05,
+      "loss": 0.8444,
+      "step": 936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764395267955947e-05,
+      "loss": 0.8822,
+      "step": 937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757115939008596e-05,
+      "loss": 1.0523,
+      "step": 938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6749830015182106e-05,
+      "loss": 0.876,
+      "step": 939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742537503587473e-05,
+      "loss": 0.8619,
+      "step": 940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673523841134211e-05,
+      "loss": 0.7732,
+      "step": 941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6727932745569866e-05,
+      "loss": 0.873,
+      "step": 942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6720620513400993e-05,
+      "loss": 0.7655,
+      "step": 943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6713301721972165e-05,
+      "loss": 0.8505,
+      "step": 944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6705976378426447e-05,
+      "loss": 0.8754,
+      "step": 945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.66986444899133e-05,
+      "loss": 0.7685,
+      "step": 946
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.8648,
+      "step": 947
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6683961106614524e-05,
+      "loss": 0.9419,
+      "step": 948
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667660962615973e-05,
+      "loss": 0.9197,
+      "step": 949
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666925162939917e-05,
+      "loss": 0.9505,
+      "step": 950
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661887123514183e-05,
+      "loss": 0.9083,
+      "step": 951
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665451611569245e-05,
+      "loss": 0.8068,
+      "step": 952
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664713861312801e-05,
+      "loss": 0.7792,
+      "step": 953
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6639754623021227e-05,
+      "loss": 0.9473,
+      "step": 954
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66323641525788e-05,
+      "loss": 0.8764,
+      "step": 955
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624967209013766e-05,
+      "loss": 0.8424,
+      "step": 956
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6617563799545462e-05,
+      "loss": 0.8822,
+      "step": 957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610153931399543e-05,
+      "loss": 0.9015,
+      "step": 958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602737611807975e-05,
+      "loss": 0.8078,
+      "step": 959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659531484800901e-05,
+      "loss": 0.8447,
+      "step": 960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587885647247186e-05,
+      "loss": 0.8438,
+      "step": 961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580450016773338e-05,
+      "loss": 0.8813,
+      "step": 962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657300796384457e-05,
+      "loss": 0.9303,
+      "step": 963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6565559495724244e-05,
+      "loss": 0.9111,
+      "step": 964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558104619682004e-05,
+      "loss": 0.7703,
+      "step": 965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6550643342993725e-05,
+      "loss": 0.752,
+      "step": 966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654317567294155e-05,
+      "loss": 0.8404,
+      "step": 967
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6535701616813845e-05,
+      "loss": 0.8686,
+      "step": 968
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528221181905217e-05,
+      "loss": 0.8744,
+      "step": 969
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65207343755165e-05,
+      "loss": 0.8815,
+      "step": 970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6513241204954745e-05,
+      "loss": 0.8185,
+      "step": 971
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6505741677533204e-05,
+      "loss": 0.7081,
+      "step": 972
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498235800571353e-05,
+      "loss": 0.9126,
+      "step": 973
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649072358139485e-05,
+      "loss": 0.9255,
+      "step": 974
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.648320502733555e-05,
+      "loss": 0.8679,
+      "step": 975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475680145731484e-05,
+      "loss": 0.8651,
+      "step": 976
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6468148943926866e-05,
+      "loss": 0.8401,
+      "step": 977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460611429272076e-05,
+      "loss": 0.7976,
+      "step": 978
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6453067609123656e-05,
+      "loss": 0.8242,
+      "step": 979
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644551749084429e-05,
+      "loss": 0.8724,
+      "step": 980
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437961081802835e-05,
+      "loss": 0.8574,
+      "step": 981
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6430398389374262e-05,
+      "loss": 0.9977,
+      "step": 982
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422829420939688e-05,
+      "loss": 0.8414,
+      "step": 983
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6415254183886356e-05,
+      "loss": 0.8441,
+      "step": 984
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6407672685607616e-05,
+      "loss": 0.8278,
+      "step": 985
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640008493350294e-05,
+      "loss": 0.9321,
+      "step": 986
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.63924909349779e-05,
+      "loss": 0.8239,
+      "step": 987
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638489069744417e-05,
+      "loss": 0.7498,
+      "step": 988
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377284228319496e-05,
+      "loss": 0.9355,
+      "step": 989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636967153502772e-05,
+      "loss": 0.8964,
+      "step": 990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362052624998767e-05,
+      "loss": 0.7296,
+      "step": 991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635442750566861e-05,
+      "loss": 0.8955,
+      "step": 992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6346796184479293e-05,
+      "loss": 0.8829,
+      "step": 993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6339158668878906e-05,
+      "loss": 0.8592,
+      "step": 994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331514966321596e-05,
+      "loss": 0.7741,
+      "step": 995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632386508426754e-05,
+      "loss": 0.8125,
+      "step": 996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631620903018295e-05,
+      "loss": 0.6888,
+      "step": 997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6308546811540055e-05,
+      "loss": 0.8037,
+      "step": 998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6300878435817115e-05,
+      "loss": 0.7843,
+      "step": 999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6293203910498375e-05,
+      "loss": 0.823,
+      "step": 1000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628552324307411e-05,
+      "loss": 0.8305,
+      "step": 1001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627783644104057e-05,
+      "loss": 0.8809,
+      "step": 1002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62701435119e-05,
+      "loss": 0.9623,
+      "step": 1003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6262444463160627e-05,
+      "loss": 0.8747,
+      "step": 1004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625473930233664e-05,
+      "loss": 1.0057,
+      "step": 1005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6247028036948206e-05,
+      "loss": 0.8932,
+      "step": 1006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239310674521443e-05,
+      "loss": 0.7859,
+      "step": 1007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6231587222588424e-05,
+      "loss": 0.8808,
+      "step": 1008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622385768868716e-05,
+      "loss": 0.9122,
+      "step": 1009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62161220803616e-05,
+      "loss": 0.9048,
+      "step": 1010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6208380405161623e-05,
+      "loss": 0.9233,
+      "step": 1011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6200632670643028e-05,
+      "loss": 0.9034,
+      "step": 1012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192878884367532e-05,
+      "loss": 0.8262,
+      "step": 1013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185119053902748e-05,
+      "loss": 0.8252,
+      "step": 1014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61773531868222e-05,
+      "loss": 0.8553,
+      "step": 1015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61695812907053e-05,
+      "loss": 1.0597,
+      "step": 1016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.616180337313733e-05,
+      "loss": 0.9358,
+      "step": 1017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6154019441709478e-05,
+      "loss": 0.8146,
+      "step": 1018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146229504018777e-05,
+      "loss": 0.8994,
+      "step": 1019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6138433567668127e-05,
+      "loss": 0.8234,
+      "step": 1020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613063164026629e-05,
+      "loss": 0.8393,
+      "step": 1021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122823729427878e-05,
+      "loss": 0.8254,
+      "step": 1022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6115009842773322e-05,
+      "loss": 0.8171,
+      "step": 1023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6107189987928906e-05,
+      "loss": 0.8455,
+      "step": 1024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6099364172526732e-05,
+      "loss": 0.7612,
+      "step": 1025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091532404204716e-05,
+      "loss": 0.9601,
+      "step": 1026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6083694690606592e-05,
+      "loss": 0.7824,
+      "step": 1027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6075851039381888e-05,
+      "loss": 0.8815,
+      "step": 1028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068001458185934e-05,
+      "loss": 0.7701,
+      "step": 1029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6060145954679846e-05,
+      "loss": 0.9174,
+      "step": 1030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605228453653051e-05,
+      "loss": 0.9313,
+      "step": 1031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044417211410606e-05,
+      "loss": 0.9573,
+      "step": 1032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036543986998558e-05,
+      "loss": 0.8465,
+      "step": 1033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602866487097855e-05,
+      "loss": 0.7907,
+      "step": 1034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020779871040538e-05,
+      "loss": 0.8897,
+      "step": 1035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012888994880194e-05,
+      "loss": 0.8811,
+      "step": 1036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600499225019894e-05,
+      "loss": 0.7693,
+      "step": 1037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599708964470392e-05,
+      "loss": 0.8879,
+      "step": 1038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5989181186108003e-05,
+      "loss": 0.8563,
+      "step": 1039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981266882129763e-05,
+      "loss": 0.8064,
+      "step": 1040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5973346740493486e-05,
+      "loss": 1.0429,
+      "step": 1041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965420768929153e-05,
+      "loss": 0.9139,
+      "step": 1042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595748897517243e-05,
+      "loss": 0.8332,
+      "step": 1043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5949551366964675e-05,
+      "loss": 0.894,
+      "step": 1044
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941607952052916e-05,
+      "loss": 0.7479,
+      "step": 1045
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5933658738189845e-05,
+      "loss": 0.7404,
+      "step": 1046
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5925703733133823e-05,
+      "loss": 0.8264,
+      "step": 1047
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591774294464885e-05,
+      "loss": 0.9119,
+      "step": 1048
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909776380504583e-05,
+      "loss": 0.8955,
+      "step": 1049
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901804048476314e-05,
+      "loss": 0.923,
+      "step": 1050
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589382595634495e-05,
+      "loss": 0.8261,
+      "step": 1051
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885842111897038e-05,
+      "loss": 0.838,
+      "step": 1052
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877852522924733e-05,
+      "loss": 0.7582,
+      "step": 1053
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5869857197225792e-05,
+      "loss": 0.7871,
+      "step": 1054
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586185614260358e-05,
+      "loss": 0.8511,
+      "step": 1055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853849366867047e-05,
+      "loss": 0.8067,
+      "step": 1056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584583687783072e-05,
+      "loss": 0.9325,
+      "step": 1057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583781868331472e-05,
+      "loss": 0.8285,
+      "step": 1058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829794791144723e-05,
+      "loss": 0.9053,
+      "step": 1059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821765209151966e-05,
+      "loss": 0.9471,
+      "step": 1060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581372994517325e-05,
+      "loss": 0.8102,
+      "step": 1061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5805689007050906e-05,
+      "loss": 0.8686,
+      "step": 1062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797642402632816e-05,
+      "loss": 0.8385,
+      "step": 1063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5789590139772384e-05,
+      "loss": 0.9445,
+      "step": 1064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5781532226328544e-05,
+      "loss": 0.7875,
+      "step": 1065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577346867016574e-05,
+      "loss": 0.8958,
+      "step": 1066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576539947915392e-05,
+      "loss": 0.9194,
+      "step": 1067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575732466116854e-05,
+      "loss": 0.9028,
+      "step": 1068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749244224090537e-05,
+      "loss": 0.8315,
+      "step": 1069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574115817580635e-05,
+      "loss": 0.8055,
+      "step": 1070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733066524207875e-05,
+      "loss": 0.9175,
+      "step": 1071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5724969277192482e-05,
+      "loss": 0.8946,
+      "step": 1072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716866442663015e-05,
+      "loss": 0.9275,
+      "step": 1073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5708758028527754e-05,
+      "loss": 0.802,
+      "step": 1074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5700644042700432e-05,
+      "loss": 0.7945,
+      "step": 1075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569252449310022e-05,
+      "loss": 0.926,
+      "step": 1076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5684399387651725e-05,
+      "loss": 0.7786,
+      "step": 1077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676268734284964e-05,
+      "loss": 1.0172,
+      "step": 1078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.566813254093538e-05,
+      "loss": 0.8698,
+      "step": 1079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.565999081554381e-05,
+      "loss": 0.7814,
+      "step": 1080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5651843566056512e-05,
+      "loss": 0.8537,
+      "step": 1081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564369080042511e-05,
+      "loss": 0.7063,
+      "step": 1082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5635532526606625e-05,
+      "loss": 0.9053,
+      "step": 1083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562736875256346e-05,
+      "loss": 0.8166,
+      "step": 1084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561919948626337e-05,
+      "loss": 0.8777,
+      "step": 1085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611024735679484e-05,
+      "loss": 0.8937,
+      "step": 1086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560284450879028e-05,
+      "loss": 0.7718,
+      "step": 1087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5594658813579574e-05,
+      "loss": 0.7814,
+      "step": 1088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586467658036526e-05,
+      "loss": 0.7766,
+      "step": 1089
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557827105015563e-05,
+      "loss": 0.8179,
+      "step": 1090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570068997936686e-05,
+      "loss": 0.8748,
+      "step": 1091
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561861509384826e-05,
+      "loss": 0.798,
+      "step": 1092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553648592510474e-05,
+      "loss": 0.7538,
+      "step": 1093
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.554543025532935e-05,
+      "loss": 0.7542,
+      "step": 1094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5537206505862486e-05,
+      "loss": 0.6429,
+      "step": 1095
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552897735213617e-05,
+      "loss": 0.6261,
+      "step": 1096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520742802181977e-05,
+      "loss": 0.6181,
+      "step": 1097
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551250286403675e-05,
+      "loss": 0.6424,
+      "step": 1098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5504257545742585e-05,
+      "loss": 0.6015,
+      "step": 1099
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549600685534684e-05,
+      "loss": 0.6441,
+      "step": 1100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5487750800902094e-05,
+      "loss": 0.6182,
+      "step": 1101
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5479489390466186e-05,
+      "loss": 0.5652,
+      "step": 1102
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5471222632102168e-05,
+      "loss": 0.6197,
+      "step": 1103
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5462950533878318e-05,
+      "loss": 0.5873,
+      "step": 1104
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5454673103868122e-05,
+      "loss": 0.6275,
+      "step": 1105
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5446390350150272e-05,
+      "loss": 0.6274,
+      "step": 1106
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5438102280808653e-05,
+      "loss": 0.5909,
+      "step": 1107
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542980890393234e-05,
+      "loss": 0.6239,
+      "step": 1108
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5421510227615587e-05,
+      "loss": 0.5954,
+      "step": 1109
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5413206259957823e-05,
+      "loss": 0.5925,
+      "step": 1110
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5404897009063636e-05,
+      "loss": 0.6054,
+      "step": 1111
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539658248304278e-05,
+      "loss": 0.6418,
+      "step": 1112
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538826269001014e-05,
+      "loss": 0.6856,
+      "step": 1113
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5379937638085768e-05,
+      "loss": 0.6441,
+      "step": 1114
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537160733539482e-05,
+      "loss": 0.6249,
+      "step": 1115
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5363271790067596e-05,
+      "loss": 0.5943,
+      "step": 1116
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5354931010239506e-05,
+      "loss": 0.6007,
+      "step": 1117
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346585004051073e-05,
+      "loss": 0.5367,
+      "step": 1118
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533823377964791e-05,
+      "loss": 0.6473,
+      "step": 1119
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5329877345180736e-05,
+      "loss": 0.6693,
+      "step": 1120
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5321515708805342e-05,
+      "loss": 0.5849,
+      "step": 1121
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313148878682617e-05,
+      "loss": 0.629,
+      "step": 1122
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5304776862978496e-05,
+      "loss": 0.5427,
+      "step": 1123
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296399669863987e-05,
+      "loss": 0.6236,
+      "step": 1124
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5288017307515142e-05,
+      "loss": 0.5716,
+      "step": 1125
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279629784113078e-05,
+      "loss": 0.6492,
+      "step": 1126
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5271237107843925e-05,
+      "loss": 0.6155,
+      "step": 1127
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262839286898857e-05,
+      "loss": 0.5674,
+      "step": 1128
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5254436329474062e-05,
+      "loss": 0.6062,
+      "step": 1129
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.524602824377075e-05,
+      "loss": 0.6233,
+      "step": 1130
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237615037995129e-05,
+      "loss": 0.5634,
+      "step": 1131
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5229196720358401e-05,
+      "loss": 0.5728,
+      "step": 1132
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5220773299076767e-05,
+      "loss": 0.5773,
+      "step": 1133
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.52123447823714e-05,
+      "loss": 0.672,
+      "step": 1134
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203911178468453e-05,
+      "loss": 0.6154,
+      "step": 1135
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5195472495599042e-05,
+      "loss": 0.5489,
+      "step": 1136
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187028741999234e-05,
+      "loss": 0.6275,
+      "step": 1137
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178579925910053e-05,
+      "loss": 0.597,
+      "step": 1138
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.517012605557746e-05,
+      "loss": 0.592,
+      "step": 1139
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161667139252354e-05,
+      "loss": 0.5233,
+      "step": 1140
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5153203185190548e-05,
+      "loss": 0.5974,
+      "step": 1141
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144734201652784e-05,
+      "loss": 0.6191,
+      "step": 1142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136260196904704e-05,
+      "loss": 0.5649,
+      "step": 1143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127781179216855e-05,
+      "loss": 0.6363,
+      "step": 1144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5119297156864675e-05,
+      "loss": 0.6225,
+      "step": 1145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511080813812849e-05,
+      "loss": 0.5761,
+      "step": 1146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102314131293494e-05,
+      "loss": 0.6143,
+      "step": 1147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093815144649755e-05,
+      "loss": 0.6287,
+      "step": 1148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5085311186492206e-05,
+      "loss": 0.5915,
+      "step": 1149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5076802265120625e-05,
+      "loss": 0.6156,
+      "step": 1150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068288388839634e-05,
+      "loss": 0.6217,
+      "step": 1151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5059769565958692e-05,
+      "loss": 0.6554,
+      "step": 1152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5051245804792094e-05,
+      "loss": 0.6426,
+      "step": 1153
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504271711365894e-05,
+      "loss": 0.6078,
+      "step": 1154
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034183500883153e-05,
+      "loss": 0.6308,
+      "step": 1155
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5025644974793457e-05,
+      "loss": 0.5335,
+      "step": 1156
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5017101543723364e-05,
+      "loss": 0.5726,
+      "step": 1157
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008553216011183e-05,
+      "loss": 0.6941,
+      "step": 1158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.5997,
+      "step": 1159
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499144190403767e-05,
+      "loss": 0.6515,
+      "step": 1160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982878936476808e-05,
+      "loss": 0.5842,
+      "step": 1161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497431110567479e-05,
+      "loss": 0.5958,
+      "step": 1162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965738419993733e-05,
+      "loss": 0.5885,
+      "step": 1163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957160887800494e-05,
+      "loss": 0.5913,
+      "step": 1164
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948578517466665e-05,
+      "loss": 0.5667,
+      "step": 1165
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4939991317368547e-05,
+      "loss": 0.5438,
+      "step": 1166
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931399295887172e-05,
+      "loss": 0.6043,
+      "step": 1167
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922802461408265e-05,
+      "loss": 0.5956,
+      "step": 1168
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4914200822322255e-05,
+      "loss": 0.6591,
+      "step": 1169
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4905594387024253e-05,
+      "loss": 0.5559,
+      "step": 1170
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489698316391406e-05,
+      "loss": 0.6296,
+      "step": 1171
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4888367161396144e-05,
+      "loss": 0.6021,
+      "step": 1172
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487974638787964e-05,
+      "loss": 0.6283,
+      "step": 1173
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871120851778338e-05,
+      "loss": 0.5892,
+      "step": 1174
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4862490561510675e-05,
+      "loss": 0.5243,
+      "step": 1175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4853855525499734e-05,
+      "loss": 0.6421,
+      "step": 1176
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845215752173225e-05,
+      "loss": 0.6515,
+      "step": 1177
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483657124996348e-05,
+      "loss": 0.5871,
+      "step": 1178
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.482792202730745e-05,
+      "loss": 0.596,
+      "step": 1179
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819268092646693e-05,
+      "loss": 0.5838,
+      "step": 1180
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810609454427362e-05,
+      "loss": 0.5954,
+      "step": 1181
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801946121100206e-05,
+      "loss": 0.5829,
+      "step": 1182
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4793278101120551e-05,
+      "loss": 0.676,
+      "step": 1183
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4784605402948302e-05,
+      "loss": 0.6389,
+      "step": 1184
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775928035047928e-05,
+      "loss": 0.5919,
+      "step": 1185
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4767246005888453e-05,
+      "loss": 0.494,
+      "step": 1186
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758559323943455e-05,
+      "loss": 0.6851,
+      "step": 1187
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4749867997691049e-05,
+      "loss": 0.7018,
+      "step": 1188
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4741172035613884e-05,
+      "loss": 0.5865,
+      "step": 1189
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4732471446199137e-05,
+      "loss": 0.6103,
+      "step": 1190
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723766237938495e-05,
+      "loss": 0.6375,
+      "step": 1191
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.471505641932816e-05,
+      "loss": 0.5902,
+      "step": 1192
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706341998868823e-05,
+      "loss": 0.5193,
+      "step": 1193
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469762298506568e-05,
+      "loss": 0.5889,
+      "step": 1194
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.46888993864284e-05,
+      "loss": 0.6091,
+      "step": 1195
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680171211471134e-05,
+      "loss": 0.5604,
+      "step": 1196
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467143846871249e-05,
+      "loss": 0.6024,
+      "step": 1197
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662701166675542e-05,
+      "loss": 0.5114,
+      "step": 1198
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653959313887813e-05,
+      "loss": 0.592,
+      "step": 1199
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645212918881265e-05,
+      "loss": 0.6572,
+      "step": 1200
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636461990192293e-05,
+      "loss": 0.6406,
+      "step": 1201
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627706536361717e-05,
+      "loss": 0.6268,
+      "step": 1202
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4618946565934775e-05,
+      "loss": 0.5932,
+      "step": 1203
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610182087461116e-05,
+      "loss": 0.678,
+      "step": 1204
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4601413109494781e-05,
+      "loss": 0.6354,
+      "step": 1205
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4592639640594211e-05,
+      "loss": 0.5767,
+      "step": 1206
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4583861689322219e-05,
+      "loss": 0.6367,
+      "step": 1207
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4575079264246005e-05,
+      "loss": 0.5704,
+      "step": 1208
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4566292373937133e-05,
+      "loss": 0.5991,
+      "step": 1209
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557501026971516e-05,
+      "loss": 0.6078,
+      "step": 1210
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548705231929426e-05,
+      "loss": 0.556,
+      "step": 1211
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539904997395468e-05,
+      "loss": 0.5778,
+      "step": 1212
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4531100331958594e-05,
+      "loss": 0.5906,
+      "step": 1213
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522291244212066e-05,
+      "loss": 0.5856,
+      "step": 1214
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4513477742753465e-05,
+      "loss": 0.615,
+      "step": 1215
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4504659836184688e-05,
+      "loss": 0.5767,
+      "step": 1216
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4495837533111921e-05,
+      "loss": 0.5416,
+      "step": 1217
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4487010842145648e-05,
+      "loss": 0.5684,
+      "step": 1218
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4478179771900634e-05,
+      "loss": 0.5941,
+      "step": 1219
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469344330995916e-05,
+      "loss": 0.5897,
+      "step": 1220
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4460504528054796e-05,
+      "loss": 0.6486,
+      "step": 1221
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4451660371704837e-05,
+      "loss": 0.5671,
+      "step": 1222
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4442811870577851e-05,
+      "loss": 0.5993,
+      "step": 1223
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4433959033309888e-05,
+      "loss": 0.6196,
+      "step": 1224
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425101868541228e-05,
+      "loss": 0.573,
+      "step": 1225
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441624038491638e-05,
+      "loss": 0.6418,
+      "step": 1226
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407374591084064e-05,
+      "loss": 0.6185,
+      "step": 1227
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439850449569721e-05,
+      "loss": 0.5481,
+      "step": 1228
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389630107412942e-05,
+      "loss": 0.5965,
+      "step": 1229
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4380751434892574e-05,
+      "loss": 0.6243,
+      "step": 1230
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4371868486801611e-05,
+      "loss": 0.5084,
+      "step": 1231
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4362981271809721e-05,
+      "loss": 0.4951,
+      "step": 1232
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354089798590735e-05,
+      "loss": 0.6734,
+      "step": 1233
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345194075822645e-05,
+      "loss": 0.5744,
+      "step": 1234
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4336294112187595e-05,
+      "loss": 0.6422,
+      "step": 1235
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4327389916371856e-05,
+      "loss": 0.5602,
+      "step": 1236
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318481497065842e-05,
+      "loss": 0.6755,
+      "step": 1237
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4309568862964077e-05,
+      "loss": 0.5762,
+      "step": 1238
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4300652022765207e-05,
+      "loss": 0.6239,
+      "step": 1239
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429173098517198e-05,
+      "loss": 0.6075,
+      "step": 1240
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.428280575889125e-05,
+      "loss": 0.5613,
+      "step": 1241
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427387635263394e-05,
+      "loss": 0.6481,
+      "step": 1242
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4264942775115065e-05,
+      "loss": 0.6014,
+      "step": 1243
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4256005035053712e-05,
+      "loss": 0.5897,
+      "step": 1244
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424706314117303e-05,
+      "loss": 0.6343,
+      "step": 1245
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238117102200215e-05,
+      "loss": 0.6137,
+      "step": 1246
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229166926866517e-05,
+      "loss": 0.6225,
+      "step": 1247
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422021262390722e-05,
+      "loss": 0.6179,
+      "step": 1248
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4211254202061632e-05,
+      "loss": 0.6588,
+      "step": 1249
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202291670073092e-05,
+      "loss": 0.6308,
+      "step": 1250
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.419332503668894e-05,
+      "loss": 0.5565,
+      "step": 1251
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418435431066052e-05,
+      "loss": 0.6504,
+      "step": 1252
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175379500743174e-05,
+      "loss": 0.5707,
+      "step": 1253
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4166400615696232e-05,
+      "loss": 0.6189,
+      "step": 1254
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4157417664282994e-05,
+      "loss": 0.6707,
+      "step": 1255
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414843065527074e-05,
+      "loss": 0.5634,
+      "step": 1256
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139439597430693e-05,
+      "loss": 0.6343,
+      "step": 1257
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4130444499538045e-05,
+      "loss": 0.6278,
+      "step": 1258
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4121445370371922e-05,
+      "loss": 0.6435,
+      "step": 1259
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112442218715385e-05,
+      "loss": 0.6017,
+      "step": 1260
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4103435053355424e-05,
+      "loss": 0.6156,
+      "step": 1261
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409442388308294e-05,
+      "loss": 0.5321,
+      "step": 1262
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.408540871669275e-05,
+      "loss": 0.6511,
+      "step": 1263
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407638956298357e-05,
+      "loss": 0.542,
+      "step": 1264
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.5906,
+      "step": 1265
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058339328822541e-05,
+      "loss": 0.5925,
+      "step": 1266
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4049308265987544e-05,
+      "loss": 0.5768,
+      "step": 1267
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4040273251067245e-05,
+      "loss": 0.6003,
+      "step": 1268
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4031234292879726e-05,
+      "loss": 0.5531,
+      "step": 1269
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022191400246923e-05,
+      "loss": 0.628,
+      "step": 1270
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4013144581994609e-05,
+      "loss": 0.6169,
+      "step": 1271
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4004093846952397e-05,
+      "loss": 0.6229,
+      "step": 1272
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995039203953712e-05,
+      "loss": 0.6103,
+      "step": 1273
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3985980661835793e-05,
+      "loss": 0.6273,
+      "step": 1274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976918229439698e-05,
+      "loss": 0.5998,
+      "step": 1275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3967851915610263e-05,
+      "loss": 0.5758,
+      "step": 1276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958781729196125e-05,
+      "loss": 0.6161,
+      "step": 1277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.39497076790497e-05,
+      "loss": 0.5963,
+      "step": 1278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394062977402717e-05,
+      "loss": 0.6201,
+      "step": 1279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393154802298848e-05,
+      "loss": 0.5594,
+      "step": 1280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922462434797335e-05,
+      "loss": 0.5543,
+      "step": 1281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3913373018321176e-05,
+      "loss": 0.5666,
+      "step": 1282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904279782431187e-05,
+      "loss": 0.6456,
+      "step": 1283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3895182736002276e-05,
+      "loss": 0.5596,
+      "step": 1284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.388608188791307e-05,
+      "loss": 0.6085,
+      "step": 1285
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3876977247045912e-05,
+      "loss": 0.6418,
+      "step": 1286
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867868822286838e-05,
+      "loss": 0.5412,
+      "step": 1287
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858756622525587e-05,
+      "loss": 0.683,
+      "step": 1288
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3849640656655572e-05,
+      "loss": 0.6661,
+      "step": 1289
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840520933573889e-05,
+      "loss": 0.5821,
+      "step": 1290
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3831397462181298e-05,
+      "loss": 0.6213,
+      "step": 1291
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3822270251382217e-05,
+      "loss": 0.6255,
+      "step": 1292
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813139310084715e-05,
+      "loss": 0.5908,
+      "step": 1293
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3804004647200504e-05,
+      "loss": 0.6125,
+      "step": 1294
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3794866271644922e-05,
+      "loss": 0.5926,
+      "step": 1295
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3785724192336938e-05,
+      "loss": 0.5526,
+      "step": 1296
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776578418199129e-05,
+      "loss": 0.5551,
+      "step": 1297
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3767428958157684e-05,
+      "loss": 0.5408,
+      "step": 1298
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758275821142382e-05,
+      "loss": 0.6619,
+      "step": 1299
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749119016086598e-05,
+      "loss": 0.5118,
+      "step": 1300
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739958551927287e-05,
+      "loss": 0.6138,
+      "step": 1301
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373079443760497e-05,
+      "loss": 0.5187,
+      "step": 1302
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721626682063733e-05,
+      "loss": 0.6259,
+      "step": 1303
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712455294251214e-05,
+      "loss": 0.6425,
+      "step": 1304
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703280283118601e-05,
+      "loss": 0.6486,
+      "step": 1305
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3694101657620615e-05,
+      "loss": 0.5951,
+      "step": 1306
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3684919426715504e-05,
+      "loss": 0.6734,
+      "step": 1307
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367573359936503e-05,
+      "loss": 0.6261,
+      "step": 1308
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3666544184534483e-05,
+      "loss": 0.6014,
+      "step": 1309
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657351191192634e-05,
+      "loss": 0.6435,
+      "step": 1310
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3648154628311754e-05,
+      "loss": 0.6664,
+      "step": 1311
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36389545048676e-05,
+      "loss": 0.551,
+      "step": 1312
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3629750829839403e-05,
+      "loss": 0.6431,
+      "step": 1313
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620543612209861e-05,
+      "loss": 0.6533,
+      "step": 1314
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361133286096513e-05,
+      "loss": 0.5569,
+      "step": 1315
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602118585094807e-05,
+      "loss": 0.6699,
+      "step": 1316
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592900793591941e-05,
+      "loss": 0.6154,
+      "step": 1317
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3583679495453e-05,
+      "loss": 0.6028,
+      "step": 1318
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574454699677893e-05,
+      "loss": 0.5176,
+      "step": 1319
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565226415269918e-05,
+      "loss": 0.6193,
+      "step": 1320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555994651235795e-05,
+      "loss": 0.614,
+      "step": 1321
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3546759416585635e-05,
+      "loss": 0.5878,
+      "step": 1322
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537520720332943e-05,
+      "loss": 0.6822,
+      "step": 1323
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3528278571494583e-05,
+      "loss": 0.5868,
+      "step": 1324
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3519032979090816e-05,
+      "loss": 0.5612,
+      "step": 1325
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.350978395214524e-05,
+      "loss": 0.6275,
+      "step": 1326
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3500531499684819e-05,
+      "loss": 0.6585,
+      "step": 1327
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3491275630739856e-05,
+      "loss": 0.6416,
+      "step": 1328
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348201635434399e-05,
+      "loss": 0.6592,
+      "step": 1329
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347275367953418e-05,
+      "loss": 0.6235,
+      "step": 1330
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346348761535071e-05,
+      "loss": 0.6497,
+      "step": 1331
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454218170837165e-05,
+      "loss": 0.6575,
+      "step": 1332
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3444945355040437e-05,
+      "loss": 0.6039,
+      "step": 1333
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34356691770107e-05,
+      "loss": 0.5956,
+      "step": 1334
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3426389645801415e-05,
+      "loss": 0.5877,
+      "step": 1335
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417106770469311e-05,
+      "loss": 0.703,
+      "step": 1336
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3407820560074385e-05,
+      "loss": 0.6432,
+      "step": 1337
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3398531023679889e-05,
+      "loss": 0.5332,
+      "step": 1338
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389238170352318e-05,
+      "loss": 0.594,
+      "step": 1339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3379942009161407e-05,
+      "loss": 0.6159,
+      "step": 1340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3370642549180117e-05,
+      "loss": 0.637,
+      "step": 1341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3361339799484634e-05,
+      "loss": 0.6058,
+      "step": 1342
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352033769154347e-05,
+      "loss": 0.6015,
+      "step": 1343
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342724467271852e-05,
+      "loss": 0.6496,
+      "step": 1344
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3333411902922934e-05,
+      "loss": 0.5375,
+      "step": 1345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324096085196572e-05,
+      "loss": 0.6265,
+      "step": 1346
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3314777023184907e-05,
+      "loss": 0.5933,
+      "step": 1347
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3305454725983251e-05,
+      "loss": 0.5738,
+      "step": 1348
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.329612920269008e-05,
+      "loss": 0.6469,
+      "step": 1349
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.328680046240701e-05,
+      "loss": 0.6037,
+      "step": 1350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3277468514238803e-05,
+      "loss": 0.6741,
+      "step": 1351
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326813336729335e-05,
+      "loss": 0.6231,
+      "step": 1352
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.325879503068166e-05,
+      "loss": 0.5547,
+      "step": 1353
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3249453513517857e-05,
+      "loss": 0.697,
+      "step": 1354
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240108824919176e-05,
+      "loss": 0.5649,
+      "step": 1355
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3230760974005938e-05,
+      "loss": 0.5432,
+      "step": 1356
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221409969901556e-05,
+      "loss": 0.5477,
+      "step": 1357
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3212055821732517e-05,
+      "loss": 0.5724,
+      "step": 1358
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202698538628376e-05,
+      "loss": 0.6639,
+      "step": 1359
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193338129721756e-05,
+      "loss": 0.5912,
+      "step": 1360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183974604148321e-05,
+      "loss": 0.6328,
+      "step": 1361
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174607971046782e-05,
+      "loss": 0.6785,
+      "step": 1362
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165238239558878e-05,
+      "loss": 0.5515,
+      "step": 1363
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.315586541882938e-05,
+      "loss": 0.6783,
+      "step": 1364
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3146489518006065e-05,
+      "loss": 0.5947,
+      "step": 1365
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137110546239722e-05,
+      "loss": 0.5517,
+      "step": 1366
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312772851268414e-05,
+      "loss": 0.6142,
+      "step": 1367
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3118343426496085e-05,
+      "loss": 0.5913,
+      "step": 1368
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3108955296835313e-05,
+      "loss": 0.6309,
+      "step": 1369
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.309956413286455e-05,
+      "loss": 0.5368,
+      "step": 1370
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.5637,
+      "step": 1371
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3080772738658728e-05,
+      "loss": 0.5687,
+      "step": 1372
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3071372526763888e-05,
+      "loss": 0.552,
+      "step": 1373
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306196931723947e-05,
+      "loss": 0.6393,
+      "step": 1374
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3052563119262915e-05,
+      "loss": 0.5877,
+      "step": 1375
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043153942014579e-05,
+      "loss": 0.5941,
+      "step": 1376
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033741794677728e-05,
+      "loss": 0.5937,
+      "step": 1377
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3024326686438529e-05,
+      "loss": 0.6271,
+      "step": 1378
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3014908626486032e-05,
+      "loss": 0.6009,
+      "step": 1379
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.300548762401217e-05,
+      "loss": 0.6451,
+      "step": 1380
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996063688211754e-05,
+      "loss": 0.6217,
+      "step": 1381
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2986636828282449e-05,
+      "loss": 0.6421,
+      "step": 1382
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2977207053424781e-05,
+      "loss": 0.5799,
+      "step": 1383
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967774372842116e-05,
+      "loss": 0.6219,
+      "step": 1384
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2958338795740662e-05,
+      "loss": 0.6024,
+      "step": 1385
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2948900331329444e-05,
+      "loss": 0.6245,
+      "step": 1386
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2939458988820317e-05,
+      "loss": 0.5077,
+      "step": 1387
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930014777427937e-05,
+      "loss": 0.5738,
+      "step": 1388
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292056770636976e-05,
+      "loss": 0.6006,
+      "step": 1389
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2911117784866037e-05,
+      "loss": 0.6615,
+      "step": 1390
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2901665022139796e-05,
+      "loss": 0.6028,
+      "step": 1391
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2892209427416847e-05,
+      "loss": 0.6881,
+      "step": 1392
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882751009925753e-05,
+      "loss": 0.5644,
+      "step": 1393
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.287328977889784e-05,
+      "loss": 0.6308,
+      "step": 1394
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2863825743567174e-05,
+      "loss": 0.6061,
+      "step": 1395
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854358913170564e-05,
+      "loss": 0.6755,
+      "step": 1396
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844889296947543e-05,
+      "loss": 0.5914,
+      "step": 1397
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2835416904140366e-05,
+      "loss": 0.6245,
+      "step": 1398
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282594174399399e-05,
+      "loss": 0.6263,
+      "step": 1399
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816463825756083e-05,
+      "loss": 0.5296,
+      "step": 1400
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806983158677e-05,
+      "loss": 0.5511,
+      "step": 1401
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.279749975200978e-05,
+      "loss": 0.5952,
+      "step": 1402
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2788013615010136e-05,
+      "loss": 0.663,
+      "step": 1403
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2778524756936438e-05,
+      "loss": 0.6453,
+      "step": 1404
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2769033187049731e-05,
+      "loss": 0.584,
+      "step": 1405
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759538914613682e-05,
+      "loss": 0.5912,
+      "step": 1406
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2750041948894621e-05,
+      "loss": 0.5962,
+      "step": 1407
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2740542299161482e-05,
+      "loss": 0.6023,
+      "step": 1408
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731039974685833e-05,
+      "loss": 0.7238,
+      "step": 1409
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2721534984741856e-05,
+      "loss": 0.5817,
+      "step": 1410
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712027338606323e-05,
+      "loss": 0.5689,
+      "step": 1411
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2702517045558604e-05,
+      "loss": 0.6716,
+      "step": 1412
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2693004114880654e-05,
+      "loss": 0.5923,
+      "step": 1413
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2683488555857e-05,
+      "loss": 0.5796,
+      "step": 1414
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673970377774733e-05,
+      "loss": 0.598,
+      "step": 1415
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2664449589923504e-05,
+      "loss": 0.5691,
+      "step": 1416
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.265492620159551e-05,
+      "loss": 0.5769,
+      "step": 1417
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645400222085477e-05,
+      "loss": 0.5731,
+      "step": 1418
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635871660690677e-05,
+      "loss": 0.5768,
+      "step": 1419
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.262634052671089e-05,
+      "loss": 0.6274,
+      "step": 1420
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2616806829448408e-05,
+      "loss": 0.5599,
+      "step": 1421
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607270578208029e-05,
+      "loss": 0.6288,
+      "step": 1422
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597731782297036e-05,
+      "loss": 0.6078,
+      "step": 1423
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2588190451025209e-05,
+      "loss": 0.6538,
+      "step": 1424
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2578646593704786e-05,
+      "loss": 0.5506,
+      "step": 1425
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2569100219650485e-05,
+      "loss": 0.6626,
+      "step": 1426
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2559551338179468e-05,
+      "loss": 0.553,
+      "step": 1427
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2549999958611351e-05,
+      "loss": 0.6475,
+      "step": 1428
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540446090268193e-05,
+      "loss": 0.5817,
+      "step": 1429
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253088974247447e-05,
+      "loss": 0.5727,
+      "step": 1430
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2521330924557087e-05,
+      "loss": 0.5598,
+      "step": 1431
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2511769645845353e-05,
+      "loss": 0.5748,
+      "step": 1432
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2502205915670985e-05,
+      "loss": 0.5932,
+      "step": 1433
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492639743368096e-05,
+      "loss": 0.5939,
+      "step": 1434
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483071138273168e-05,
+      "loss": 0.5932,
+      "step": 1435
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247350010972507e-05,
+      "loss": 0.6073,
+      "step": 1436
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2463926667065031e-05,
+      "loss": 0.6456,
+      "step": 1437
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2454350819636641e-05,
+      "loss": 0.6855,
+      "step": 1438
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444772576785828e-05,
+      "loss": 0.5521,
+      "step": 1439
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2435191947860867e-05,
+      "loss": 0.5471,
+      "step": 1440
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2425608942212358e-05,
+      "loss": 0.6239,
+      "step": 1441
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241602356919322e-05,
+      "loss": 0.6355,
+      "step": 1442
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406435838158686e-05,
+      "loss": 0.606,
+      "step": 1443
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2396845758466282e-05,
+      "loss": 0.5457,
+      "step": 1444
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2387253339475837e-05,
+      "loss": 0.6137,
+      "step": 1445
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237765859054946e-05,
+      "loss": 0.5987,
+      "step": 1446
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368061521051526e-05,
+      "loss": 0.6,
+      "step": 1447
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2358462140348685e-05,
+      "loss": 0.6611,
+      "step": 1448
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.234886045780984e-05,
+      "loss": 0.6011,
+      "step": 1449
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2339256482806132e-05,
+      "loss": 0.5394,
+      "step": 1450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2329650224710956e-05,
+      "loss": 0.6134,
+      "step": 1451
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2320041692899922e-05,
+      "loss": 0.596,
+      "step": 1452
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310430896750864e-05,
+      "loss": 0.6007,
+      "step": 1453
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2300817845643822e-05,
+      "loss": 0.6017,
+      "step": 1454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291202548961042e-05,
+      "loss": 0.6107,
+      "step": 1455
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281585016086965e-05,
+      "loss": 0.6556,
+      "step": 1456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271965256408201e-05,
+      "loss": 0.572,
+      "step": 1457
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2262343279313545e-05,
+      "loss": 0.5692,
+      "step": 1458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225271909419395e-05,
+      "loss": 0.6288,
+      "step": 1459
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2243092710442533e-05,
+      "loss": 0.5168,
+      "step": 1460
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2233464137454542e-05,
+      "loss": 0.604,
+      "step": 1461
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.222383338462738e-05,
+      "loss": 0.5979,
+      "step": 1462
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221420046136056e-05,
+      "loss": 0.6709,
+      "step": 1463
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220456537705572e-05,
+      "loss": 0.5597,
+      "step": 1464
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2194928141116613e-05,
+      "loss": 0.5789,
+      "step": 1465
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185288762949088e-05,
+      "loss": 0.5953,
+      "step": 1466
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.217564725196108e-05,
+      "loss": 0.6131,
+      "step": 1467
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166003617562608e-05,
+      "loss": 0.5408,
+      "step": 1468
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156357869165771e-05,
+      "loss": 0.6615,
+      "step": 1469
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146710016184722e-05,
+      "loss": 0.6221,
+      "step": 1470
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2137060068035672e-05,
+      "loss": 0.6196,
+      "step": 1471
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127408034136873e-05,
+      "loss": 0.5943,
+      "step": 1472
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117753923908617e-05,
+      "loss": 0.6228,
+      "step": 1473
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2108097746773227e-05,
+      "loss": 0.5796,
+      "step": 1474
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2098439512155028e-05,
+      "loss": 0.5911,
+      "step": 1475
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088779229480371e-05,
+      "loss": 0.5754,
+      "step": 1476
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2079116908177592e-05,
+      "loss": 0.5564,
+      "step": 1477
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069452557677028e-05,
+      "loss": 0.6432,
+      "step": 1478
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2059786187410984e-05,
+      "loss": 0.5779,
+      "step": 1479
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2050117806813748e-05,
+      "loss": 0.6572,
+      "step": 1480
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2040447425321561e-05,
+      "loss": 0.4951,
+      "step": 1481
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2030775052372623e-05,
+      "loss": 0.5692,
+      "step": 1482
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2021100697407075e-05,
+      "loss": 0.5863,
+      "step": 1483
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2011424369866995e-05,
+      "loss": 0.6459,
+      "step": 1484
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2001746079196381e-05,
+      "loss": 0.5356,
+      "step": 1485
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199206583484115e-05,
+      "loss": 0.5727,
+      "step": 1486
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.198238364624913e-05,
+      "loss": 0.6243,
+      "step": 1487
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1972699522870039e-05,
+      "loss": 0.5789,
+      "step": 1488
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1963013474155487e-05,
+      "loss": 0.5778,
+      "step": 1489
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1953325509558964e-05,
+      "loss": 0.5912,
+      "step": 1490
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1943635638535827e-05,
+      "loss": 0.6278,
+      "step": 1491
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1933943870543299e-05,
+      "loss": 0.6139,
+      "step": 1492
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1924250215040447e-05,
+      "loss": 0.521,
+      "step": 1493
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1914554681488188e-05,
+      "loss": 0.5664,
+      "step": 1494
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1904857279349265e-05,
+      "loss": 0.5358,
+      "step": 1495
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1895158018088244e-05,
+      "loss": 0.63,
+      "step": 1496
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1885456907171517e-05,
+      "loss": 0.6307,
+      "step": 1497
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1875753956067272e-05,
+      "loss": 0.6523,
+      "step": 1498
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186604917424549e-05,
+      "loss": 0.5903,
+      "step": 1499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856342571177945e-05,
+      "loss": 0.6258,
+      "step": 1500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1846634156338193e-05,
+      "loss": 0.5943,
+      "step": 1501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836923939201543e-05,
+      "loss": 0.6171,
+      "step": 1502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1827211929245075e-05,
+      "loss": 0.6311,
+      "step": 1503
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817498135947619e-05,
+      "loss": 0.6245,
+      "step": 1504
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807782568789736e-05,
+      "loss": 0.5719,
+      "step": 1505
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1798065237253731e-05,
+      "loss": 0.6327,
+      "step": 1506
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1788346150823625e-05,
+      "loss": 0.5859,
+      "step": 1507
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1778625318985143e-05,
+      "loss": 0.6101,
+      "step": 1508
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176890275122573e-05,
+      "loss": 0.5885,
+      "step": 1509
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1759178457034515e-05,
+      "loss": 0.5301,
+      "step": 1510
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1749452445902315e-05,
+      "loss": 0.6001,
+      "step": 1511
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739724727321618e-05,
+      "loss": 0.532,
+      "step": 1512
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1729995310786586e-05,
+      "loss": 0.6136,
+      "step": 1513
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1720264205793033e-05,
+      "loss": 0.5741,
+      "step": 1514
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710531421838422e-05,
+      "loss": 0.6428,
+      "step": 1515
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700796968421856e-05,
+      "loss": 0.5817,
+      "step": 1516
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1691060855044067e-05,
+      "loss": 0.578,
+      "step": 1517
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681323091207406e-05,
+      "loss": 0.6365,
+      "step": 1518
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671583686415833e-05,
+      "loss": 0.5746,
+      "step": 1519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1661842650174917e-05,
+      "loss": 0.6516,
+      "step": 1520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.165209999199181e-05,
+      "loss": 0.5535,
+      "step": 1521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1642355721375253e-05,
+      "loss": 0.5627,
+      "step": 1522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632609847835556e-05,
+      "loss": 0.5576,
+      "step": 1523
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.16228623808846e-05,
+      "loss": 0.5306,
+      "step": 1524
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613113330035816e-05,
+      "loss": 0.5695,
+      "step": 1525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603362704804185e-05,
+      "loss": 0.5968,
+      "step": 1526
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593610514706217e-05,
+      "loss": 0.6042,
+      "step": 1527
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1583856769259956e-05,
+      "loss": 0.6138,
+      "step": 1528
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574101477984966e-05,
+      "loss": 0.5502,
+      "step": 1529
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.156434465040231e-05,
+      "loss": 0.5912,
+      "step": 1530
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155458629603456e-05,
+      "loss": 0.5666,
+      "step": 1531
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544826424405775e-05,
+      "loss": 0.595,
+      "step": 1532
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535065045041492e-05,
+      "loss": 0.5701,
+      "step": 1533
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1525302167468726e-05,
+      "loss": 0.6294,
+      "step": 1534
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515537801215944e-05,
+      "loss": 0.5393,
+      "step": 1535
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1505771955813079e-05,
+      "loss": 0.5456,
+      "step": 1536
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496004640791491e-05,
+      "loss": 0.5991,
+      "step": 1537
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1486235865683994e-05,
+      "loss": 0.5773,
+      "step": 1538
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476465640024814e-05,
+      "loss": 0.6075,
+      "step": 1539
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1466693973349594e-05,
+      "loss": 0.634,
+      "step": 1540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456920875195383e-05,
+      "loss": 0.569,
+      "step": 1541
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1447146355100638e-05,
+      "loss": 0.5047,
+      "step": 1542
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1437370422605184e-05,
+      "loss": 0.624,
+      "step": 1543
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1427593087250246e-05,
+      "loss": 0.5817,
+      "step": 1544
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.14178143585784e-05,
+      "loss": 0.6439,
+      "step": 1545
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1408034246133593e-05,
+      "loss": 0.6436,
+      "step": 1546
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1398252759461119e-05,
+      "loss": 0.5701,
+      "step": 1547
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1388469908107616e-05,
+      "loss": 0.6006,
+      "step": 1548
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378685701621047e-05,
+      "loss": 0.5351,
+      "step": 1549
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13689001495507e-05,
+      "loss": 0.5432,
+      "step": 1550
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359113261447183e-05,
+      "loss": 0.6216,
+      "step": 1551
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349325046862403e-05,
+      "loss": 0.5671,
+      "step": 1552
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1339535515349563e-05,
+      "loss": 0.6448,
+      "step": 1553
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1329744676463144e-05,
+      "loss": 0.5859,
+      "step": 1554
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1319952539758912e-05,
+      "loss": 0.5919,
+      "step": 1555
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.13101591147939e-05,
+      "loss": 0.6087,
+      "step": 1556
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1300364411126395e-05,
+      "loss": 0.6267,
+      "step": 1557
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1290568438315926e-05,
+      "loss": 0.5999,
+      "step": 1558
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280771205923269e-05,
+      "loss": 0.5703,
+      "step": 1559
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1270972723510432e-05,
+      "loss": 0.5922,
+      "step": 1560
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261173000640636e-05,
+      "loss": 0.5805,
+      "step": 1561
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125137204687831e-05,
+      "loss": 0.4981,
+      "step": 1562
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1241569871789096e-05,
+      "loss": 0.5583,
+      "step": 1563
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1231766484939812e-05,
+      "loss": 0.6526,
+      "step": 1564
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221961895898478e-05,
+      "loss": 0.6267,
+      "step": 1565
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212156114234272e-05,
+      "loss": 0.5818,
+      "step": 1566
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1202349149517541e-05,
+      "loss": 0.632,
+      "step": 1567
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1192541011319788e-05,
+      "loss": 0.5454,
+      "step": 1568
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1182731709213658e-05,
+      "loss": 0.5529,
+      "step": 1569
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1172921252772938e-05,
+      "loss": 0.5355,
+      "step": 1570
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1163109651572535e-05,
+      "loss": 0.5999,
+      "step": 1571
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1153296915188474e-05,
+      "loss": 0.6227,
+      "step": 1572
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1143483053197892e-05,
+      "loss": 0.5877,
+      "step": 1573
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1133668075179023e-05,
+      "loss": 0.6072,
+      "step": 1574
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.112385199071119e-05,
+      "loss": 0.5423,
+      "step": 1575
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1114034809374794e-05,
+      "loss": 0.6496,
+      "step": 1576
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1104216540751309e-05,
+      "loss": 0.6024,
+      "step": 1577
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094397194423266e-05,
+      "loss": 0.5266,
+      "step": 1578
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084576779974257e-05,
+      "loss": 0.5659,
+      "step": 1579
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1074755306988904e-05,
+      "loss": 0.6368,
+      "step": 1580
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064932785052877e-05,
+      "loss": 0.5854,
+      "step": 1581
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1055109223752852e-05,
+      "loss": 0.5808,
+      "step": 1582
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1045284632676535e-05,
+      "loss": 0.5536,
+      "step": 1583
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1035459021412631e-05,
+      "loss": 0.5646,
+      "step": 1584
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102563239955084e-05,
+      "loss": 0.6488,
+      "step": 1585
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015804776681845e-05,
+      "loss": 0.5201,
+      "step": 1586
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1005976162397309e-05,
+      "loss": 0.5942,
+      "step": 1587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996146566289871e-05,
+      "loss": 0.5853,
+      "step": 1588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0986315997953118e-05,
+      "loss": 0.4848,
+      "step": 1589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.097648446698159e-05,
+      "loss": 0.6109,
+      "step": 1590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0966651982970757e-05,
+      "loss": 0.5435,
+      "step": 1591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0956818555517035e-05,
+      "loss": 0.6706,
+      "step": 1592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0946984194217753e-05,
+      "loss": 0.5974,
+      "step": 1593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937148908671149e-05,
+      "loss": 0.5944,
+      "step": 1594
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927312708476367e-05,
+      "loss": 0.5891,
+      "step": 1595
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0917475603233441e-05,
+      "loss": 0.5333,
+      "step": 1596
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090763760254329e-05,
+      "loss": 0.6177,
+      "step": 1597
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897798716007708e-05,
+      "loss": 0.5744,
+      "step": 1598
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0887958953229349e-05,
+      "loss": 0.533,
+      "step": 1599
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087811832381173e-05,
+      "loss": 0.5589,
+      "step": 1600
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868276837359204e-05,
+      "loss": 0.6214,
+      "step": 1601
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0858434503476969e-05,
+      "loss": 0.5452,
+      "step": 1602
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0848591331771045e-05,
+      "loss": 0.5819,
+      "step": 1603
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0838747331848275e-05,
+      "loss": 0.5787,
+      "step": 1604
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828902513316299e-05,
+      "loss": 0.6067,
+      "step": 1605
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0819056885783572e-05,
+      "loss": 0.5616,
+      "step": 1606
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809210458859327e-05,
+      "loss": 0.578,
+      "step": 1607
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0799363242153583e-05,
+      "loss": 0.5371,
+      "step": 1608
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.078951524527712e-05,
+      "loss": 0.5529,
+      "step": 1609
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0779666477841492e-05,
+      "loss": 0.6208,
+      "step": 1610
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0769816949459002e-05,
+      "loss": 0.5724,
+      "step": 1611
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759966669742689e-05,
+      "loss": 0.5893,
+      "step": 1612
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0750115648306333e-05,
+      "loss": 0.6314,
+      "step": 1613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074026389476443e-05,
+      "loss": 0.6007,
+      "step": 1614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730411418732198e-05,
+      "loss": 0.5557,
+      "step": 1615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0720558229825562e-05,
+      "loss": 0.6213,
+      "step": 1616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0710704337661131e-05,
+      "loss": 0.5196,
+      "step": 1617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700849751856212e-05,
+      "loss": 0.5565,
+      "step": 1618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069099448202878e-05,
+      "loss": 0.5241,
+      "step": 1619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681138537797487e-05,
+      "loss": 0.5775,
+      "step": 1620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671281928781636e-05,
+      "loss": 0.6101,
+      "step": 1621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0661424664601181e-05,
+      "loss": 0.5981,
+      "step": 1622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651566754876715e-05,
+      "loss": 0.5494,
+      "step": 1623
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0641708209229463e-05,
+      "loss": 0.5429,
+      "step": 1624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631849037281267e-05,
+      "loss": 0.6059,
+      "step": 1625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0621989248654585e-05,
+      "loss": 0.6474,
+      "step": 1626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612128852972474e-05,
+      "loss": 0.5786,
+      "step": 1627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602267859858581e-05,
+      "loss": 0.6027,
+      "step": 1628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0592406278937143e-05,
+      "loss": 0.5475,
+      "step": 1629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0582544119832967e-05,
+      "loss": 0.6362,
+      "step": 1630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572681392171417e-05,
+      "loss": 0.5317,
+      "step": 1631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562818105578426e-05,
+      "loss": 0.5908,
+      "step": 1632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0552954269680462e-05,
+      "loss": 0.6311,
+      "step": 1633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543089894104534e-05,
+      "loss": 0.6351,
+      "step": 1634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0533224988478176e-05,
+      "loss": 0.6174,
+      "step": 1635
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523359562429441e-05,
+      "loss": 0.5969,
+      "step": 1636
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051349362558688e-05,
+      "loss": 0.7705,
+      "step": 1637
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503627187579564e-05,
+      "loss": 0.606,
+      "step": 1638
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049376025803703e-05,
+      "loss": 0.6259,
+      "step": 1639
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0483892846589312e-05,
+      "loss": 0.6078,
+      "step": 1640
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.04740249628669e-05,
+      "loss": 0.5534,
+      "step": 1641
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0464156616500756e-05,
+      "loss": 0.6488,
+      "step": 1642
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0454287817122291e-05,
+      "loss": 0.6022,
+      "step": 1643
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444418574363354e-05,
+      "loss": 0.5911,
+      "step": 1644
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.043454889785623e-05,
+      "loss": 0.494,
+      "step": 1645
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0424678797233626e-05,
+      "loss": 0.6272,
+      "step": 1646
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0414808282128668e-05,
+      "loss": 0.581,
+      "step": 1647
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404937362174878e-05,
+      "loss": 0.5812,
+      "step": 1648
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039506604700618e-05,
+      "loss": 0.6745,
+      "step": 1649
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038519434625688e-05,
+      "loss": 0.5888,
+      "step": 1650
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375322269561658e-05,
+      "loss": 0.5842,
+      "step": 1651
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365449826555571e-05,
+      "loss": 0.5661,
+      "step": 1652
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0355577026874026e-05,
+      "loss": 0.5571,
+      "step": 1653
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345703880152776e-05,
+      "loss": 0.5797,
+      "step": 1654
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335830396027912e-05,
+      "loss": 0.5597,
+      "step": 1655
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0325956584135866e-05,
+      "loss": 0.5496,
+      "step": 1656
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0316082454113379e-05,
+      "loss": 0.5503,
+      "step": 1657
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0306208015597505e-05,
+      "loss": 0.6193,
+      "step": 1658
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0296333278225599e-05,
+      "loss": 0.6,
+      "step": 1659
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0286458251635307e-05,
+      "loss": 0.5619,
+      "step": 1660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276582945464564e-05,
+      "loss": 0.5833,
+      "step": 1661
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0266707369351565e-05,
+      "loss": 0.6088,
+      "step": 1662
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0256831532934783e-05,
+      "loss": 0.5404,
+      "step": 1663
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246955445852928e-05,
+      "loss": 0.5983,
+      "step": 1664
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0237079117744977e-05,
+      "loss": 0.583,
+      "step": 1665
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0227202558250121e-05,
+      "loss": 0.6075,
+      "step": 1666
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.021732577700779e-05,
+      "loss": 0.5635,
+      "step": 1667
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207448783657621e-05,
+      "loss": 0.5295,
+      "step": 1668
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0197571587839466e-05,
+      "loss": 0.5815,
+      "step": 1669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0187694199193371e-05,
+      "loss": 0.5778,
+      "step": 1670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0177816627359575e-05,
+      "loss": 0.5662,
+      "step": 1671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167938881978488e-05,
+      "loss": 0.5523,
+      "step": 1672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015806097269069e-05,
+      "loss": 0.657,
+      "step": 1673
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0148182909136928e-05,
+      "loss": 0.5574,
+      "step": 1674
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0138304700958096e-05,
+      "loss": 0.6156,
+      "step": 1675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0128426357795225e-05,
+      "loss": 0.6251,
+      "step": 1676
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118547889289485e-05,
+      "loss": 0.6271,
+      "step": 1677
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0108669305082159e-05,
+      "loss": 0.5749,
+      "step": 1678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0098790614814658e-05,
+      "loss": 0.5677,
+      "step": 1679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008891182812848e-05,
+      "loss": 0.5616,
+      "step": 1680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007903295466523e-05,
+      "loss": 0.6119,
+      "step": 1681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0069154004066586e-05,
+      "loss": 0.6798,
+      "step": 1682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059274985974305e-05,
+      "loss": 0.5169,
+      "step": 1683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0049395910030223e-05,
+      "loss": 0.6043,
+      "step": 1684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039516785876213e-05,
+      "loss": 0.5601,
+      "step": 1685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0029637623154209e-05,
+      "loss": 0.5946,
+      "step": 1686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001975843150617e-05,
+      "loss": 0.5919,
+      "step": 1687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009879220574096e-05,
+      "loss": 0.5992,
+      "step": 1688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.6458,
+      "step": 1689
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990120779425904e-06,
+      "loss": 0.5533,
+      "step": 1690
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980241568493834e-06,
+      "loss": 0.5951,
+      "step": 1691
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.970362376845795e-06,
+      "loss": 0.5533,
+      "step": 1692
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960483214123789e-06,
+      "loss": 0.5962,
+      "step": 1693
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950604089969778e-06,
+      "loss": 0.6048,
+      "step": 1694
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940725014025696e-06,
+      "loss": 0.6218,
+      "step": 1695
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.930845995933418e-06,
+      "loss": 0.5736,
+      "step": 1696
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920967045334774e-06,
+      "loss": 0.5905,
+      "step": 1697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91108817187152e-06,
+      "loss": 0.6104,
+      "step": 1698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.901209385185345e-06,
+      "loss": 0.5896,
+      "step": 1699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.891330694917841e-06,
+      "loss": 0.5728,
+      "step": 1700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881452110710519e-06,
+      "loss": 0.6391,
+      "step": 1701
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.871573642204777e-06,
+      "loss": 0.6238,
+      "step": 1702
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.86169529904191e-06,
+      "loss": 0.5431,
+      "step": 1703
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851817090863073e-06,
+      "loss": 0.6126,
+      "step": 1704
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841939027309313e-06,
+      "loss": 0.6055,
+      "step": 1705
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832061118021514e-06,
+      "loss": 0.4992,
+      "step": 1706
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822183372640426e-06,
+      "loss": 0.5472,
+      "step": 1707
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.812305800806627e-06,
+      "loss": 0.5625,
+      "step": 1708
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.802428412160538e-06,
+      "loss": 0.535,
+      "step": 1709
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792551216342382e-06,
+      "loss": 0.6961,
+      "step": 1710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.782674222992214e-06,
+      "loss": 0.5962,
+      "step": 1711
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.77279744174988e-06,
+      "loss": 0.5359,
+      "step": 1712
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.762920882255026e-06,
+      "loss": 0.685,
+      "step": 1713
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75304455414707e-06,
+      "loss": 0.5011,
+      "step": 1714
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74316846706522e-06,
+      "loss": 0.5923,
+      "step": 1715
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.733292630648434e-06,
+      "loss": 0.5988,
+      "step": 1716
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72341705453544e-06,
+      "loss": 0.6055,
+      "step": 1717
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.713541748364693e-06,
+      "loss": 0.605,
+      "step": 1718
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.703666721774403e-06,
+      "loss": 0.636,
+      "step": 1719
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.693791984402495e-06,
+      "loss": 0.5531,
+      "step": 1720
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.683917545886623e-06,
+      "loss": 0.5116,
+      "step": 1721
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.674043415864134e-06,
+      "loss": 0.5947,
+      "step": 1722
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664169603972091e-06,
+      "loss": 0.6456,
+      "step": 1723
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654296119847228e-06,
+      "loss": 0.5832,
+      "step": 1724
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.644422973125977e-06,
+      "loss": 0.5739,
+      "step": 1725
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634550173444428e-06,
+      "loss": 0.5794,
+      "step": 1726
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624677730438344e-06,
+      "loss": 0.5967,
+      "step": 1727
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614805653743124e-06,
+      "loss": 0.5566,
+      "step": 1728
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.604933952993822e-06,
+      "loss": 0.5883,
+      "step": 1729
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.595062637825122e-06,
+      "loss": 0.5759,
+      "step": 1730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.585191717871336e-06,
+      "loss": 0.554,
+      "step": 1731
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.575321202766374e-06,
+      "loss": 0.5635,
+      "step": 1732
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.565451102143772e-06,
+      "loss": 0.6766,
+      "step": 1733
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555581425636648e-06,
+      "loss": 0.5814,
+      "step": 1734
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.545712182877714e-06,
+      "loss": 0.5703,
+      "step": 1735
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.535843383499244e-06,
+      "loss": 0.5666,
+      "step": 1736
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525975037133101e-06,
+      "loss": 0.6245,
+      "step": 1737
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51610715341069e-06,
+      "loss": 0.6209,
+      "step": 1738
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.506239741962971e-06,
+      "loss": 0.557,
+      "step": 1739
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.496372812420436e-06,
+      "loss": 0.6475,
+      "step": 1740
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.486506374413121e-06,
+      "loss": 0.6383,
+      "step": 1741
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.476640437570562e-06,
+      "loss": 0.6019,
+      "step": 1742
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.466775011521825e-06,
+      "loss": 0.5794,
+      "step": 1743
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.456910105895464e-06,
+      "loss": 0.5985,
+      "step": 1744
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44704573031954e-06,
+      "loss": 0.5807,
+      "step": 1745
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437181894421576e-06,
+      "loss": 0.6433,
+      "step": 1746
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.427318607828584e-06,
+      "loss": 0.6553,
+      "step": 1747
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.417455880167034e-06,
+      "loss": 0.574,
+      "step": 1748
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.407593721062858e-06,
+      "loss": 0.5655,
+      "step": 1749
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.397732140141419e-06,
+      "loss": 0.6027,
+      "step": 1750
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387871147027528e-06,
+      "loss": 0.5673,
+      "step": 1751
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378010751345414e-06,
+      "loss": 0.5856,
+      "step": 1752
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368150962718736e-06,
+      "loss": 0.6206,
+      "step": 1753
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.358291790770539e-06,
+      "loss": 0.569,
+      "step": 1754
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.34843324512329e-06,
+      "loss": 0.5994,
+      "step": 1755
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.338575335398822e-06,
+      "loss": 0.5735,
+      "step": 1756
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328718071218366e-06,
+      "loss": 0.5738,
+      "step": 1757
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.318861462202513e-06,
+      "loss": 0.4931,
+      "step": 1758
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309005517971222e-06,
+      "loss": 0.5264,
+      "step": 1759
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.299150248143791e-06,
+      "loss": 0.5592,
+      "step": 1760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.28929566233887e-06,
+      "loss": 0.5907,
+      "step": 1761
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.279441770174438e-06,
+      "loss": 0.6046,
+      "step": 1762
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269588581267804e-06,
+      "loss": 0.5894,
+      "step": 1763
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259736105235572e-06,
+      "loss": 0.6063,
+      "step": 1764
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.24988435169367e-06,
+      "loss": 0.5219,
+      "step": 1765
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.240033330257313e-06,
+      "loss": 0.5862,
+      "step": 1766
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230183050541001e-06,
+      "loss": 0.5568,
+      "step": 1767
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.22033352215851e-06,
+      "loss": 0.6429,
+      "step": 1768
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.210484754722882e-06,
+      "loss": 0.5988,
+      "step": 1769
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200636757846419e-06,
+      "loss": 0.5602,
+      "step": 1770
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190789541140675e-06,
+      "loss": 0.6432,
+      "step": 1771
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.180943114216428e-06,
+      "loss": 0.6066,
+      "step": 1772
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171097486683703e-06,
+      "loss": 0.5584,
+      "step": 1773
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.161252668151729e-06,
+      "loss": 0.5391,
+      "step": 1774
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.151408668228958e-06,
+      "loss": 0.5548,
+      "step": 1775
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.141565496523033e-06,
+      "loss": 0.5531,
+      "step": 1776
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.1317231626408e-06,
+      "loss": 0.5323,
+      "step": 1777
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.121881676188273e-06,
+      "loss": 0.6243,
+      "step": 1778
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112041046770653e-06,
+      "loss": 0.6057,
+      "step": 1779
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102201283992294e-06,
+      "loss": 0.535,
+      "step": 1780
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.092362397456713e-06,
+      "loss": 0.5822,
+      "step": 1781
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08252439676656e-06,
+      "loss": 0.6456,
+      "step": 1782
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072687291523636e-06,
+      "loss": 0.5772,
+      "step": 1783
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.062851091328851e-06,
+      "loss": 0.5662,
+      "step": 1784
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.053015805782249e-06,
+      "loss": 0.5524,
+      "step": 1785
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043181444482965e-06,
+      "loss": 0.5766,
+      "step": 1786
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.033348017029247e-06,
+      "loss": 0.5715,
+      "step": 1787
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.023515533018415e-06,
+      "loss": 0.5809,
+      "step": 1788
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.013684002046883e-06,
+      "loss": 0.6515,
+      "step": 1789
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003853433710127e-06,
+      "loss": 0.5928,
+      "step": 1790
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.994023837602694e-06,
+      "loss": 0.6113,
+      "step": 1791
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984195223318159e-06,
+      "loss": 0.5632,
+      "step": 1792
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974367600449166e-06,
+      "loss": 0.6112,
+      "step": 1793
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.96454097858737e-06,
+      "loss": 0.6166,
+      "step": 1794
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.954715367323468e-06,
+      "loss": 0.6557,
+      "step": 1795
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.94489077624715e-06,
+      "loss": 0.5804,
+      "step": 1796
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935067214947126e-06,
+      "loss": 0.5358,
+      "step": 1797
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925244693011094e-06,
+      "loss": 0.5921,
+      "step": 1798
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.915423220025747e-06,
+      "loss": 0.629,
+      "step": 1799
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905602805576734e-06,
+      "loss": 0.536,
+      "step": 1800
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.895783459248695e-06,
+      "loss": 0.5468,
+      "step": 1801
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.885965190625207e-06,
+      "loss": 0.5863,
+      "step": 1802
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876148009288813e-06,
+      "loss": 0.515,
+      "step": 1803
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866331924820977e-06,
+      "loss": 0.6563,
+      "step": 1804
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.856516946802111e-06,
+      "loss": 0.5476,
+      "step": 1805
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.846703084811528e-06,
+      "loss": 0.663,
+      "step": 1806
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836890348427468e-06,
+      "loss": 0.5815,
+      "step": 1807
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.827078747227062e-06,
+      "loss": 0.6593,
+      "step": 1808
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817268290786343e-06,
+      "loss": 0.5695,
+      "step": 1809
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.807458988680214e-06,
+      "loss": 0.5777,
+      "step": 1810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79765085048246e-06,
+      "loss": 0.6933,
+      "step": 1811
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78784388576573e-06,
+      "loss": 0.5965,
+      "step": 1812
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778038104101525e-06,
+      "loss": 0.7332,
+      "step": 1813
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.768233515060188e-06,
+      "loss": 0.5689,
+      "step": 1814
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.758430128210908e-06,
+      "loss": 0.5954,
+      "step": 1815
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.748627953121691e-06,
+      "loss": 0.6103,
+      "step": 1816
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.73882699935937e-06,
+      "loss": 0.6175,
+      "step": 1817
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.729027276489568e-06,
+      "loss": 0.5823,
+      "step": 1818
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719228794076733e-06,
+      "loss": 0.5655,
+      "step": 1819
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.709431561684078e-06,
+      "loss": 0.6327,
+      "step": 1820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.69963558887361e-06,
+      "loss": 0.642,
+      "step": 1821
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6898408852061e-06,
+      "loss": 0.6411,
+      "step": 1822
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.68004746024109e-06,
+      "loss": 0.6077,
+      "step": 1823
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.670255323536858e-06,
+      "loss": 0.5744,
+      "step": 1824
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.660464484650442e-06,
+      "loss": 0.6262,
+      "step": 1825
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650674953137597e-06,
+      "loss": 0.6411,
+      "step": 1826
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.64088673855282e-06,
+      "loss": 0.5123,
+      "step": 1827
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.631099850449302e-06,
+      "loss": 0.6584,
+      "step": 1828
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621314298378958e-06,
+      "loss": 0.5809,
+      "step": 1829
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.611530091892386e-06,
+      "loss": 0.5896,
+      "step": 1830
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.601747240538883e-06,
+      "loss": 0.571,
+      "step": 1831
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.591965753866407e-06,
+      "loss": 0.628,
+      "step": 1832
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582185641421601e-06,
+      "loss": 0.5979,
+      "step": 1833
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.572406912749754e-06,
+      "loss": 0.6639,
+      "step": 1834
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.562629577394817e-06,
+      "loss": 0.5658,
+      "step": 1835
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.552853644899363e-06,
+      "loss": 0.5509,
+      "step": 1836
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.543079124804618e-06,
+      "loss": 0.5773,
+      "step": 1837
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.53330602665041e-06,
+      "loss": 0.5447,
+      "step": 1838
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52353435997519e-06,
+      "loss": 0.6189,
+      "step": 1839
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.513764134316006e-06,
+      "loss": 0.6687,
+      "step": 1840
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50399535920851e-06,
+      "loss": 0.5598,
+      "step": 1841
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.494228044186925e-06,
+      "loss": 0.5481,
+      "step": 1842
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484462198784058e-06,
+      "loss": 0.5632,
+      "step": 1843
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.474697832531276e-06,
+      "loss": 0.6044,
+      "step": 1844
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.46493495495851e-06,
+      "loss": 0.6139,
+      "step": 1845
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455173575594227e-06,
+      "loss": 0.5683,
+      "step": 1846
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.445413703965441e-06,
+      "loss": 0.5946,
+      "step": 1847
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43565534959769e-06,
+      "loss": 0.5632,
+      "step": 1848
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.425898522015038e-06,
+      "loss": 0.6584,
+      "step": 1849
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.416143230740045e-06,
+      "loss": 0.6658,
+      "step": 1850
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406389485293786e-06,
+      "loss": 0.5223,
+      "step": 1851
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396637295195817e-06,
+      "loss": 0.5387,
+      "step": 1852
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.386886669964185e-06,
+      "loss": 0.6141,
+      "step": 1853
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377137619115401e-06,
+      "loss": 0.5968,
+      "step": 1854
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.367390152164448e-06,
+      "loss": 0.6049,
+      "step": 1855
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35764427862475e-06,
+      "loss": 0.5926,
+      "step": 1856
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347900008008194e-06,
+      "loss": 0.5882,
+      "step": 1857
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.338157349825083e-06,
+      "loss": 0.6159,
+      "step": 1858
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328416313584169e-06,
+      "loss": 0.5192,
+      "step": 1859
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318676908792596e-06,
+      "loss": 0.5139,
+      "step": 1860
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.308939144955934e-06,
+      "loss": 0.5612,
+      "step": 1861
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299203031578144e-06,
+      "loss": 0.5551,
+      "step": 1862
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289468578161581e-06,
+      "loss": 0.5578,
+      "step": 1863
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.279735794206969e-06,
+      "loss": 0.5449,
+      "step": 1864
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.270004689213416e-06,
+      "loss": 0.6624,
+      "step": 1865
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260275272678382e-06,
+      "loss": 0.5738,
+      "step": 1866
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25054755409769e-06,
+      "loss": 0.6274,
+      "step": 1867
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.240821542965487e-06,
+      "loss": 0.6387,
+      "step": 1868
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231097248774273e-06,
+      "loss": 0.6376,
+      "step": 1869
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221374681014858e-06,
+      "loss": 0.6239,
+      "step": 1870
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21165384917638e-06,
+      "loss": 0.5985,
+      "step": 1871
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.201934762746269e-06,
+      "loss": 0.5441,
+      "step": 1872
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.192217431210266e-06,
+      "loss": 0.6092,
+      "step": 1873
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.182501864052383e-06,
+      "loss": 0.5018,
+      "step": 1874
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.172788070754927e-06,
+      "loss": 0.5875,
+      "step": 1875
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.163076060798459e-06,
+      "loss": 0.6219,
+      "step": 1876
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.153365843661812e-06,
+      "loss": 0.5642,
+      "step": 1877
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143657428822055e-06,
+      "loss": 0.5784,
+      "step": 1878
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.133950825754511e-06,
+      "loss": 0.562,
+      "step": 1879
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12424604393273e-06,
+      "loss": 0.5827,
+      "step": 1880
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.114543092828485e-06,
+      "loss": 0.658,
+      "step": 1881
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.104841981911756e-06,
+      "loss": 0.5838,
+      "step": 1882
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.095142720650739e-06,
+      "loss": 0.5839,
+      "step": 1883
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.085445318511814e-06,
+      "loss": 0.5506,
+      "step": 1884
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.075749784959556e-06,
+      "loss": 0.5353,
+      "step": 1885
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.066056129456703e-06,
+      "loss": 0.586,
+      "step": 1886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.056364361464176e-06,
+      "loss": 0.6318,
+      "step": 1887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.046674490441037e-06,
+      "loss": 0.6284,
+      "step": 1888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.036986525844516e-06,
+      "loss": 0.6072,
+      "step": 1889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.027300477129963e-06,
+      "loss": 0.593,
+      "step": 1890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.017616353750874e-06,
+      "loss": 0.5908,
+      "step": 1891
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.007934165158851e-06,
+      "loss": 0.554,
+      "step": 1892
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99825392080362e-06,
+      "loss": 0.5881,
+      "step": 1893
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.988575630133007e-06,
+      "loss": 0.5043,
+      "step": 1894
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.978899302592927e-06,
+      "loss": 0.5622,
+      "step": 1895
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969224947627379e-06,
+      "loss": 0.5669,
+      "step": 1896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.95955257467844e-06,
+      "loss": 0.567,
+      "step": 1897
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.949882193186252e-06,
+      "loss": 0.5361,
+      "step": 1898
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940213812589018e-06,
+      "loss": 0.6025,
+      "step": 1899
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.930547442322973e-06,
+      "loss": 0.5928,
+      "step": 1900
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92088309182241e-06,
+      "loss": 0.5611,
+      "step": 1901
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91122077051963e-06,
+      "loss": 0.6102,
+      "step": 1902
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.901560487844973e-06,
+      "loss": 0.5727,
+      "step": 1903
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.891902253226775e-06,
+      "loss": 0.6066,
+      "step": 1904
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882246076091385e-06,
+      "loss": 0.5935,
+      "step": 1905
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87259196586313e-06,
+      "loss": 0.6098,
+      "step": 1906
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862939931964333e-06,
+      "loss": 0.578,
+      "step": 1907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.85328998381528e-06,
+      "loss": 0.5909,
+      "step": 1908
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843642130834232e-06,
+      "loss": 0.5431,
+      "step": 1909
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833996382437394e-06,
+      "loss": 0.5895,
+      "step": 1910
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.824352748038924e-06,
+      "loss": 0.5914,
+      "step": 1911
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814711237050913e-06,
+      "loss": 0.6193,
+      "step": 1912
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.80507185888339e-06,
+      "loss": 0.5929,
+      "step": 1913
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.795434622944282e-06,
+      "loss": 0.6125,
+      "step": 1914
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785799538639445e-06,
+      "loss": 0.5726,
+      "step": 1915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.776166615372622e-06,
+      "loss": 0.5481,
+      "step": 1916
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76653586254546e-06,
+      "loss": 0.6245,
+      "step": 1917
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.756907289557469e-06,
+      "loss": 0.5653,
+      "step": 1918
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.747280905806051e-06,
+      "loss": 0.6765,
+      "step": 1919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737656720686457e-06,
+      "loss": 0.5412,
+      "step": 1920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.7280347435918e-06,
+      "loss": 0.5438,
+      "step": 1921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718414983913037e-06,
+      "loss": 0.5934,
+      "step": 1922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70879745103896e-06,
+      "loss": 0.524,
+      "step": 1923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69918215435618e-06,
+      "loss": 0.54,
+      "step": 1924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.689569103249139e-06,
+      "loss": 0.5464,
+      "step": 1925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.679958307100078e-06,
+      "loss": 0.5827,
+      "step": 1926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.670349775289047e-06,
+      "loss": 0.6351,
+      "step": 1927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.660743517193868e-06,
+      "loss": 0.6905,
+      "step": 1928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651139542190164e-06,
+      "loss": 0.6463,
+      "step": 1929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.641537859651316e-06,
+      "loss": 0.5906,
+      "step": 1930
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631938478948478e-06,
+      "loss": 0.5524,
+      "step": 1931
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.622341409450542e-06,
+      "loss": 0.5259,
+      "step": 1932
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.612746660524165e-06,
+      "loss": 0.6032,
+      "step": 1933
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.603154241533719e-06,
+      "loss": 0.5523,
+      "step": 1934
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593564161841318e-06,
+      "loss": 0.5998,
+      "step": 1935
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.583976430806781e-06,
+      "loss": 0.6784,
+      "step": 1936
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.574391057787646e-06,
+      "loss": 0.621,
+      "step": 1937
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5648080521391344e-06,
+      "loss": 0.5681,
+      "step": 1938
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555227423214174e-06,
+      "loss": 0.6107,
+      "step": 1939
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.545649180363361e-06,
+      "loss": 0.5364,
+      "step": 1940
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.536073332934972e-06,
+      "loss": 0.6933,
+      "step": 1941
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.526499890274932e-06,
+      "loss": 0.5687,
+      "step": 1942
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516928861726834e-06,
+      "loss": 0.6181,
+      "step": 1943
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5073602566319045e-06,
+      "loss": 0.5733,
+      "step": 1944
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.497794084329015e-06,
+      "loss": 0.6089,
+      "step": 1945
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.488230354154648e-06,
+      "loss": 0.6065,
+      "step": 1946
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.478669075442917e-06,
+      "loss": 0.5724,
+      "step": 1947
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469110257525531e-06,
+      "loss": 0.5553,
+      "step": 1948
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.45955390973181e-06,
+      "loss": 0.5572,
+      "step": 1949
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.450000041388649e-06,
+      "loss": 0.57,
+      "step": 1950
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.440448661820536e-06,
+      "loss": 0.5886,
+      "step": 1951
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.430899780349519e-06,
+      "loss": 0.6486,
+      "step": 1952
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.421353406295215e-06,
+      "loss": 0.5447,
+      "step": 1953
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.411809548974792e-06,
+      "loss": 0.6042,
+      "step": 1954
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.402268217702966e-06,
+      "loss": 0.5418,
+      "step": 1955
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3927294217919735e-06,
+      "loss": 0.5908,
+      "step": 1956
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.383193170551595e-06,
+      "loss": 0.6206,
+      "step": 1957
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.373659473289111e-06,
+      "loss": 0.5293,
+      "step": 1958
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364128339309326e-06,
+      "loss": 0.5896,
+      "step": 1959
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354599777914524e-06,
+      "loss": 0.564,
+      "step": 1960
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345073798404494e-06,
+      "loss": 0.6167,
+      "step": 1961
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3355504100764956e-06,
+      "loss": 0.5858,
+      "step": 1962
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.326029622225269e-06,
+      "loss": 0.5862,
+      "step": 1963
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316511444143001e-06,
+      "loss": 0.551,
+      "step": 1964
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306995885119349e-06,
+      "loss": 0.4908,
+      "step": 1965
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.297482954441397e-06,
+      "loss": 0.6605,
+      "step": 1966
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.28797266139368e-06,
+      "loss": 0.5006,
+      "step": 1967
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278465015258145e-06,
+      "loss": 0.5688,
+      "step": 1968
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268960025314169e-06,
+      "loss": 0.497,
+      "step": 1969
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.259457700838521e-06,
+      "loss": 0.6667,
+      "step": 1970
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.249958051105383e-06,
+      "loss": 0.6103,
+      "step": 1971
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240461085386317e-06,
+      "loss": 0.5978,
+      "step": 1972
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.230966812950273e-06,
+      "loss": 0.7119,
+      "step": 1973
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2214752430635625e-06,
+      "loss": 0.6096,
+      "step": 1974
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2119863849898684e-06,
+      "loss": 0.4956,
+      "step": 1975
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.20250024799022e-06,
+      "loss": 0.6648,
+      "step": 1976
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1930168413230025e-06,
+      "loss": 0.5593,
+      "step": 1977
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183536174243919e-06,
+      "loss": 0.5813,
+      "step": 1978
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.174058256006012e-06,
+      "loss": 0.6,
+      "step": 1979
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.164583095859636e-06,
+      "loss": 0.5068,
+      "step": 1980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.155110703052459e-06,
+      "loss": 0.5392,
+      "step": 1981
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145641086829436e-06,
+      "loss": 0.6056,
+      "step": 1982
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.136174256432828e-06,
+      "loss": 0.5741,
+      "step": 1983
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.126710221102162e-06,
+      "loss": 0.6141,
+      "step": 1984
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1172489900742495e-06,
+      "loss": 0.5633,
+      "step": 1985
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107790572583154e-06,
+      "loss": 0.5856,
+      "step": 1986
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0983349778602064e-06,
+      "loss": 0.6066,
+      "step": 1987
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0888822151339654e-06,
+      "loss": 0.5192,
+      "step": 1988
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079432293630244e-06,
+      "loss": 0.5697,
+      "step": 1989
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.069985222572064e-06,
+      "loss": 0.5314,
+      "step": 1990
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0605410111796855e-06,
+      "loss": 0.5158,
+      "step": 1991
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0510996686705555e-06,
+      "loss": 0.5624,
+      "step": 1992
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0416612042593405e-06,
+      "loss": 0.5283,
+      "step": 1993
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.032225627157883e-06,
+      "loss": 0.6033,
+      "step": 1994
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.022792946575222e-06,
+      "loss": 0.6631,
+      "step": 1995
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.013363171717552e-06,
+      "loss": 0.6509,
+      "step": 1996
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003936311788251e-06,
+      "loss": 0.6223,
+      "step": 1997
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.994512375987832e-06,
+      "loss": 0.481,
+      "step": 1998
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.985091373513972e-06,
+      "loss": 0.551,
+      "step": 1999
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.975673313561473e-06,
+      "loss": 0.5677,
+      "step": 2000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966258205322274e-06,
+      "loss": 0.5763,
+      "step": 2001
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956846057985423e-06,
+      "loss": 0.6085,
+      "step": 2002
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.947436880737089e-06,
+      "loss": 0.5829,
+      "step": 2003
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938030682760532e-06,
+      "loss": 0.5649,
+      "step": 2004
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928627473236116e-06,
+      "loss": 0.71,
+      "step": 2005
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.919227261341275e-06,
+      "loss": 0.5745,
+      "step": 2006
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.63,
+      "step": 2007
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900435867135451e-06,
+      "loss": 0.5865,
+      "step": 2008
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8910447031646884e-06,
+      "loss": 0.552,
+      "step": 2009
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8816565735039164e-06,
+      "loss": 0.5444,
+      "step": 2010
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8722714873158635e-06,
+      "loss": 0.5207,
+      "step": 2011
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862889453760278e-06,
+      "loss": 0.486,
+      "step": 2012
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.853510481993939e-06,
+      "loss": 0.598,
+      "step": 2013
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844134581170623e-06,
+      "loss": 0.549,
+      "step": 2014
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834761760441127e-06,
+      "loss": 0.5811,
+      "step": 2015
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8253920289532214e-06,
+      "loss": 0.5927,
+      "step": 2016
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.816025395851682e-06,
+      "loss": 0.5593,
+      "step": 2017
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806661870278244e-06,
+      "loss": 0.568,
+      "step": 2018
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797301461371626e-06,
+      "loss": 0.5343,
+      "step": 2019
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787944178267486e-06,
+      "loss": 0.4862,
+      "step": 2020
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7785900300984475e-06,
+      "loss": 0.6664,
+      "step": 2021
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769239025994063e-06,
+      "loss": 0.5388,
+      "step": 2022
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759891175080827e-06,
+      "loss": 0.6207,
+      "step": 2023
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.750546486482143e-06,
+      "loss": 0.5667,
+      "step": 2024
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741204969318343e-06,
+      "loss": 0.544,
+      "step": 2025
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.731866632706652e-06,
+      "loss": 0.5862,
+      "step": 2026
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.722531485761199e-06,
+      "loss": 0.6168,
+      "step": 2027
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.713199537592991e-06,
+      "loss": 0.4779,
+      "step": 2028
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703870797309922e-06,
+      "loss": 0.5697,
+      "step": 2029
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694545274016748e-06,
+      "loss": 0.6281,
+      "step": 2030
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6852229768150976e-06,
+      "loss": 0.5831,
+      "step": 2031
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.67590391480343e-06,
+      "loss": 0.5689,
+      "step": 2032
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.666588097077068e-06,
+      "loss": 0.6451,
+      "step": 2033
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6572755327281506e-06,
+      "loss": 0.612,
+      "step": 2034
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.647966230845655e-06,
+      "loss": 0.6088,
+      "step": 2035
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.638660200515366e-06,
+      "loss": 0.531,
+      "step": 2036
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.629357450819885e-06,
+      "loss": 0.5208,
+      "step": 2037
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6200579908385955e-06,
+      "loss": 0.5816,
+      "step": 2038
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.610761829647685e-06,
+      "loss": 0.5808,
+      "step": 2039
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.601468976320111e-06,
+      "loss": 0.5791,
+      "step": 2040
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592179439925619e-06,
+      "loss": 0.5526,
+      "step": 2041
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5828932295306916e-06,
+      "loss": 0.5423,
+      "step": 2042
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.573610354198587e-06,
+      "loss": 0.5547,
+      "step": 2043
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5643308229893e-06,
+      "loss": 0.6098,
+      "step": 2044
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.555054644959565e-06,
+      "loss": 0.584,
+      "step": 2045
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545781829162834e-06,
+      "loss": 0.5625,
+      "step": 2046
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.536512384649294e-06,
+      "loss": 0.5244,
+      "step": 2047
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527246320465822e-06,
+      "loss": 0.5357,
+      "step": 2048
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.517983645656014e-06,
+      "loss": 0.5148,
+      "step": 2049
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.508724369260144e-06,
+      "loss": 0.5021,
+      "step": 2050
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499468500315185e-06,
+      "loss": 0.5792,
+      "step": 2051
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4902160478547626e-06,
+      "loss": 0.6126,
+      "step": 2052
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.480967020909188e-06,
+      "loss": 0.5888,
+      "step": 2053
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4717214285054165e-06,
+      "loss": 0.5622,
+      "step": 2054
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4624792796670624e-06,
+      "loss": 0.6189,
+      "step": 2055
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.453240583414364e-06,
+      "loss": 0.608,
+      "step": 2056
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444005348764207e-06,
+      "loss": 0.5663,
+      "step": 2057
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434773584730083e-06,
+      "loss": 0.5346,
+      "step": 2058
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4255453003221115e-06,
+      "loss": 0.5095,
+      "step": 2059
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4163205045469975e-06,
+      "loss": 0.5277,
+      "step": 2060
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4070992064080606e-06,
+      "loss": 0.559,
+      "step": 2061
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397881414905193e-06,
+      "loss": 0.5268,
+      "step": 2062
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.388667139034873e-06,
+      "loss": 0.6109,
+      "step": 2063
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3794563877901385e-06,
+      "loss": 0.6014,
+      "step": 2064
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.370249170160598e-06,
+      "loss": 0.5824,
+      "step": 2065
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.361045495132401e-06,
+      "loss": 0.5496,
+      "step": 2066
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35184537168825e-06,
+      "loss": 0.5995,
+      "step": 2067
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342648808807368e-06,
+      "loss": 0.5407,
+      "step": 2068
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.33345581546552e-06,
+      "loss": 0.5959,
+      "step": 2069
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.324266400634969e-06,
+      "loss": 0.5999,
+      "step": 2070
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3150805732845e-06,
+      "loss": 0.5178,
+      "step": 2071
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.305898342379386e-06,
+      "loss": 0.5278,
+      "step": 2072
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.296719716881401e-06,
+      "loss": 0.6058,
+      "step": 2073
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287544705748787e-06,
+      "loss": 0.5285,
+      "step": 2074
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278373317936269e-06,
+      "loss": 0.6086,
+      "step": 2075
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.26920556239503e-06,
+      "loss": 0.6142,
+      "step": 2076
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.260041448072714e-06,
+      "loss": 0.5923,
+      "step": 2077
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.250880983913401e-06,
+      "loss": 0.631,
+      "step": 2078
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241724178857621e-06,
+      "loss": 0.52,
+      "step": 2079
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.232571041842319e-06,
+      "loss": 0.5951,
+      "step": 2080
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223421581800873e-06,
+      "loss": 0.6687,
+      "step": 2081
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2142758076630624e-06,
+      "loss": 0.5139,
+      "step": 2082
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.205133728355081e-06,
+      "loss": 0.6649,
+      "step": 2083
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.195995352799498e-06,
+      "loss": 0.5355,
+      "step": 2084
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186860689915286e-06,
+      "loss": 0.5869,
+      "step": 2085
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.177729748617784e-06,
+      "loss": 0.5304,
+      "step": 2086
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.168602537818706e-06,
+      "loss": 0.6029,
+      "step": 2087
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159479066426114e-06,
+      "loss": 0.5378,
+      "step": 2088
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1503593433444316e-06,
+      "loss": 0.5907,
+      "step": 2089
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.141243377474415e-06,
+      "loss": 0.538,
+      "step": 2090
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132131177713165e-06,
+      "loss": 0.5516,
+      "step": 2091
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123022752954091e-06,
+      "loss": 0.658,
+      "step": 2092
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.113918112086933e-06,
+      "loss": 0.5542,
+      "step": 2093
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.104817263997727e-06,
+      "loss": 0.6554,
+      "step": 2094
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095720217568819e-06,
+      "loss": 0.5103,
+      "step": 2095
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.086626981678827e-06,
+      "loss": 0.5812,
+      "step": 2096
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07753756520267e-06,
+      "loss": 0.5992,
+      "step": 2097
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068451977011522e-06,
+      "loss": 0.5026,
+      "step": 2098
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.059370225972834e-06,
+      "loss": 0.5716,
+      "step": 2099
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050292320950302e-06,
+      "loss": 0.5473,
+      "step": 2100
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0412182708038794e-06,
+      "loss": 0.5985,
+      "step": 2101
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.032148084389742e-06,
+      "loss": 0.5631,
+      "step": 2102
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023081770560307e-06,
+      "loss": 0.605,
+      "step": 2103
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0140193381642075e-06,
+      "loss": 0.5571,
+      "step": 2104
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.004960796046293e-06,
+      "loss": 0.5217,
+      "step": 2105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.995906153047606e-06,
+      "loss": 0.5735,
+      "step": 2106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.986855418005393e-06,
+      "loss": 0.5181,
+      "step": 2107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97780859975308e-06,
+      "loss": 0.5952,
+      "step": 2108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96876570712028e-06,
+      "loss": 0.6004,
+      "step": 2109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.959726748932759e-06,
+      "loss": 0.5564,
+      "step": 2110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95069173401246e-06,
+      "loss": 0.5529,
+      "step": 2111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941660671177463e-06,
+      "loss": 0.6156,
+      "step": 2112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.932633569242e-06,
+      "loss": 0.6598,
+      "step": 2113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9236104370164316e-06,
+      "loss": 0.626,
+      "step": 2114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9145912833072535e-06,
+      "loss": 0.5902,
+      "step": 2115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.905576116917065e-06,
+      "loss": 0.5763,
+      "step": 2116
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89656494664458e-06,
+      "loss": 0.5681,
+      "step": 2117
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887557781284616e-06,
+      "loss": 0.6511,
+      "step": 2118
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.878554629628081e-06,
+      "loss": 0.5336,
+      "step": 2119
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.869555500461956e-06,
+      "loss": 0.6089,
+      "step": 2120
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.860560402569308e-06,
+      "loss": 0.5412,
+      "step": 2121
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.851569344729263e-06,
+      "loss": 0.6165,
+      "step": 2122
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.842582335717009e-06,
+      "loss": 0.5588,
+      "step": 2123
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8335993843037695e-06,
+      "loss": 0.5275,
+      "step": 2124
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.82462049925683e-06,
+      "loss": 0.5612,
+      "step": 2125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815645689339484e-06,
+      "loss": 0.6036,
+      "step": 2126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8066749633110675e-06,
+      "loss": 0.5679,
+      "step": 2127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797708329926911e-06,
+      "loss": 0.6099,
+      "step": 2128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788745797938372e-06,
+      "loss": 0.5515,
+      "step": 2129
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779787376092785e-06,
+      "loss": 0.4854,
+      "step": 2130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.770833073133488e-06,
+      "loss": 0.537,
+      "step": 2131
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761882897799789e-06,
+      "loss": 0.6535,
+      "step": 2132
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7529368588269745e-06,
+      "loss": 0.5233,
+      "step": 2133
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.743994964946289e-06,
+      "loss": 0.651,
+      "step": 2134
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.735057224884939e-06,
+      "loss": 0.5467,
+      "step": 2135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.726123647366064e-06,
+      "loss": 0.5813,
+      "step": 2136
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.717194241108756e-06,
+      "loss": 0.5509,
+      "step": 2137
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.708269014828021e-06,
+      "loss": 0.5425,
+      "step": 2138
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.699347977234799e-06,
+      "loss": 0.6107,
+      "step": 2139
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690431137035928e-06,
+      "loss": 0.5616,
+      "step": 2140
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681518502934162e-06,
+      "loss": 0.6543,
+      "step": 2141
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6726100836281435e-06,
+      "loss": 0.669,
+      "step": 2142
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6637058878124075e-06,
+      "loss": 0.5634,
+      "step": 2143
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654805924177355e-06,
+      "loss": 0.579,
+      "step": 2144
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645910201409268e-06,
+      "loss": 0.5636,
+      "step": 2145
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637018728190282e-06,
+      "loss": 0.5355,
+      "step": 2146
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.628131513198392e-06,
+      "loss": 0.6147,
+      "step": 2147
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.619248565107427e-06,
+      "loss": 0.5641,
+      "step": 2148
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.610369892587064e-06,
+      "loss": 0.6186,
+      "step": 2149
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601495504302795e-06,
+      "loss": 0.5067,
+      "step": 2150
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592625408915939e-06,
+      "loss": 0.594,
+      "step": 2151
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.583759615083621e-06,
+      "loss": 0.4636,
+      "step": 2152
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.574898131458776e-06,
+      "loss": 0.5855,
+      "step": 2153
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.566040966690115e-06,
+      "loss": 0.581,
+      "step": 2154
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.557188129422153e-06,
+      "loss": 0.5843,
+      "step": 2155
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.548339628295165e-06,
+      "loss": 0.6221,
+      "step": 2156
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53949547194521e-06,
+      "loss": 0.6654,
+      "step": 2157
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.530655669004091e-06,
+      "loss": 0.6047,
+      "step": 2158
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5218202280993725e-06,
+      "loss": 0.5861,
+      "step": 2159
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.512989157854356e-06,
+      "loss": 0.5259,
+      "step": 2160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5041624668880825e-06,
+      "loss": 0.5318,
+      "step": 2161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.495340163815315e-06,
+      "loss": 0.5462,
+      "step": 2162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.486522257246538e-06,
+      "loss": 0.6086,
+      "step": 2163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477708755787939e-06,
+      "loss": 0.5911,
+      "step": 2164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468899668041411e-06,
+      "loss": 0.5655,
+      "step": 2165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460095002604533e-06,
+      "loss": 0.4891,
+      "step": 2166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451294768070581e-06,
+      "loss": 0.5898,
+      "step": 2167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442498973028489e-06,
+      "loss": 0.5586,
+      "step": 2168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.43370762606287e-06,
+      "loss": 0.5406,
+      "step": 2169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.424920735753994e-06,
+      "loss": 0.5114,
+      "step": 2170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.416138310677784e-06,
+      "loss": 0.5696,
+      "step": 2171
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.407360359405793e-06,
+      "loss": 0.5817,
+      "step": 2172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.398586890505224e-06,
+      "loss": 0.5895,
+      "step": 2173
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389817912538887e-06,
+      "loss": 0.4998,
+      "step": 2174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.381053434065229e-06,
+      "loss": 0.5197,
+      "step": 2175
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3722934636382875e-06,
+      "loss": 0.5776,
+      "step": 2176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363538009807714e-06,
+      "loss": 0.6221,
+      "step": 2177
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35478708111874e-06,
+      "loss": 0.5056,
+      "step": 2178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346040686112189e-06,
+      "loss": 0.5586,
+      "step": 2179
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337298833324458e-06,
+      "loss": 0.6132,
+      "step": 2180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.328561531287513e-06,
+      "loss": 0.5423,
+      "step": 2181
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.319828788528868e-06,
+      "loss": 0.5697,
+      "step": 2182
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311100613571603e-06,
+      "loss": 0.5846,
+      "step": 2183
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302377014934322e-06,
+      "loss": 0.5003,
+      "step": 2184
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293658001131182e-06,
+      "loss": 0.5397,
+      "step": 2185
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284943580671846e-06,
+      "loss": 0.5551,
+      "step": 2186
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.276233762061507e-06,
+      "loss": 0.6136,
+      "step": 2187
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.267528553800865e-06,
+      "loss": 0.4506,
+      "step": 2188
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.25882796438612e-06,
+      "loss": 0.3685,
+      "step": 2189
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250132002308954e-06,
+      "loss": 0.4477,
+      "step": 2190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.24144067605655e-06,
+      "loss": 0.4135,
+      "step": 2191
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.232753994111549e-06,
+      "loss": 0.4045,
+      "step": 2192
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.224071964952078e-06,
+      "loss": 0.4163,
+      "step": 2193
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2153945970517e-06,
+      "loss": 0.4295,
+      "step": 2194
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.206721898879454e-06,
+      "loss": 0.4323,
+      "step": 2195
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.198053878899798e-06,
+      "loss": 0.4228,
+      "step": 2196
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.18939054557264e-06,
+      "loss": 0.3948,
+      "step": 2197
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180731907353309e-06,
+      "loss": 0.42,
+      "step": 2198
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172077972692553e-06,
+      "loss": 0.4669,
+      "step": 2199
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163428750036521e-06,
+      "loss": 0.4757,
+      "step": 2200
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.15478424782678e-06,
+      "loss": 0.4215,
+      "step": 2201
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.146144474500268e-06,
+      "loss": 0.3722,
+      "step": 2202
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.13750943848933e-06,
+      "loss": 0.4191,
+      "step": 2203
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128879148221666e-06,
+      "loss": 0.4156,
+      "step": 2204
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.120253612120363e-06,
+      "loss": 0.4017,
+      "step": 2205
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.111632838603857e-06,
+      "loss": 0.3883,
+      "step": 2206
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.103016836085943e-06,
+      "loss": 0.3629,
+      "step": 2207
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.094405612975748e-06,
+      "loss": 0.3985,
+      "step": 2208
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08579917767775e-06,
+      "loss": 0.4543,
+      "step": 2209
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077197538591736e-06,
+      "loss": 0.4031,
+      "step": 2210
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068600704112832e-06,
+      "loss": 0.4508,
+      "step": 2211
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.060008682631454e-06,
+      "loss": 0.4207,
+      "step": 2212
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.051421482533342e-06,
+      "loss": 0.3489,
+      "step": 2213
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.042839112199509e-06,
+      "loss": 0.4209,
+      "step": 2214
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.03426158000627e-06,
+      "loss": 0.4061,
+      "step": 2215
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02568889432521e-06,
+      "loss": 0.4028,
+      "step": 2216
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.017121063523194e-06,
+      "loss": 0.3883,
+      "step": 2217
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.00855809596233e-06,
+      "loss": 0.4767,
+      "step": 2218
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.4086,
+      "step": 2219
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.991446783988818e-06,
+      "loss": 0.435,
+      "step": 2220
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9828984562766415e-06,
+      "loss": 0.3815,
+      "step": 2221
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.974355025206549e-06,
+      "loss": 0.4059,
+      "step": 2222
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.965816499116849e-06,
+      "loss": 0.395,
+      "step": 2223
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.95728288634106e-06,
+      "loss": 0.4211,
+      "step": 2224
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.948754195207908e-06,
+      "loss": 0.4465,
+      "step": 2225
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940230434041308e-06,
+      "loss": 0.3823,
+      "step": 2226
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.93171161116037e-06,
+      "loss": 0.4424,
+      "step": 2227
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.923197734879378e-06,
+      "loss": 0.3869,
+      "step": 2228
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.914688813507798e-06,
+      "loss": 0.4285,
+      "step": 2229
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906184855350247e-06,
+      "loss": 0.3421,
+      "step": 2230
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.897685868706512e-06,
+      "loss": 0.4132,
+      "step": 2231
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.889191861871515e-06,
+      "loss": 0.3222,
+      "step": 2232
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880702843135326e-06,
+      "loss": 0.4536,
+      "step": 2233
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872218820783145e-06,
+      "loss": 0.4418,
+      "step": 2234
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.863739803095299e-06,
+      "loss": 0.3933,
+      "step": 2235
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.855265798347217e-06,
+      "loss": 0.4436,
+      "step": 2236
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.846796814809455e-06,
+      "loss": 0.4223,
+      "step": 2237
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838332860747649e-06,
+      "loss": 0.4269,
+      "step": 2238
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829873944422544e-06,
+      "loss": 0.4568,
+      "step": 2239
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8214200740899505e-06,
+      "loss": 0.4145,
+      "step": 2240
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8129712580007725e-06,
+      "loss": 0.4089,
+      "step": 2241
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8045275044009635e-06,
+      "loss": 0.4248,
+      "step": 2242
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796088821531549e-06,
+      "loss": 0.4041,
+      "step": 2243
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7876552176286005e-06,
+      "loss": 0.4668,
+      "step": 2244
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.779226700923238e-06,
+      "loss": 0.4072,
+      "step": 2245
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.770803279641602e-06,
+      "loss": 0.4221,
+      "step": 2246
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.762384962004877e-06,
+      "loss": 0.3854,
+      "step": 2247
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.753971756229252e-06,
+      "loss": 0.4267,
+      "step": 2248
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.745563670525942e-06,
+      "loss": 0.4192,
+      "step": 2249
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737160713101149e-06,
+      "loss": 0.4148,
+      "step": 2250
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.728762892156079e-06,
+      "loss": 0.37,
+      "step": 2251
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.720370215886925e-06,
+      "loss": 0.4139,
+      "step": 2252
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.71198269248486e-06,
+      "loss": 0.349,
+      "step": 2253
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703600330136017e-06,
+      "loss": 0.415,
+      "step": 2254
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.695223137021509e-06,
+      "loss": 0.3442,
+      "step": 2255
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.686851121317385e-06,
+      "loss": 0.3937,
+      "step": 2256
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.67848429119466e-06,
+      "loss": 0.4006,
+      "step": 2257
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.670122654819269e-06,
+      "loss": 0.4469,
+      "step": 2258
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.661766220352098e-06,
+      "loss": 0.4044,
+      "step": 2259
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653414995948934e-06,
+      "loss": 0.3936,
+      "step": 2260
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6450689897604965e-06,
+      "loss": 0.3798,
+      "step": 2261
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.636728209932404e-06,
+      "loss": 0.3422,
+      "step": 2262
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.628392664605184e-06,
+      "loss": 0.4389,
+      "step": 2263
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.620062361914235e-06,
+      "loss": 0.4389,
+      "step": 2264
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.611737309989861e-06,
+      "loss": 0.3793,
+      "step": 2265
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.603417516957224e-06,
+      "loss": 0.4046,
+      "step": 2266
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.595102990936367e-06,
+      "loss": 0.4162,
+      "step": 2267
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.586793740042181e-06,
+      "loss": 0.3793,
+      "step": 2268
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.578489772384415e-06,
+      "loss": 0.411,
+      "step": 2269
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.570191096067661e-06,
+      "loss": 0.3987,
+      "step": 2270
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.561897719191349e-06,
+      "loss": 0.4009,
+      "step": 2271
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5536096498497295e-06,
+      "loss": 0.3969,
+      "step": 2272
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545326896131881e-06,
+      "loss": 0.4492,
+      "step": 2273
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5370494661216835e-06,
+      "loss": 0.3524,
+      "step": 2274
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.528777367897837e-06,
+      "loss": 0.4288,
+      "step": 2275
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520510609533819e-06,
+      "loss": 0.42,
+      "step": 2276
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.512249199097914e-06,
+      "loss": 0.3548,
+      "step": 2277
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.503993144653169e-06,
+      "loss": 0.4478,
+      "step": 2278
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.495742454257418e-06,
+      "loss": 0.4446,
+      "step": 2279
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.487497135963253e-06,
+      "loss": 0.39,
+      "step": 2280
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.479257197818026e-06,
+      "loss": 0.372,
+      "step": 2281
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.471022647863833e-06,
+      "loss": 0.4102,
+      "step": 2282
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4627934941375185e-06,
+      "loss": 0.4121,
+      "step": 2283
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45456974467065e-06,
+      "loss": 0.4611,
+      "step": 2284
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.446351407489533e-06,
+      "loss": 0.4125,
+      "step": 2285
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.438138490615178e-06,
+      "loss": 0.4305,
+      "step": 2286
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429931002063315e-06,
+      "loss": 0.3884,
+      "step": 2287
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.421728949844373e-06,
+      "loss": 0.3681,
+      "step": 2288
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413532341963477e-06,
+      "loss": 0.4034,
+      "step": 2289
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.40534118642043e-06,
+      "loss": 0.3945,
+      "step": 2290
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.397155491209727e-06,
+      "loss": 0.4196,
+      "step": 2291
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388975264320519e-06,
+      "loss": 0.4431,
+      "step": 2292
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380800513736635e-06,
+      "loss": 0.4328,
+      "step": 2293
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.372631247436545e-06,
+      "loss": 0.3854,
+      "step": 2294
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.36446747339338e-06,
+      "loss": 0.4571,
+      "step": 2295
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356309199574895e-06,
+      "loss": 0.3911,
+      "step": 2296
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.348156433943491e-06,
+      "loss": 0.3706,
+      "step": 2297
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340009184456188e-06,
+      "loss": 0.4773,
+      "step": 2298
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.331867459064623e-06,
+      "loss": 0.4369,
+      "step": 2299
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3237312657150364e-06,
+      "loss": 0.3961,
+      "step": 2300
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.315600612348278e-06,
+      "loss": 0.4191,
+      "step": 2301
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.30747550689978e-06,
+      "loss": 0.4411,
+      "step": 2302
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299355957299573e-06,
+      "loss": 0.3836,
+      "step": 2303
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.29124197147225e-06,
+      "loss": 0.5268,
+      "step": 2304
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2831335573369906e-06,
+      "loss": 0.4216,
+      "step": 2305
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.27503072280752e-06,
+      "loss": 0.3684,
+      "step": 2306
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2669334757921284e-06,
+      "loss": 0.4498,
+      "step": 2307
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.258841824193651e-06,
+      "loss": 0.4,
+      "step": 2308
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.250755775909465e-06,
+      "loss": 0.3991,
+      "step": 2309
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.242675338831464e-06,
+      "loss": 0.395,
+      "step": 2310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.234600520846085e-06,
+      "loss": 0.3928,
+      "step": 2311
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2265313298342645e-06,
+      "loss": 0.3773,
+      "step": 2312
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.218467773671461e-06,
+      "loss": 0.4155,
+      "step": 2313
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.210409860227619e-06,
+      "loss": 0.3237,
+      "step": 2314
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.202357597367187e-06,
+      "loss": 0.4125,
+      "step": 2315
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.194310992949095e-06,
+      "loss": 0.4564,
+      "step": 2316
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.186270054826754e-06,
+      "loss": 0.3977,
+      "step": 2317
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178234790848034e-06,
+      "loss": 0.4308,
+      "step": 2318
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170205208855281e-06,
+      "loss": 0.3872,
+      "step": 2319
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.162181316685282e-06,
+      "loss": 0.415,
+      "step": 2320
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.154163122169284e-06,
+      "loss": 0.4218,
+      "step": 2321
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146150633132958e-06,
+      "loss": 0.3754,
+      "step": 2322
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.138143857396425e-06,
+      "loss": 0.3823,
+      "step": 2323
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.13014280277421e-06,
+      "loss": 0.3971,
+      "step": 2324
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.12214747707527e-06,
+      "loss": 0.3721,
+      "step": 2325
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.114157888102963e-06,
+      "loss": 0.4065,
+      "step": 2326
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.106174043655054e-06,
+      "loss": 0.4222,
+      "step": 2327
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098195951523691e-06,
+      "loss": 0.3654,
+      "step": 2328
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090223619495419e-06,
+      "loss": 0.4106,
+      "step": 2329
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.082257055351152e-06,
+      "loss": 0.4478,
+      "step": 2330
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0742962668661826e-06,
+      "loss": 0.4167,
+      "step": 2331
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066341261810157e-06,
+      "loss": 0.3659,
+      "step": 2332
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0583920479470875e-06,
+      "loss": 0.4487,
+      "step": 2333
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.050448633035326e-06,
+      "loss": 0.4168,
+      "step": 2334
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.042511024827573e-06,
+      "loss": 0.339,
+      "step": 2335
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034579231070851e-06,
+      "loss": 0.3773,
+      "step": 2336
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026653259506518e-06,
+      "loss": 0.3976,
+      "step": 2337
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.018733117870238e-06,
+      "loss": 0.4942,
+      "step": 2338
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010818813892e-06,
+      "loss": 0.4534,
+      "step": 2339
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.00291035529608e-06,
+      "loss": 0.4496,
+      "step": 2340
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995007749801063e-06,
+      "loss": 0.4018,
+      "step": 2341
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987111005119808e-06,
+      "loss": 0.4126,
+      "step": 2342
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.979220128959463e-06,
+      "loss": 0.4265,
+      "step": 2343
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971335129021449e-06,
+      "loss": 0.4171,
+      "step": 2344
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.963456013001449e-06,
+      "loss": 0.3648,
+      "step": 2345
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.955582788589397e-06,
+      "loss": 0.3967,
+      "step": 2346
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947715463469493e-06,
+      "loss": 0.4165,
+      "step": 2347
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.939854045320158e-06,
+      "loss": 0.4446,
+      "step": 2348
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931998541814069e-06,
+      "loss": 0.3801,
+      "step": 2349
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.924148960618114e-06,
+      "loss": 0.3958,
+      "step": 2350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.91630530939341e-06,
+      "loss": 0.3722,
+      "step": 2351
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9084675957952845e-06,
+      "loss": 0.4208,
+      "step": 2352
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900635827473273e-06,
+      "loss": 0.3825,
+      "step": 2353
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892810012071096e-06,
+      "loss": 0.4043,
+      "step": 2354
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.884990157226683e-06,
+      "loss": 0.3835,
+      "step": 2355
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.877176270572126e-06,
+      "loss": 0.3436,
+      "step": 2356
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.869368359733711e-06,
+      "loss": 0.4226,
+      "step": 2357
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.861566432331875e-06,
+      "loss": 0.3872,
+      "step": 2358
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85377049598123e-06,
+      "loss": 0.389,
+      "step": 2359
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8459805582905265e-06,
+      "loss": 0.451,
+      "step": 2360
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.83819662686267e-06,
+      "loss": 0.4853,
+      "step": 2361
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8304187092947046e-06,
+      "loss": 0.3795,
+      "step": 2362
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822646813177803e-06,
+      "loss": 0.3891,
+      "step": 2363
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.814880946097253e-06,
+      "loss": 0.3936,
+      "step": 2364
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8071211156324727e-06,
+      "loss": 0.3824,
+      "step": 2365
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.799367329356973e-06,
+      "loss": 0.3928,
+      "step": 2366
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7916195948383817e-06,
+      "loss": 0.3987,
+      "step": 2367
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7838779196384036e-06,
+      "loss": 0.3799,
+      "step": 2368
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7761423113128427e-06,
+      "loss": 0.4344,
+      "step": 2369
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7684127774115766e-06,
+      "loss": 0.3795,
+      "step": 2370
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.760689325478559e-06,
+      "loss": 0.4123,
+      "step": 2371
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7529719630517956e-06,
+      "loss": 0.3703,
+      "step": 2372
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7452606976633644e-06,
+      "loss": 0.4016,
+      "step": 2373
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7375555368393767e-06,
+      "loss": 0.3546,
+      "step": 2374
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.729856488100003e-06,
+      "loss": 0.3487,
+      "step": 2375
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.722163558959433e-06,
+      "loss": 0.3692,
+      "step": 2376
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7144767569258957e-06,
+      "loss": 0.3974,
+      "step": 2377
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7067960895016277e-06,
+      "loss": 0.4165,
+      "step": 2378
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6991215641828903e-06,
+      "loss": 0.3992,
+      "step": 2379
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6914531884599437e-06,
+      "loss": 0.413,
+      "step": 2380
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6837909698170526e-06,
+      "loss": 0.3822,
+      "step": 2381
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.67613491573246e-06,
+      "loss": 0.3764,
+      "step": 2382
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.668485033678406e-06,
+      "loss": 0.4065,
+      "step": 2383
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6608413311210955e-06,
+      "loss": 0.3824,
+      "step": 2384
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.653203815520714e-06,
+      "loss": 0.3962,
+      "step": 2385
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6455724943313943e-06,
+      "loss": 0.476,
+      "step": 2386
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6379473750012375e-06,
+      "loss": 0.4251,
+      "step": 2387
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6303284649722804e-06,
+      "loss": 0.4268,
+      "step": 2388
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.622715771680508e-06,
+      "loss": 0.3753,
+      "step": 2389
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6151093025558336e-06,
+      "loss": 0.3952,
+      "step": 2390
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.607509065022101e-06,
+      "loss": 0.3701,
+      "step": 2391
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5999150664970617e-06,
+      "loss": 0.4653,
+      "step": 2392
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5923273143923885e-06,
+      "loss": 0.3967,
+      "step": 2393
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5847458161136485e-06,
+      "loss": 0.4477,
+      "step": 2394
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5771705790603163e-06,
+      "loss": 0.405,
+      "step": 2395
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5696016106257416e-06,
+      "loss": 0.4218,
+      "step": 2396
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.562038918197168e-06,
+      "loss": 0.4258,
+      "step": 2397
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.55448250915571e-06,
+      "loss": 0.43,
+      "step": 2398
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5469323908763507e-06,
+      "loss": 0.3547,
+      "step": 2399
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.539388570727925e-06,
+      "loss": 0.4172,
+      "step": 2400
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5318510560731366e-06,
+      "loss": 0.4463,
+      "step": 2401
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5243198542685176e-06,
+      "loss": 0.369,
+      "step": 2402
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5167949726644545e-06,
+      "loss": 0.3729,
+      "step": 2403
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.509276418605152e-06,
+      "loss": 0.3785,
+      "step": 2404
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.501764199428651e-06,
+      "loss": 0.3857,
+      "step": 2405
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4942583224667994e-06,
+      "loss": 0.4054,
+      "step": 2406
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.48675879504526e-06,
+      "loss": 0.4651,
+      "step": 2407
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.479265624483501e-06,
+      "loss": 0.4331,
+      "step": 2408
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4717788180947855e-06,
+      "loss": 0.4044,
+      "step": 2409
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4642983831861577e-06,
+      "loss": 0.4076,
+      "step": 2410
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4568243270584545e-06,
+      "loss": 0.4619,
+      "step": 2411
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4493566570062763e-06,
+      "loss": 0.4138,
+      "step": 2412
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4418953803180022e-06,
+      "loss": 0.4537,
+      "step": 2413
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.434440504275759e-06,
+      "loss": 0.3848,
+      "step": 2414
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4269920361554342e-06,
+      "loss": 0.4369,
+      "step": 2415
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4195499832266632e-06,
+      "loss": 0.4258,
+      "step": 2416
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4121143527528178e-06,
+      "loss": 0.4496,
+      "step": 2417
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4046851519909953e-06,
+      "loss": 0.4156,
+      "step": 2418
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3972623881920296e-06,
+      "loss": 0.3948,
+      "step": 2419
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3898460686004574e-06,
+      "loss": 0.4247,
+      "step": 2420
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382436200454543e-06,
+      "loss": 0.3877,
+      "step": 2421
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3750327909862378e-06,
+      "loss": 0.384,
+      "step": 2422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3676358474212035e-06,
+      "loss": 0.3997,
+      "step": 2423
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.360245376978779e-06,
+      "loss": 0.4005,
+      "step": 2424
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.352861386871993e-06,
+      "loss": 0.4915,
+      "step": 2425
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3454838843075487e-06,
+      "loss": 0.4402,
+      "step": 2426
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.338112876485821e-06,
+      "loss": 0.3163,
+      "step": 2427
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3307483706008314e-06,
+      "loss": 0.3871,
+      "step": 2428
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.323390373840276e-06,
+      "loss": 0.3945,
+      "step": 2429
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3160388933854792e-06,
+      "loss": 0.4144,
+      "step": 2430
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.308693936411421e-06,
+      "loss": 0.3853,
+      "step": 2431
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301355510086701e-06,
+      "loss": 0.4202,
+      "step": 2432
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2940236215735554e-06,
+      "loss": 0.4229,
+      "step": 2433
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2866982780278357e-06,
+      "loss": 0.3762,
+      "step": 2434
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2793794865990092e-06,
+      "loss": 0.4198,
+      "step": 2435
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.272067254430137e-06,
+      "loss": 0.4173,
+      "step": 2436
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.264761588657893e-06,
+      "loss": 0.3659,
+      "step": 2437
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.25746249641253e-06,
+      "loss": 0.3531,
+      "step": 2438
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.250169984817897e-06,
+      "loss": 0.408,
+      "step": 2439
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.242884060991408e-06,
+      "loss": 0.354,
+      "step": 2440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2356047320440586e-06,
+      "loss": 0.3963,
+      "step": 2441
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2283320050803966e-06,
+      "loss": 0.4204,
+      "step": 2442
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.221065887198537e-06,
+      "loss": 0.4615,
+      "step": 2443
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2138063854901392e-06,
+      "loss": 0.3735,
+      "step": 2444
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2065535070404085e-06,
+      "loss": 0.3502,
+      "step": 2445
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.199307258928077e-06,
+      "loss": 0.4197,
+      "step": 2446
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1920676482254186e-06,
+      "loss": 0.3596,
+      "step": 2447
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.184834681998218e-06,
+      "loss": 0.3691,
+      "step": 2448
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1776083673057834e-06,
+      "loss": 0.3971,
+      "step": 2449
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1703887112009226e-06,
+      "loss": 0.3726,
+      "step": 2450
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163175720729954e-06,
+      "loss": 0.4765,
+      "step": 2451
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1559694029326816e-06,
+      "loss": 0.4014,
+      "step": 2452
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1487697648424033e-06,
+      "loss": 0.3708,
+      "step": 2453
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1415768134858947e-06,
+      "loss": 0.4125,
+      "step": 2454
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.134390555883412e-06,
+      "loss": 0.4165,
+      "step": 2455
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1272109990486634e-06,
+      "loss": 0.3943,
+      "step": 2456
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.120038149988832e-06,
+      "loss": 0.3699,
+      "step": 2457
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1128720157045456e-06,
+      "loss": 0.4269,
+      "step": 2458
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1057126031898843e-06,
+      "loss": 0.4366,
+      "step": 2459
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.098559919432359e-06,
+      "loss": 0.4334,
+      "step": 2460
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0914139714129245e-06,
+      "loss": 0.3759,
+      "step": 2461
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0842747661059537e-06,
+      "loss": 0.4625,
+      "step": 2462
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0771423104792454e-06,
+      "loss": 0.3729,
+      "step": 2463
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.070016611494e-06,
+      "loss": 0.4013,
+      "step": 2464
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.062897676104837e-06,
+      "loss": 0.3918,
+      "step": 2465
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.05578551125976e-06,
+      "loss": 0.3619,
+      "step": 2466
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0486801239001806e-06,
+      "loss": 0.4259,
+      "step": 2467
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.041581520960881e-06,
+      "loss": 0.4203,
+      "step": 2468
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0344897093700333e-06,
+      "loss": 0.4126,
+      "step": 2469
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.027404696049171e-06,
+      "loss": 0.3527,
+      "step": 2470
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0203264879132e-06,
+      "loss": 0.4797,
+      "step": 2471
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.013255091870384e-06,
+      "loss": 0.3676,
+      "step": 2472
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.006190514822338e-06,
+      "loss": 0.4146,
+      "step": 2473
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9991327636640143e-06,
+      "loss": 0.4143,
+      "step": 2474
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.992081845283715e-06,
+      "loss": 0.3744,
+      "step": 2475
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9850377665630624e-06,
+      "loss": 0.4577,
+      "step": 2476
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9780005343770145e-06,
+      "loss": 0.4531,
+      "step": 2477
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9709701555938343e-06,
+      "loss": 0.4544,
+      "step": 2478
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.963946637075107e-06,
+      "loss": 0.4309,
+      "step": 2479
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.956929985675716e-06,
+      "loss": 0.3903,
+      "step": 2480
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9499202082438493e-06,
+      "loss": 0.4408,
+      "step": 2481
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9429173116209743e-06,
+      "loss": 0.4562,
+      "step": 2482
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9359213026418567e-06,
+      "loss": 0.3985,
+      "step": 2483
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9289321881345257e-06,
+      "loss": 0.4134,
+      "step": 2484
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.921949974920293e-06,
+      "loss": 0.4494,
+      "step": 2485
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9149746698137314e-06,
+      "loss": 0.4005,
+      "step": 2486
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.908006279622667e-06,
+      "loss": 0.4386,
+      "step": 2487
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9010448111481857e-06,
+      "loss": 0.4219,
+      "step": 2488
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8940902711846052e-06,
+      "loss": 0.4283,
+      "step": 2489
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8871426665194945e-06,
+      "loss": 0.3866,
+      "step": 2490
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.880202003933645e-06,
+      "loss": 0.3622,
+      "step": 2491
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87326829020108e-06,
+      "loss": 0.4054,
+      "step": 2492
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.866341532089031e-06,
+      "loss": 0.3935,
+      "step": 2493
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8594217363579504e-06,
+      "loss": 0.3852,
+      "step": 2494
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8525089097614867e-06,
+      "loss": 0.4271,
+      "step": 2495
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8456030590464967e-06,
+      "loss": 0.3724,
+      "step": 2496
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8387041909530164e-06,
+      "loss": 0.3757,
+      "step": 2497
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.831812312214278e-06,
+      "loss": 0.367,
+      "step": 2498
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8249274295566863e-06,
+      "loss": 0.4328,
+      "step": 2499
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.818049549699825e-06,
+      "loss": 0.4288,
+      "step": 2500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.811178679356429e-06,
+      "loss": 0.4038,
+      "step": 2501
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8043148252324083e-06,
+      "loss": 0.3788,
+      "step": 2502
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7974579940268096e-06,
+      "loss": 0.3462,
+      "step": 2503
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7906081924318416e-06,
+      "loss": 0.4309,
+      "step": 2504
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.783765427132837e-06,
+      "loss": 0.3606,
+      "step": 2505
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.776929704808272e-06,
+      "loss": 0.3728,
+      "step": 2506
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7701010321297416e-06,
+      "loss": 0.4469,
+      "step": 2507
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.763279415761965e-06,
+      "loss": 0.3969,
+      "step": 2508
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.756464862362772e-06,
+      "loss": 0.3981,
+      "step": 2509
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.749657378583106e-06,
+      "loss": 0.444,
+      "step": 2510
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742856971066996e-06,
+      "loss": 0.3528,
+      "step": 2511
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.736063646451581e-06,
+      "loss": 0.3015,
+      "step": 2512
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7292774113670738e-06,
+      "loss": 0.4143,
+      "step": 2513
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7224982724367775e-06,
+      "loss": 0.4469,
+      "step": 2514
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715726236277061e-06,
+      "loss": 0.4478,
+      "step": 2515
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7089613094973675e-06,
+      "loss": 0.4052,
+      "step": 2516
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702203498700201e-06,
+      "loss": 0.434,
+      "step": 2517
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6954528104811194e-06,
+      "loss": 0.4715,
+      "step": 2518
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688709251428725e-06,
+      "loss": 0.4384,
+      "step": 2519
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6819728281246706e-06,
+      "loss": 0.4478,
+      "step": 2520
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.675243547143632e-06,
+      "loss": 0.3795,
+      "step": 2521
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6685214150533303e-06,
+      "loss": 0.4101,
+      "step": 2522
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6618064384144925e-06,
+      "loss": 0.4111,
+      "step": 2523
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6550986237808775e-06,
+      "loss": 0.4832,
+      "step": 2524
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6483979776992406e-06,
+      "loss": 0.3616,
+      "step": 2525
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6417045067093495e-06,
+      "loss": 0.3938,
+      "step": 2526
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6350182173439666e-06,
+      "loss": 0.4296,
+      "step": 2527
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6283391161288484e-06,
+      "loss": 0.4535,
+      "step": 2528
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6216672095827267e-06,
+      "loss": 0.442,
+      "step": 2529
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.615002504217322e-06,
+      "loss": 0.4206,
+      "step": 2530
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6083450065373163e-06,
+      "loss": 0.3977,
+      "step": 2531
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.601694723040368e-06,
+      "loss": 0.4102,
+      "step": 2532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.595051660217084e-06,
+      "loss": 0.3395,
+      "step": 2533
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5884158245510325e-06,
+      "loss": 0.3905,
+      "step": 2534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.58178722251872e-06,
+      "loss": 0.3621,
+      "step": 2535
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5751658605896003e-06,
+      "loss": 0.3877,
+      "step": 2536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5685517452260566e-06,
+      "loss": 0.431,
+      "step": 2537
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.561944882883405e-06,
+      "loss": 0.3622,
+      "step": 2538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.555345280009872e-06,
+      "loss": 0.3457,
+      "step": 2539
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.548752943046613e-06,
+      "loss": 0.3901,
+      "step": 2540
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5421678784276772e-06,
+      "loss": 0.4601,
+      "step": 2541
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5355900925800282e-06,
+      "loss": 0.3413,
+      "step": 2542
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5290195919235173e-06,
+      "loss": 0.4431,
+      "step": 2543
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5224563828708904e-06,
+      "loss": 0.4266,
+      "step": 2544
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.515900471827775e-06,
+      "loss": 0.4061,
+      "step": 2545
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5093518651926795e-06,
+      "loss": 0.3743,
+      "step": 2546
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.502810569356976e-06,
+      "loss": 0.3476,
+      "step": 2547
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496276590704908e-06,
+      "loss": 0.3893,
+      "step": 2548
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4897499356135712e-06,
+      "loss": 0.3231,
+      "step": 2549
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4832306104529234e-06,
+      "loss": 0.4477,
+      "step": 2550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4767186215857542e-06,
+      "loss": 0.3923,
+      "step": 2551
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.470213975367709e-06,
+      "loss": 0.4626,
+      "step": 2552
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.463716678147251e-06,
+      "loss": 0.4123,
+      "step": 2553
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.457226736265684e-06,
+      "loss": 0.409,
+      "step": 2554
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4507441560571275e-06,
+      "loss": 0.4584,
+      "step": 2555
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.444268943848518e-06,
+      "loss": 0.4737,
+      "step": 2556
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.437801105959594e-06,
+      "loss": 0.4369,
+      "step": 2557
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.431340648702909e-06,
+      "loss": 0.379,
+      "step": 2558
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.424887578383799e-06,
+      "loss": 0.4225,
+      "step": 2559
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4184419013004047e-06,
+      "loss": 0.3725,
+      "step": 2560
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.412003623743636e-06,
+      "loss": 0.3679,
+      "step": 2561
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4055727519971948e-06,
+      "loss": 0.4918,
+      "step": 2562
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.399149292337547e-06,
+      "loss": 0.4011,
+      "step": 2563
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3927332510339296e-06,
+      "loss": 0.375,
+      "step": 2564
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3863246343483306e-06,
+      "loss": 0.4285,
+      "step": 2565
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379923448535503e-06,
+      "loss": 0.3898,
+      "step": 2566
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.373529699842936e-06,
+      "loss": 0.3863,
+      "step": 2567
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3671433945108712e-06,
+      "loss": 0.3964,
+      "step": 2568
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3607645387722753e-06,
+      "loss": 0.4208,
+      "step": 2569
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.354393138852855e-06,
+      "loss": 0.4141,
+      "step": 2570
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3480292009710282e-06,
+      "loss": 0.4247,
+      "step": 2571
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.341672731337937e-06,
+      "loss": 0.4232,
+      "step": 2572
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3353237361574376e-06,
+      "loss": 0.4512,
+      "step": 2573
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.328982221626087e-06,
+      "loss": 0.3785,
+      "step": 2574
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.322648193933137e-06,
+      "loss": 0.4337,
+      "step": 2575
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.316321659260542e-06,
+      "loss": 0.347,
+      "step": 2576
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.310002623782933e-06,
+      "loss": 0.4,
+      "step": 2577
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3036910936676325e-06,
+      "loss": 0.4042,
+      "step": 2578
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2973870750746253e-06,
+      "loss": 0.4679,
+      "step": 2579
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.291090574156576e-06,
+      "loss": 0.4154,
+      "step": 2580
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284801597058807e-06,
+      "loss": 0.4244,
+      "step": 2581
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2785201499193022e-06,
+      "loss": 0.3603,
+      "step": 2582
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.272246238868687e-06,
+      "loss": 0.4174,
+      "step": 2583
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2659798700302425e-06,
+      "loss": 0.4141,
+      "step": 2584
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2597210495198785e-06,
+      "loss": 0.4431,
+      "step": 2585
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2534697834461474e-06,
+      "loss": 0.3871,
+      "step": 2586
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2472260779102185e-06,
+      "loss": 0.4225,
+      "step": 2587
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240989939005893e-06,
+      "loss": 0.4059,
+      "step": 2588
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.234761372819577e-06,
+      "loss": 0.44,
+      "step": 2589
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2285403854302912e-06,
+      "loss": 0.3896,
+      "step": 2590
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2223269829096593e-06,
+      "loss": 0.4329,
+      "step": 2591
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2161211713219044e-06,
+      "loss": 0.3764,
+      "step": 2592
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2099229567238325e-06,
+      "loss": 0.4124,
+      "step": 2593
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2037323451648463e-06,
+      "loss": 0.4514,
+      "step": 2594
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1975493426869155e-06,
+      "loss": 0.4004,
+      "step": 2595
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.191373955324596e-06,
+      "loss": 0.3397,
+      "step": 2596
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1852061891050013e-06,
+      "loss": 0.4609,
+      "step": 2597
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.179046050047815e-06,
+      "loss": 0.3573,
+      "step": 2598
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1728935441652687e-06,
+      "loss": 0.384,
+      "step": 2599
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.166748677462148e-06,
+      "loss": 0.3844,
+      "step": 2600
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160611455935786e-06,
+      "loss": 0.3998,
+      "step": 2601
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.154481885576052e-06,
+      "loss": 0.4105,
+      "step": 2602
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1483599723653415e-06,
+      "loss": 0.411,
+      "step": 2603
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1422457222785875e-06,
+      "loss": 0.418,
+      "step": 2604
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1361391412832323e-06,
+      "loss": 0.3512,
+      "step": 2605
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.130040235339246e-06,
+      "loss": 0.3545,
+      "step": 2606
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1239490103990946e-06,
+      "loss": 0.4086,
+      "step": 2607
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1178654724077576e-06,
+      "loss": 0.3911,
+      "step": 2608
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.111789627302707e-06,
+      "loss": 0.4106,
+      "step": 2609
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.105721481013915e-06,
+      "loss": 0.4274,
+      "step": 2610
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0996610394638228e-06,
+      "loss": 0.4212,
+      "step": 2611
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.093608308567371e-06,
+      "loss": 0.4845,
+      "step": 2612
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.087563294231958e-06,
+      "loss": 0.4426,
+      "step": 2613
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0815260023574657e-06,
+      "loss": 0.4224,
+      "step": 2614
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0754964388362264e-06,
+      "loss": 0.4137,
+      "step": 2615
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0694746095530405e-06,
+      "loss": 0.427,
+      "step": 2616
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0634605203851486e-06,
+      "loss": 0.3977,
+      "step": 2617
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0574541772022472e-06,
+      "loss": 0.4419,
+      "step": 2618
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0514555858664663e-06,
+      "loss": 0.448,
+      "step": 2619
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045464752232377e-06,
+      "loss": 0.4287,
+      "step": 2620
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039481682146968e-06,
+      "loss": 0.4124,
+      "step": 2621
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0335063814496625e-06,
+      "loss": 0.4206,
+      "step": 2622
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027538855972291e-06,
+      "loss": 0.375,
+      "step": 2623
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0215791115391047e-06,
+      "loss": 0.4025,
+      "step": 2624
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0156271539667517e-06,
+      "loss": 0.3804,
+      "step": 2625
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0096829890642856e-06,
+      "loss": 0.4189,
+      "step": 2626
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003746622633155e-06,
+      "loss": 0.3976,
+      "step": 2627
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997818060467196e-06,
+      "loss": 0.4266,
+      "step": 2628
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.991897308352624e-06,
+      "loss": 0.3886,
+      "step": 2629
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9859843720680394e-06,
+      "loss": 0.4232,
+      "step": 2630
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.980079257384405e-06,
+      "loss": 0.3804,
+      "step": 2631
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.974181970065061e-06,
+      "loss": 0.4365,
+      "step": 2632
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9682925158656972e-06,
+      "loss": 0.4392,
+      "step": 2633
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962410900534367e-06,
+      "loss": 0.3976,
+      "step": 2634
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9565371298114666e-06,
+      "loss": 0.4447,
+      "step": 2635
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9506712094297386e-06,
+      "loss": 0.374,
+      "step": 2636
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.944813145114266e-06,
+      "loss": 0.4442,
+      "step": 2637
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938962942582465e-06,
+      "loss": 0.4924,
+      "step": 2638
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.93312060754407e-06,
+      "loss": 0.4046,
+      "step": 2639
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9272861457011482e-06,
+      "loss": 0.4164,
+      "step": 2640
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9214595627480726e-06,
+      "loss": 0.3634,
+      "step": 2641
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.915640864371536e-06,
+      "loss": 0.4254,
+      "step": 2642
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.4323,
+      "step": 2643
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9040271440563384e-06,
+      "loss": 0.3667,
+      "step": 2644
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8982321334525567e-06,
+      "loss": 0.4733,
+      "step": 2645
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8924450300950593e-06,
+      "loss": 0.4152,
+      "step": 2646
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8866658396319947e-06,
+      "loss": 0.3457,
+      "step": 2647
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.880894567703805e-06,
+      "loss": 0.3609,
+      "step": 2648
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.875131219943187e-06,
+      "loss": 0.3577,
+      "step": 2649
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8693758019751186e-06,
+      "loss": 0.4669,
+      "step": 2650
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.863628319416826e-06,
+      "loss": 0.4041,
+      "step": 2651
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.857888777877801e-06,
+      "loss": 0.3625,
+      "step": 2652
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.852157182959774e-06,
+      "loss": 0.3493,
+      "step": 2653
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.84643354025673e-06,
+      "loss": 0.4542,
+      "step": 2654
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8407178553548876e-06,
+      "loss": 0.3881,
+      "step": 2655
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.835010133832702e-06,
+      "loss": 0.4069,
+      "step": 2656
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.829310381260848e-06,
+      "loss": 0.4578,
+      "step": 2657
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.823618603202235e-06,
+      "loss": 0.5616,
+      "step": 2658
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.817934805211976e-06,
+      "loss": 0.4438,
+      "step": 2659
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.812258992837408e-06,
+      "loss": 0.4212,
+      "step": 2660
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8065911716180639e-06,
+      "loss": 0.419,
+      "step": 2661
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8009313470856838e-06,
+      "loss": 0.3949,
+      "step": 2662
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7952795247642008e-06,
+      "loss": 0.3585,
+      "step": 2663
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7896357101697405e-06,
+      "loss": 0.4216,
+      "step": 2664
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7839999088106074e-06,
+      "loss": 0.4476,
+      "step": 2665
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7783721261872933e-06,
+      "loss": 0.4059,
+      "step": 2666
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.772752367792452e-06,
+      "loss": 0.4316,
+      "step": 2667
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7671406391109204e-06,
+      "loss": 0.3486,
+      "step": 2668
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.761536945619684e-06,
+      "loss": 0.414,
+      "step": 2669
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7559412927878983e-06,
+      "loss": 0.364,
+      "step": 2670
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750353686076861e-06,
+      "loss": 0.417,
+      "step": 2671
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7447741309400246e-06,
+      "loss": 0.4002,
+      "step": 2672
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7392026328229804e-06,
+      "loss": 0.3746,
+      "step": 2673
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.733639197163458e-06,
+      "loss": 0.4223,
+      "step": 2674
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7280838293913116e-06,
+      "loss": 0.4563,
+      "step": 2675
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7225365349285294e-06,
+      "loss": 0.4302,
+      "step": 2676
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7169973191892142e-06,
+      "loss": 0.4258,
+      "step": 2677
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7114661875795902e-06,
+      "loss": 0.4196,
+      "step": 2678
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7059431454979825e-06,
+      "loss": 0.4355,
+      "step": 2679
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7004281983348313e-06,
+      "loss": 0.4329,
+      "step": 2680
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.694921351472667e-06,
+      "loss": 0.3774,
+      "step": 2681
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.68942261028612e-06,
+      "loss": 0.3802,
+      "step": 2682
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6839319801419073e-06,
+      "loss": 0.3834,
+      "step": 2683
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.678449466398835e-06,
+      "loss": 0.469,
+      "step": 2684
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6729750744077755e-06,
+      "loss": 0.3743,
+      "step": 2685
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6675088095116888e-06,
+      "loss": 0.4676,
+      "step": 2686
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.662050677045589e-06,
+      "loss": 0.4127,
+      "step": 2687
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6566006823365678e-06,
+      "loss": 0.4625,
+      "step": 2688
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6511588307037596e-06,
+      "loss": 0.4014,
+      "step": 2689
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6457251274583608e-06,
+      "loss": 0.3575,
+      "step": 2690
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6402995779036146e-06,
+      "loss": 0.413,
+      "step": 2691
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6348821873348053e-06,
+      "loss": 0.3273,
+      "step": 2692
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.629472961039248e-06,
+      "loss": 0.3987,
+      "step": 2693
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6240719042963004e-06,
+      "loss": 0.3548,
+      "step": 2694
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6186790223773375e-06,
+      "loss": 0.3767,
+      "step": 2695
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6132943205457607e-06,
+      "loss": 0.4114,
+      "step": 2696
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6079178040569853e-06,
+      "loss": 0.4135,
+      "step": 2697
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6025494781584416e-06,
+      "loss": 0.3936,
+      "step": 2698
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5971893480895583e-06,
+      "loss": 0.4475,
+      "step": 2699
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.591837419081773e-06,
+      "loss": 0.3412,
+      "step": 2700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5864936963585164e-06,
+      "loss": 0.4391,
+      "step": 2701
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5811581851352121e-06,
+      "loss": 0.4001,
+      "step": 2702
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.575830890619261e-06,
+      "loss": 0.4133,
+      "step": 2703
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5705118180100553e-06,
+      "loss": 0.419,
+      "step": 2704
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5652009724989547e-06,
+      "loss": 0.3736,
+      "step": 2705
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5598983592692974e-06,
+      "loss": 0.4096,
+      "step": 2706
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5546039834963745e-06,
+      "loss": 0.3781,
+      "step": 2707
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5493178503474505e-06,
+      "loss": 0.3899,
+      "step": 2708
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5440399649817384e-06,
+      "loss": 0.4005,
+      "step": 2709
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5387703325504045e-06,
+      "loss": 0.3961,
+      "step": 2710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5335089581965556e-06,
+      "loss": 0.3529,
+      "step": 2711
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5282558470552444e-06,
+      "loss": 0.35,
+      "step": 2712
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.523011004253453e-06,
+      "loss": 0.3623,
+      "step": 2713
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5177744349100997e-06,
+      "loss": 0.3524,
+      "step": 2714
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5125461441360223e-06,
+      "loss": 0.4487,
+      "step": 2715
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5073261370339853e-06,
+      "loss": 0.4824,
+      "step": 2716
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5021144186986592e-06,
+      "loss": 0.4403,
+      "step": 2717
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4969109942166338e-06,
+      "loss": 0.3839,
+      "step": 2718
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4917158686663992e-06,
+      "loss": 0.4184,
+      "step": 2719
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4865290471183503e-06,
+      "loss": 0.3455,
+      "step": 2720
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4813505346347701e-06,
+      "loss": 0.49,
+      "step": 2721
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4761803362698401e-06,
+      "loss": 0.3616,
+      "step": 2722
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4710184570696184e-06,
+      "loss": 0.474,
+      "step": 2723
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4658649020720539e-06,
+      "loss": 0.4076,
+      "step": 2724
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.460719676306962e-06,
+      "loss": 0.4241,
+      "step": 2725
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4555827847960336e-06,
+      "loss": 0.3806,
+      "step": 2726
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.450454232552826e-06,
+      "loss": 0.4366,
+      "step": 2727
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4453340245827573e-06,
+      "loss": 0.4239,
+      "step": 2728
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4402221658830963e-06,
+      "loss": 0.4413,
+      "step": 2729
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4351186614429725e-06,
+      "loss": 0.4004,
+      "step": 2730
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4300235162433496e-06,
+      "loss": 0.3688,
+      "step": 2731
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4249367352570464e-06,
+      "loss": 0.3678,
+      "step": 2732
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4198583234487052e-06,
+      "loss": 0.41,
+      "step": 2733
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.414788285774813e-06,
+      "loss": 0.4122,
+      "step": 2734
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4097266271836695e-06,
+      "loss": 0.4253,
+      "step": 2735
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4046733526154088e-06,
+      "loss": 0.4213,
+      "step": 2736
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3996284670019767e-06,
+      "loss": 0.4055,
+      "step": 2737
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3945919752671345e-06,
+      "loss": 0.4503,
+      "step": 2738
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3895638823264447e-06,
+      "loss": 0.3625,
+      "step": 2739
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3845441930872805e-06,
+      "loss": 0.39,
+      "step": 2740
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3795329124488055e-06,
+      "loss": 0.4023,
+      "step": 2741
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.374530045301986e-06,
+      "loss": 0.3874,
+      "step": 2742
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3695355965295653e-06,
+      "loss": 0.4391,
+      "step": 2743
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.364549571006083e-06,
+      "loss": 0.4227,
+      "step": 2744
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3595719735978451e-06,
+      "loss": 0.4108,
+      "step": 2745
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3546028091629414e-06,
+      "loss": 0.439,
+      "step": 2746
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.349642082551227e-06,
+      "loss": 0.4282,
+      "step": 2747
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.344689798604326e-06,
+      "loss": 0.4117,
+      "step": 2748
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.4154,
+      "step": 2749
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3348105780302301e-06,
+      "loss": 0.4233,
+      "step": 2750
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3298836510450597e-06,
+      "loss": 0.3795,
+      "step": 2751
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3249651860087364e-06,
+      "loss": 0.4198,
+      "step": 2752
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3200551877216327e-06,
+      "loss": 0.4013,
+      "step": 2753
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3151536609758586e-06,
+      "loss": 0.509,
+      "step": 2754
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3102606105552585e-06,
+      "loss": 0.3704,
+      "step": 2755
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3053760412354021e-06,
+      "loss": 0.4381,
+      "step": 2756
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3004999577835786e-06,
+      "loss": 0.4524,
+      "step": 2757
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2956323649588009e-06,
+      "loss": 0.4162,
+      "step": 2758
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2907732675117878e-06,
+      "loss": 0.3974,
+      "step": 2759
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2859226701849758e-06,
+      "loss": 0.3938,
+      "step": 2760
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2810805777124969e-06,
+      "loss": 0.5194,
+      "step": 2761
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2762469948201883e-06,
+      "loss": 0.4102,
+      "step": 2762
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2714219262255777e-06,
+      "loss": 0.4015,
+      "step": 2763
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.266605376637885e-06,
+      "loss": 0.4915,
+      "step": 2764
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2617973507580162e-06,
+      "loss": 0.425,
+      "step": 2765
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.256997853278561e-06,
+      "loss": 0.3876,
+      "step": 2766
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2522068888837758e-06,
+      "loss": 0.4053,
+      "step": 2767
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2474244622496002e-06,
+      "loss": 0.5085,
+      "step": 2768
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2426505780436326e-06,
+      "loss": 0.4916,
+      "step": 2769
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2378852409251396e-06,
+      "loss": 0.4372,
+      "step": 2770
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2331284555450406e-06,
+      "loss": 0.384,
+      "step": 2771
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2283802265459144e-06,
+      "loss": 0.349,
+      "step": 2772
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2236405585619847e-06,
+      "loss": 0.4704,
+      "step": 2773
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.218909456219124e-06,
+      "loss": 0.4459,
+      "step": 2774
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.214186924134838e-06,
+      "loss": 0.3582,
+      "step": 2775
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2094729669182748e-06,
+      "loss": 0.4526,
+      "step": 2776
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2047675891702082e-06,
+      "loss": 0.4046,
+      "step": 2777
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2000707954830438e-06,
+      "loss": 0.4098,
+      "step": 2778
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1953825904408033e-06,
+      "loss": 0.3735,
+      "step": 2779
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1907029786191316e-06,
+      "loss": 0.4045,
+      "step": 2780
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1860319645852814e-06,
+      "loss": 0.3717,
+      "step": 2781
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1813695528981172e-06,
+      "loss": 0.4188,
+      "step": 2782
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1767157481081092e-06,
+      "loss": 0.4503,
+      "step": 2783
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1720705547573263e-06,
+      "loss": 0.3727,
+      "step": 2784
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1674339773794286e-06,
+      "loss": 0.4758,
+      "step": 2785
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1628060204996738e-06,
+      "loss": 0.3285,
+      "step": 2786
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.158186688634898e-06,
+      "loss": 0.4266,
+      "step": 2787
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.153575986293528e-06,
+      "loss": 0.4008,
+      "step": 2788
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1489739179755622e-06,
+      "loss": 0.382,
+      "step": 2789
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.144380488172574e-06,
+      "loss": 0.3753,
+      "step": 2790
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1397957013677064e-06,
+      "loss": 0.4282,
+      "step": 2791
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1352195620356698e-06,
+      "loss": 0.4046,
+      "step": 2792
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1306520746427263e-06,
+      "loss": 0.3803,
+      "step": 2793
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1260932436467042e-06,
+      "loss": 0.4187,
+      "step": 2794
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1215430734969723e-06,
+      "loss": 0.4284,
+      "step": 2795
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1170015686344571e-06,
+      "loss": 0.4073,
+      "step": 2796
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1124687334916195e-06,
+      "loss": 0.4482,
+      "step": 2797
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1079445724924653e-06,
+      "loss": 0.3956,
+      "step": 2798
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1034290900525279e-06,
+      "loss": 0.4276,
+      "step": 2799
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0989222905788767e-06,
+      "loss": 0.4411,
+      "step": 2800
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0944241784701025e-06,
+      "loss": 0.355,
+      "step": 2801
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0899347581163222e-06,
+      "loss": 0.3939,
+      "step": 2802
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0854540338991615e-06,
+      "loss": 0.3707,
+      "step": 2803
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0809820101917666e-06,
+      "loss": 0.4777,
+      "step": 2804
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0765186913587866e-06,
+      "loss": 0.4432,
+      "step": 2805
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.072064081756381e-06,
+      "loss": 0.4416,
+      "step": 2806
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0676181857321998e-06,
+      "loss": 0.4581,
+      "step": 2807
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0631810076253967e-06,
+      "loss": 0.3596,
+      "step": 2808
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0587525517666142e-06,
+      "loss": 0.3683,
+      "step": 2809
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0543328224779848e-06,
+      "loss": 0.4187,
+      "step": 2810
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0499218240731157e-06,
+      "loss": 0.4218,
+      "step": 2811
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0455195608571046e-06,
+      "loss": 0.4424,
+      "step": 2812
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0411260371265096e-06,
+      "loss": 0.3463,
+      "step": 2813
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0367412571693746e-06,
+      "loss": 0.4791,
+      "step": 2814
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.032365225265196e-06,
+      "loss": 0.4212,
+      "step": 2815
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.027997945684942e-06,
+      "loss": 0.3894,
+      "step": 2816
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0236394226910329e-06,
+      "loss": 0.3486,
+      "step": 2817
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0192896605373436e-06,
+      "loss": 0.4695,
+      "step": 2818
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0149486634692019e-06,
+      "loss": 0.3653,
+      "step": 2819
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0106164357233794e-06,
+      "loss": 0.4126,
+      "step": 2820
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0062929815280842e-06,
+      "loss": 0.4565,
+      "step": 2821
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0019783051029696e-06,
+      "loss": 0.4043,
+      "step": 2822
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976724106591128e-07,
+      "loss": 0.3962,
+      "step": 2823
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.9337530239903e-07,
+      "loss": 0.3904,
+      "step": 2824
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.890869845166518e-07,
+      "loss": 0.4085,
+      "step": 2825
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.848074611973391e-07,
+      "loss": 0.4112,
+      "step": 2826
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.805367366178608e-07,
+      "loss": 0.3557,
+      "step": 2827
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.762748149464051e-07,
+      "loss": 0.4128,
+      "step": 2828
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.720217003425648e-07,
+      "loss": 0.3892,
+      "step": 2829
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.677773969573402e-07,
+      "loss": 0.4188,
+      "step": 2830
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.635419089331255e-07,
+      "loss": 0.4302,
+      "step": 2831
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59315240403721e-07,
+      "loss": 0.4353,
+      "step": 2832
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.5509739549431e-07,
+      "loss": 0.4552,
+      "step": 2833
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.508883783214695e-07,
+      "loss": 0.3774,
+      "step": 2834
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.466881929931582e-07,
+      "loss": 0.4052,
+      "step": 2835
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.424968436087167e-07,
+      "loss": 0.4035,
+      "step": 2836
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.383143342588596e-07,
+      "loss": 0.4236,
+      "step": 2837
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.341406690256793e-07,
+      "loss": 0.4187,
+      "step": 2838
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.299758519826274e-07,
+      "loss": 0.3497,
+      "step": 2839
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.25819887194529e-07,
+      "loss": 0.3669,
+      "step": 2840
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.216727787175605e-07,
+      "loss": 0.417,
+      "step": 2841
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.175345305992633e-07,
+      "loss": 0.3859,
+      "step": 2842
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.134051468785243e-07,
+      "loss": 0.4175,
+      "step": 2843
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.092846315855841e-07,
+      "loss": 0.4207,
+      "step": 2844
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.051729887420224e-07,
+      "loss": 0.4093,
+      "step": 2845
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.010702223607637e-07,
+      "loss": 0.4071,
+      "step": 2846
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.969763364460682e-07,
+      "loss": 0.3982,
+      "step": 2847
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.928913349935309e-07,
+      "loss": 0.4926,
+      "step": 2848
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.888152219900692e-07,
+      "loss": 0.4431,
+      "step": 2849
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.847480014139332e-07,
+      "loss": 0.4625,
+      "step": 2850
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.806896772346873e-07,
+      "loss": 0.434,
+      "step": 2851
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.76640253413219e-07,
+      "loss": 0.3871,
+      "step": 2852
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.725997339017233e-07,
+      "loss": 0.4425,
+      "step": 2853
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.685681226437093e-07,
+      "loss": 0.4567,
+      "step": 2854
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.645454235739903e-07,
+      "loss": 0.4269,
+      "step": 2855
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.605316406186836e-07,
+      "loss": 0.4295,
+      "step": 2856
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.56526777695198e-07,
+      "loss": 0.3833,
+      "step": 2857
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.525308387122433e-07,
+      "loss": 0.4324,
+      "step": 2858
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.485438275698154e-07,
+      "loss": 0.3647,
+      "step": 2859
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.445657481592007e-07,
+      "loss": 0.4309,
+      "step": 2860
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.405966043629621e-07,
+      "loss": 0.429,
+      "step": 2861
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.366364000549487e-07,
+      "loss": 0.3715,
+      "step": 2862
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.326851391002777e-07,
+      "loss": 0.4189,
+      "step": 2863
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.287428253553432e-07,
+      "loss": 0.3611,
+      "step": 2864
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.24809462667805e-07,
+      "loss": 0.3651,
+      "step": 2865
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.208850548765879e-07,
+      "loss": 0.3968,
+      "step": 2866
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.169696058118725e-07,
+      "loss": 0.4171,
+      "step": 2867
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.130631192951022e-07,
+      "loss": 0.4103,
+      "step": 2868
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.091655991389668e-07,
+      "loss": 0.3218,
+      "step": 2869
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.052770491474104e-07,
+      "loss": 0.3845,
+      "step": 2870
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.01397473115616e-07,
+      "loss": 0.4402,
+      "step": 2871
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.975268748300146e-07,
+      "loss": 0.4424,
+      "step": 2872
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.936652580682724e-07,
+      "loss": 0.436,
+      "step": 2873
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.898126265992912e-07,
+      "loss": 0.3744,
+      "step": 2874
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.859689841831975e-07,
+      "loss": 0.3855,
+      "step": 2875
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.821343345713528e-07,
+      "loss": 0.4142,
+      "step": 2876
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.783086815063346e-07,
+      "loss": 0.4247,
+      "step": 2877
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.744920287219459e-07,
+      "loss": 0.4646,
+      "step": 2878
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.706843799431985e-07,
+      "loss": 0.3995,
+      "step": 2879
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.668857388863249e-07,
+      "loss": 0.4061,
+      "step": 2880
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.630961092587575e-07,
+      "loss": 0.446,
+      "step": 2881
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.593154947591385e-07,
+      "loss": 0.4333,
+      "step": 2882
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.555438990773134e-07,
+      "loss": 0.3913,
+      "step": 2883
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.51781325894323e-07,
+      "loss": 0.4828,
+      "step": 2884
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.480277788823986e-07,
+      "loss": 0.5051,
+      "step": 2885
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.442832617049678e-07,
+      "loss": 0.3938,
+      "step": 2886
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.405477780166415e-07,
+      "loss": 0.4408,
+      "step": 2887
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.368213314632167e-07,
+      "loss": 0.457,
+      "step": 2888
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.331039256816664e-07,
+      "loss": 0.3854,
+      "step": 2889
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.293955643001449e-07,
+      "loss": 0.388,
+      "step": 2890
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.25696250937975e-07,
+      "loss": 0.4036,
+      "step": 2891
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.220059892056508e-07,
+      "loss": 0.3602,
+      "step": 2892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.183247827048323e-07,
+      "loss": 0.4634,
+      "step": 2893
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.146526350283411e-07,
+      "loss": 0.4242,
+      "step": 2894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.109895497601571e-07,
+      "loss": 0.4043,
+      "step": 2895
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.073355304754159e-07,
+      "loss": 0.4147,
+      "step": 2896
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.036905807404037e-07,
+      "loss": 0.4227,
+      "step": 2897
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.000547041125583e-07,
+      "loss": 0.4278,
+      "step": 2898
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.964279041404553e-07,
+      "loss": 0.3758,
+      "step": 2899
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.928101843638202e-07,
+      "loss": 0.4186,
+      "step": 2900
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.892015483135095e-07,
+      "loss": 0.4528,
+      "step": 2901
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.856019995115204e-07,
+      "loss": 0.342,
+      "step": 2902
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.820115414709727e-07,
+      "loss": 0.3903,
+      "step": 2903
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.784301776961222e-07,
+      "loss": 0.3265,
+      "step": 2904
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.748579116823417e-07,
+      "loss": 0.387,
+      "step": 2905
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.712947469161313e-07,
+      "loss": 0.4256,
+      "step": 2906
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.677406868751013e-07,
+      "loss": 0.46,
+      "step": 2907
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.641957350279838e-07,
+      "loss": 0.3489,
+      "step": 2908
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.606598948346132e-07,
+      "loss": 0.4808,
+      "step": 2909
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.57133169745936e-07,
+      "loss": 0.3882,
+      "step": 2910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.536155632040031e-07,
+      "loss": 0.4855,
+      "step": 2911
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.501070786419639e-07,
+      "loss": 0.4312,
+      "step": 2912
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.466077194840637e-07,
+      "loss": 0.4562,
+      "step": 2913
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.431174891456438e-07,
+      "loss": 0.4216,
+      "step": 2914
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.396363910331338e-07,
+      "loss": 0.4311,
+      "step": 2915
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.361644285440527e-07,
+      "loss": 0.3338,
+      "step": 2916
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.327016050670009e-07,
+      "loss": 0.4369,
+      "step": 2917
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.292479239816607e-07,
+      "loss": 0.4003,
+      "step": 2918
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.258033886587911e-07,
+      "loss": 0.429,
+      "step": 2919
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.223680024602263e-07,
+      "loss": 0.4462,
+      "step": 2920
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.189417687388665e-07,
+      "loss": 0.4915,
+      "step": 2921
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.155246908386869e-07,
+      "loss": 0.3962,
+      "step": 2922
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.121167720947174e-07,
+      "loss": 0.4059,
+      "step": 2923
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.087180158330564e-07,
+      "loss": 0.3783,
+      "step": 2924
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.053284253708547e-07,
+      "loss": 0.4111,
+      "step": 2925
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.019480040163228e-07,
+      "loss": 0.4076,
+      "step": 2926
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.98576755068715e-07,
+      "loss": 0.4305,
+      "step": 2927
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.952146818183391e-07,
+      "loss": 0.3894,
+      "step": 2928
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.918617875465449e-07,
+      "loss": 0.3928,
+      "step": 2929
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.88518075525728e-07,
+      "loss": 0.385,
+      "step": 2930
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.851835490193136e-07,
+      "loss": 0.4151,
+      "step": 2931
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.818582112817717e-07,
+      "loss": 0.4016,
+      "step": 2932
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.785420655585949e-07,
+      "loss": 0.4164,
+      "step": 2933
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.752351150863133e-07,
+      "loss": 0.3816,
+      "step": 2934
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.719373630924741e-07,
+      "loss": 0.3518,
+      "step": 2935
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.686488127956535e-07,
+      "loss": 0.4087,
+      "step": 2936
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.653694674054444e-07,
+      "loss": 0.353,
+      "step": 2937
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.620993301224564e-07,
+      "loss": 0.3806,
+      "step": 2938
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.588384041383089e-07,
+      "loss": 0.3503,
+      "step": 2939
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.555866926356379e-07,
+      "loss": 0.3928,
+      "step": 2940
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.523441987880774e-07,
+      "loss": 0.359,
+      "step": 2941
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.491109257602744e-07,
+      "loss": 0.4949,
+      "step": 2942
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.458868767078673e-07,
+      "loss": 0.3382,
+      "step": 2943
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.426720547775e-07,
+      "loss": 0.45,
+      "step": 2944
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.394664631068047e-07,
+      "loss": 0.4325,
+      "step": 2945
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.36270104824409e-07,
+      "loss": 0.402,
+      "step": 2946
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.330829830499263e-07,
+      "loss": 0.3733,
+      "step": 2947
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.299051008939593e-07,
+      "loss": 0.3802,
+      "step": 2948
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.267364614580861e-07,
+      "loss": 0.3606,
+      "step": 2949
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.235770678348706e-07,
+      "loss": 0.4116,
+      "step": 2950
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.204269231078484e-07,
+      "loss": 0.3818,
+      "step": 2951
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.172860303515314e-07,
+      "loss": 0.4029,
+      "step": 2952
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.141543926313974e-07,
+      "loss": 0.3771,
+      "step": 2953
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.110320130038982e-07,
+      "loss": 0.4414,
+      "step": 2954
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.079188945164426e-07,
+      "loss": 0.3926,
+      "step": 2955
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.048150402074047e-07,
+      "loss": 0.3893,
+      "step": 2956
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.017204531061159e-07,
+      "loss": 0.3803,
+      "step": 2957
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.986351362328656e-07,
+      "loss": 0.3998,
+      "step": 2958
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.955590925988896e-07,
+      "loss": 0.3811,
+      "step": 2959
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.924923252063785e-07,
+      "loss": 0.3973,
+      "step": 2960
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894348370484648e-07,
+      "loss": 0.3585,
+      "step": 2961
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.863866311092302e-07,
+      "loss": 0.4494,
+      "step": 2962
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.833477103636908e-07,
+      "loss": 0.3857,
+      "step": 2963
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.80318077777805e-07,
+      "loss": 0.3872,
+      "step": 2964
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.772977363084652e-07,
+      "loss": 0.4261,
+      "step": 2965
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7428668890349584e-07,
+      "loss": 0.485,
+      "step": 2966
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7128493850164715e-07,
+      "loss": 0.3476,
+      "step": 2967
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.682924880326012e-07,
+      "loss": 0.4436,
+      "step": 2968
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.653093404169573e-07,
+      "loss": 0.3865,
+      "step": 2969
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.623354985662409e-07,
+      "loss": 0.3803,
+      "step": 2970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5937096538289147e-07,
+      "loss": 0.3996,
+      "step": 2971
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5641574376026567e-07,
+      "loss": 0.3344,
+      "step": 2972
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.534698365826273e-07,
+      "loss": 0.3904,
+      "step": 2973
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.505332467251566e-07,
+      "loss": 0.4544,
+      "step": 2974
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.476059770539354e-07,
+      "loss": 0.3837,
+      "step": 2975
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4468803042595153e-07,
+      "loss": 0.4288,
+      "step": 2976
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4177940968909037e-07,
+      "loss": 0.4006,
+      "step": 2977
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.388801176821389e-07,
+      "loss": 0.4188,
+      "step": 2978
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.359901572347758e-07,
+      "loss": 0.3971,
+      "step": 2979
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3310953116757704e-07,
+      "loss": 0.3946,
+      "step": 2980
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3023824229200375e-07,
+      "loss": 0.4527,
+      "step": 2981
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.273762934104053e-07,
+      "loss": 0.4888,
+      "step": 2982
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.245236873160163e-07,
+      "loss": 0.3515,
+      "step": 2983
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.216804267929542e-07,
+      "loss": 0.406,
+      "step": 2984
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.188465146162113e-07,
+      "loss": 0.4321,
+      "step": 2985
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.160219535516596e-07,
+      "loss": 0.3957,
+      "step": 2986
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1320674635604186e-07,
+      "loss": 0.3868,
+      "step": 2987
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.104008957769745e-07,
+      "loss": 0.405,
+      "step": 2988
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0760440455293703e-07,
+      "loss": 0.381,
+      "step": 2989
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0481727541328173e-07,
+      "loss": 0.4228,
+      "step": 2990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.02039511078216e-07,
+      "loss": 0.4465,
+      "step": 2991
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9927111425881237e-07,
+      "loss": 0.4504,
+      "step": 2992
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9651208765699835e-07,
+      "loss": 0.4075,
+      "step": 2993
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.937624339655599e-07,
+      "loss": 0.4203,
+      "step": 2994
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.910221558681271e-07,
+      "loss": 0.4223,
+      "step": 2995
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.882912560391883e-07,
+      "loss": 0.4344,
+      "step": 2996
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8556973714407366e-07,
+      "loss": 0.3997,
+      "step": 2997
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8285760183895957e-07,
+      "loss": 0.305,
+      "step": 2998
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.801548527708621e-07,
+      "loss": 0.3842,
+      "step": 2999
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7746149257763984e-07,
+      "loss": 0.4107,
+      "step": 3000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.747775238879847e-07,
+      "loss": 0.355,
+      "step": 3001
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7210294932142566e-07,
+      "loss": 0.3673,
+      "step": 3002
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6943777148831907e-07,
+      "loss": 0.3988,
+      "step": 3003
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.667819929898553e-07,
+      "loss": 0.3674,
+      "step": 3004
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.641356164180454e-07,
+      "loss": 0.4633,
+      "step": 3005
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.614986443557289e-07,
+      "loss": 0.3663,
+      "step": 3006
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.588710793765626e-07,
+      "loss": 0.4374,
+      "step": 3007
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.562529240450274e-07,
+      "loss": 0.389,
+      "step": 3008
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5364418091641374e-07,
+      "loss": 0.4375,
+      "step": 3009
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.510448525368315e-07,
+      "loss": 0.4232,
+      "step": 3010
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4845494144320036e-07,
+      "loss": 0.3822,
+      "step": 3011
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.458744501632483e-07,
+      "loss": 0.3633,
+      "step": 3012
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.433033812155073e-07,
+      "loss": 0.4356,
+      "step": 3013
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4074173710931804e-07,
+      "loss": 0.4158,
+      "step": 3014
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.381895203448182e-07,
+      "loss": 0.3863,
+      "step": 3015
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3564673341294765e-07,
+      "loss": 0.4146,
+      "step": 3016
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3311337879544104e-07,
+      "loss": 0.4772,
+      "step": 3017
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.305894589648273e-07,
+      "loss": 0.3932,
+      "step": 3018
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.280749763844293e-07,
+      "loss": 0.3604,
+      "step": 3019
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.255699335083562e-07,
+      "loss": 0.4351,
+      "step": 3020
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2307433278150445e-07,
+      "loss": 0.4379,
+      "step": 3021
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.205881766395591e-07,
+      "loss": 0.3904,
+      "step": 3022
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1811146750898025e-07,
+      "loss": 0.4831,
+      "step": 3023
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.1564420780701435e-07,
+      "loss": 0.3948,
+      "step": 3024
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1318639994168176e-07,
+      "loss": 0.3357,
+      "step": 3025
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.107380463117804e-07,
+      "loss": 0.3717,
+      "step": 3026
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0829914930687767e-07,
+      "loss": 0.4273,
+      "step": 3027
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.058697113073128e-07,
+      "loss": 0.3699,
+      "step": 3028
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.034497346841958e-07,
+      "loss": 0.4297,
+      "step": 3029
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.010392217993996e-07,
+      "loss": 0.4513,
+      "step": 3030
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9863817500556e-07,
+      "loss": 0.4508,
+      "step": 3031
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9624659664607704e-07,
+      "loss": 0.4419,
+      "step": 3032
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.938644890551057e-07,
+      "loss": 0.434,
+      "step": 3033
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.914918545575618e-07,
+      "loss": 0.4136,
+      "step": 3034
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.891286954691108e-07,
+      "loss": 0.3719,
+      "step": 3035
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.867750140961756e-07,
+      "loss": 0.3975,
+      "step": 3036
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.844308127359241e-07,
+      "loss": 0.4057,
+      "step": 3037
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.820960936762762e-07,
+      "loss": 0.3925,
+      "step": 3038
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7977085919589253e-07,
+      "loss": 0.4323,
+      "step": 3039
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.77455111564181e-07,
+      "loss": 0.3737,
+      "step": 3040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7514885304128713e-07,
+      "loss": 0.3787,
+      "step": 3041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.728520858780992e-07,
+      "loss": 0.3739,
+      "step": 3042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.705648123162363e-07,
+      "loss": 0.3396,
+      "step": 3043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6828703458805725e-07,
+      "loss": 0.348,
+      "step": 3044
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.660187549166504e-07,
+      "loss": 0.3875,
+      "step": 3045
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.637599755158338e-07,
+      "loss": 0.4326,
+      "step": 3046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6151069859015386e-07,
+      "loss": 0.3803,
+      "step": 3047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.592709263348858e-07,
+      "loss": 0.369,
+      "step": 3048
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.570406609360221e-07,
+      "loss": 0.3478,
+      "step": 3049
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5481990457028147e-07,
+      "loss": 0.4071,
+      "step": 3050
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5260865940510027e-07,
+      "loss": 0.4115,
+      "step": 3051
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.504069275986332e-07,
+      "loss": 0.4032,
+      "step": 3052
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.482147112997468e-07,
+      "loss": 0.408,
+      "step": 3053
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4603201264802423e-07,
+      "loss": 0.338,
+      "step": 3054
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4385883377375683e-07,
+      "loss": 0.4498,
+      "step": 3055
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4169517679794694e-07,
+      "loss": 0.404,
+      "step": 3056
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.3954104383230316e-07,
+      "loss": 0.3991,
+      "step": 3057
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3739643697923699e-07,
+      "loss": 0.3931,
+      "step": 3058
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3526135833186527e-07,
+      "loss": 0.3769,
+      "step": 3059
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.331358099740022e-07,
+      "loss": 0.3781,
+      "step": 3060
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3101979398016284e-07,
+      "loss": 0.4167,
+      "step": 3061
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2891331241555958e-07,
+      "loss": 0.4357,
+      "step": 3062
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2681636733609457e-07,
+      "loss": 0.411,
+      "step": 3063
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2472896078836847e-07,
+      "loss": 0.3809,
+      "step": 3064
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2265109480966828e-07,
+      "loss": 0.3759,
+      "step": 3065
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.205827714279729e-07,
+      "loss": 0.3351,
+      "step": 3066
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1852399266194312e-07,
+      "loss": 0.4078,
+      "step": 3067
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.164747605209294e-07,
+      "loss": 0.4298,
+      "step": 3068
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1443507700495968e-07,
+      "loss": 0.4044,
+      "step": 3069
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.124049441047482e-07,
+      "loss": 0.4168,
+      "step": 3070
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1038436380168114e-07,
+      "loss": 0.3862,
+      "step": 3071
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0837333806782877e-07,
+      "loss": 0.3866,
+      "step": 3072
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0637186886592998e-07,
+      "loss": 0.4482,
+      "step": 3073
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.043799581493988e-07,
+      "loss": 0.3972,
+      "step": 3074
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0239760786232355e-07,
+      "loss": 0.3954,
+      "step": 3075
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.004248199394565e-07,
+      "loss": 0.4287,
+      "step": 3076
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9846159630621976e-07,
+      "loss": 0.3287,
+      "step": 3077
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9650793887870167e-07,
+      "loss": 0.4666,
+      "step": 3078
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9456384956365149e-07,
+      "loss": 0.4653,
+      "step": 3079
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.926293302584814e-07,
+      "loss": 0.4194,
+      "step": 3080
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9070438285126337e-07,
+      "loss": 0.3968,
+      "step": 3081
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8878900922072674e-07,
+      "loss": 0.3918,
+      "step": 3082
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8688321123625842e-07,
+      "loss": 0.3603,
+      "step": 3083
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8498699075789938e-07,
+      "loss": 0.3671,
+      "step": 3084
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8310034963634148e-07,
+      "loss": 0.3662,
+      "step": 3085
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8122328971292735e-07,
+      "loss": 0.411,
+      "step": 3086
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.793558128196493e-07,
+      "loss": 0.3898,
+      "step": 3087
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.774979207791472e-07,
+      "loss": 0.3751,
+      "step": 3088
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7564961540470492e-07,
+      "loss": 0.4145,
+      "step": 3089
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.738108985002518e-07,
+      "loss": 0.4156,
+      "step": 3090
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7198177186035447e-07,
+      "loss": 0.3538,
+      "step": 3091
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7016223727022606e-07,
+      "loss": 0.4563,
+      "step": 3092
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6835229650571272e-07,
+      "loss": 0.4318,
+      "step": 3093
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6655195133330027e-07,
+      "loss": 0.4556,
+      "step": 3094
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.647612035101054e-07,
+      "loss": 0.4149,
+      "step": 3095
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6298005478388445e-07,
+      "loss": 0.4644,
+      "step": 3096
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6120850689301692e-07,
+      "loss": 0.4072,
+      "step": 3097
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5944656156651973e-07,
+      "loss": 0.4236,
+      "step": 3098
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5769422052403172e-07,
+      "loss": 0.4214,
+      "step": 3099
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5595148547582373e-07,
+      "loss": 0.3471,
+      "step": 3100
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5421835812278408e-07,
+      "loss": 0.3826,
+      "step": 3101
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5249484015643078e-07,
+      "loss": 0.4191,
+      "step": 3102
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5078093325889943e-07,
+      "loss": 0.414,
+      "step": 3103
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4907663910294856e-07,
+      "loss": 0.3695,
+      "step": 3104
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4738195935194987e-07,
+      "loss": 0.3504,
+      "step": 3105
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4569689565989587e-07,
+      "loss": 0.4,
+      "step": 3106
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4402144967139098e-07,
+      "loss": 0.4408,
+      "step": 3107
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4235562302165605e-07,
+      "loss": 0.4349,
+      "step": 3108
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4069941733651948e-07,
+      "loss": 0.3898,
+      "step": 3109
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3905283423242267e-07,
+      "loss": 0.4106,
+      "step": 3110
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3741587531641566e-07,
+      "loss": 0.4568,
+      "step": 3111
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3578854218615379e-07,
+      "loss": 0.3934,
+      "step": 3112
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.341708364298966e-07,
+      "loss": 0.3587,
+      "step": 3113
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3256275962651223e-07,
+      "loss": 0.3482,
+      "step": 3114
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.309643133454641e-07,
+      "loss": 0.4227,
+      "step": 3115
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2937549914682434e-07,
+      "loss": 0.3628,
+      "step": 3116
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2779631858125696e-07,
+      "loss": 0.469,
+      "step": 3117
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2622677319002796e-07,
+      "loss": 0.3787,
+      "step": 3118
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2466686450499866e-07,
+      "loss": 0.337,
+      "step": 3119
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.231165940486234e-07,
+      "loss": 0.3732,
+      "step": 3120
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2157596333395084e-07,
+      "loss": 0.379,
+      "step": 3121
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2004497386462476e-07,
+      "loss": 0.4088,
+      "step": 3122
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.185236271348722e-07,
+      "loss": 0.4303,
+      "step": 3123
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1701192462951538e-07,
+      "loss": 0.4135,
+      "step": 3124
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1550986782395857e-07,
+      "loss": 0.4335,
+      "step": 3125
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1401745818419685e-07,
+      "loss": 0.398,
+      "step": 3126
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.125346971668051e-07,
+      "loss": 0.4214,
+      "step": 3127
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.110615862189457e-07,
+      "loss": 0.3703,
+      "step": 3128
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0959812677835968e-07,
+      "loss": 0.4058,
+      "step": 3129
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0814432027337008e-07,
+      "loss": 0.4304,
+      "step": 3130
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0670016812287631e-07,
+      "loss": 0.3833,
+      "step": 3131
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.052656717363587e-07,
+      "loss": 0.3883,
+      "step": 3132
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0384083251386957e-07,
+      "loss": 0.4382,
+      "step": 3133
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.024256518460387e-07,
+      "loss": 0.4119,
+      "step": 3134
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0102013111406905e-07,
+      "loss": 0.3921,
+      "step": 3135
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.962427168973554e-08,
+      "loss": 0.3583,
+      "step": 3136
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.823807493538284e-08,
+      "loss": 0.3517,
+      "step": 3137
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.68615422039243e-08,
+      "loss": 0.4075,
+      "step": 3138
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.549467483884412e-08,
+      "loss": 0.4028,
+      "step": 3139
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.413747417419184e-08,
+      "loss": 0.3722,
+      "step": 3140
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.278994153457898e-08,
+      "loss": 0.4121,
+      "step": 3141
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.145207823518798e-08,
+      "loss": 0.399,
+      "step": 3142
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.012388558175877e-08,
+      "loss": 0.4408,
+      "step": 3143
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.880536487059333e-08,
+      "loss": 0.4263,
+      "step": 3144
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.749651738855669e-08,
+      "loss": 0.4426,
+      "step": 3145
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.619734441307038e-08,
+      "loss": 0.4251,
+      "step": 3146
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.490784721211454e-08,
+      "loss": 0.3548,
+      "step": 3147
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.362802704422579e-08,
+      "loss": 0.3778,
+      "step": 3148
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.235788515849607e-08,
+      "loss": 0.4409,
+      "step": 3149
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.109742279457266e-08,
+      "loss": 0.3805,
+      "step": 3150
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.984664118265262e-08,
+      "loss": 0.3524,
+      "step": 3151
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.860554154348831e-08,
+      "loss": 0.3978,
+      "step": 3152
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.73741250883786e-08,
+      "loss": 0.4216,
+      "step": 3153
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.615239301917654e-08,
+      "loss": 0.4107,
+      "step": 3154
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.494034652827942e-08,
+      "loss": 0.4451,
+      "step": 3155
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.373798679863209e-08,
+      "loss": 0.4425,
+      "step": 3156
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.254531500372697e-08,
+      "loss": 0.3676,
+      "step": 3157
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.13623323076007e-08,
+      "loss": 0.3631,
+      "step": 3158
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.018903986483083e-08,
+      "loss": 0.3998,
+      "step": 3159
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.902543882054024e-08,
+      "loss": 0.4354,
+      "step": 3160
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.787153031039273e-08,
+      "loss": 0.4337,
+      "step": 3161
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.672731546059185e-08,
+      "loss": 0.4579,
+      "step": 3162
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.559279538787877e-08,
+      "loss": 0.4266,
+      "step": 3163
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.44679711995344e-08,
+      "loss": 0.3983,
+      "step": 3164
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.335284399337726e-08,
+      "loss": 0.4073,
+      "step": 3165
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.224741485776008e-08,
+      "loss": 0.4123,
+      "step": 3166
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.115168487157097e-08,
+      "loss": 0.3634,
+      "step": 3167
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.006565510423112e-08,
+      "loss": 0.3746,
+      "step": 3168
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.89893266156949e-08,
+      "loss": 0.3842,
+      "step": 3169
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.792270045644976e-08,
+      "loss": 0.4376,
+      "step": 3170
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.686577766751078e-08,
+      "loss": 0.4444,
+      "step": 3171
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5818559280425014e-08,
+      "loss": 0.357,
+      "step": 3172
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4781046317267103e-08,
+      "loss": 0.4542,
+      "step": 3173
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.375323979063929e-08,
+      "loss": 0.4088,
+      "step": 3174
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.273514070367247e-08,
+      "loss": 0.4113,
+      "step": 3175
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.172675005001959e-08,
+      "loss": 0.4344,
+      "step": 3176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.072806881386005e-08,
+      "loss": 0.3995,
+      "step": 3177
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.973909796989862e-08,
+      "loss": 0.3617,
+      "step": 3178
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8759838483358745e-08,
+      "loss": 0.4155,
+      "step": 3179
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.779029130999036e-08,
+      "loss": 0.4347,
+      "step": 3180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.683045739606096e-08,
+      "loss": 0.4452,
+      "step": 3181
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.588033767836009e-08,
+      "loss": 0.3877,
+      "step": 3182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4939933084192646e-08,
+      "loss": 0.31,
+      "step": 3183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.400924453138777e-08,
+      "loss": 0.4711,
+      "step": 3184
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3088272928285545e-08,
+      "loss": 0.3744,
+      "step": 3185
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2177019173748054e-08,
+      "loss": 0.3997,
+      "step": 3186
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.1275484157147216e-08,
+      "loss": 0.4035,
+      "step": 3187
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.038366875837363e-08,
+      "loss": 0.399,
+      "step": 3188
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.950157384783104e-08,
+      "loss": 0.4275,
+      "step": 3189
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.862920028643524e-08,
+      "loss": 0.4184,
+      "step": 3190
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.776654892561293e-08,
+      "loss": 0.4148,
+      "step": 3191
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.691362060730397e-08,
+      "loss": 0.3791,
+      "step": 3192
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.607041616396023e-08,
+      "loss": 0.4454,
+      "step": 3193
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.523693641853787e-08,
+      "loss": 0.352,
+      "step": 3194
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4413182184507285e-08,
+      "loss": 0.3947,
+      "step": 3195
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.359915426584315e-08,
+      "loss": 0.5129,
+      "step": 3196
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.279485345702993e-08,
+      "loss": 0.3729,
+      "step": 3197
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.200028054305637e-08,
+      "loss": 0.3855,
+      "step": 3198
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.12154362994177e-08,
+      "loss": 0.3777,
+      "step": 3199
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.044032149211562e-08,
+      "loss": 0.4618,
+      "step": 3200
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.967493687765277e-08,
+      "loss": 0.4006,
+      "step": 3201
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8919283203038274e-08,
+      "loss": 0.3908,
+      "step": 3202
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.817336120578329e-08,
+      "loss": 0.3892,
+      "step": 3203
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7437171613898806e-08,
+      "loss": 0.3865,
+      "step": 3204
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.671071514589896e-08,
+      "loss": 0.3357,
+      "step": 3205
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5993992510799925e-08,
+      "loss": 0.4049,
+      "step": 3206
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.528700440811438e-08,
+      "loss": 0.4743,
+      "step": 3207
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.458975152785592e-08,
+      "loss": 0.4173,
+      "step": 3208
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3902234550536862e-08,
+      "loss": 0.3815,
+      "step": 3209
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.322445414716712e-08,
+      "loss": 0.4196,
+      "step": 3210
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2556410979253095e-08,
+      "loss": 0.3886,
+      "step": 3211
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.18981056987988e-08,
+      "loss": 0.4726,
+      "step": 3212
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1249538948304726e-08,
+      "loss": 0.4131,
+      "step": 3213
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0610711360764536e-08,
+      "loss": 0.3476,
+      "step": 3214
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.998162355966726e-08,
+      "loss": 0.3902,
+      "step": 3215
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.936227615899733e-08,
+      "loss": 0.4094,
+      "step": 3216
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.875266976323231e-08,
+      "loss": 0.4182,
+      "step": 3217
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.815280496734073e-08,
+      "loss": 0.3581,
+      "step": 3218
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7562682356786488e-08,
+      "loss": 0.4264,
+      "step": 3219
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.698230250752109e-08,
+      "loss": 0.3908,
+      "step": 3220
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.641166598599142e-08,
+      "loss": 0.3952,
+      "step": 3221
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.58507733491331e-08,
+      "loss": 0.4129,
+      "step": 3222
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5299625144370444e-08,
+      "loss": 0.4412,
+      "step": 3223
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4758221909620951e-08,
+      "loss": 0.4575,
+      "step": 3224
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4226564173286383e-08,
+      "loss": 0.3719,
+      "step": 3225
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.370465245426167e-08,
+      "loss": 0.4361,
+      "step": 3226
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3192487261926013e-08,
+      "loss": 0.4128,
+      "step": 3227
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2690069096148449e-08,
+      "loss": 0.3937,
+      "step": 3228
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2197398447283404e-08,
+      "loss": 0.3843,
+      "step": 3229
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.171447579617402e-08,
+      "loss": 0.3886,
+      "step": 3230
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1241301614147715e-08,
+      "loss": 0.4483,
+      "step": 3231
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.0777876363016193e-08,
+      "loss": 0.4629,
+      "step": 3232
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0324200495079873e-08,
+      "loss": 0.3608,
+      "step": 3233
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.88027445312123e-09,
+      "loss": 0.4464,
+      "step": 3234
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.446098670408132e-09,
+      "loss": 0.3898,
+      "step": 3235
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.021673570690504e-09,
+      "loss": 0.5015,
+      "step": 3236
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.606999568204766e-09,
+      "loss": 0.3825,
+      "step": 3237
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.20207706766829e-09,
+      "loss": 0.4409,
+      "step": 3238
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.806906464281617e-09,
+      "loss": 0.4126,
+      "step": 3239
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.42148814372623e-09,
+      "loss": 0.419,
+      "step": 3240
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.045822482167896e-09,
+      "loss": 0.4312,
+      "step": 3241
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.6799098462522195e-09,
+      "loss": 0.3651,
+      "step": 3242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.323750593106859e-09,
+      "loss": 0.43,
+      "step": 3243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.977345070339314e-09,
+      "loss": 0.4008,
+      "step": 3244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6406936160369184e-09,
+      "loss": 0.3904,
+      "step": 3245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.313796558770179e-09,
+      "loss": 0.3937,
+      "step": 3246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.996654217584995e-09,
+      "loss": 0.4284,
+      "step": 3247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.689266902011547e-09,
+      "loss": 0.391,
+      "step": 3248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.39163491205652e-09,
+      "loss": 0.4064,
+      "step": 3249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.103758538205327e-09,
+      "loss": 0.3999,
+      "step": 3250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.825638061421e-09,
+      "loss": 0.3948,
+      "step": 3251
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.557273753149737e-09,
+      "loss": 0.3987,
+      "step": 3252
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.298665875309803e-09,
+      "loss": 0.3715,
+      "step": 3253
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0498146803004113e-09,
+      "loss": 0.4121,
+      "step": 3254
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.810720410998391e-09,
+      "loss": 0.3441,
+      "step": 3255
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.581383300757079e-09,
+      "loss": 0.3667,
+      "step": 3256
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3618035734074285e-09,
+      "loss": 0.4026,
+      "step": 3257
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.151981443258011e-09,
+      "loss": 0.3519,
+      "step": 3258
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.951917115091684e-09,
+      "loss": 0.3633,
+      "step": 3259
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7616107841700313e-09,
+      "loss": 0.4023,
+      "step": 3260
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5810626362300351e-09,
+      "loss": 0.3803,
+      "step": 3261
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4102728474840733e-09,
+      "loss": 0.4233,
+      "step": 3262
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.249241584623251e-09,
+      "loss": 0.39,
+      "step": 3263
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.0979690048107395e-09,
+      "loss": 0.4895,
+      "step": 3264
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.564552556884377e-10,
+      "loss": 0.469,
+      "step": 3265
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.247004753714205e-10,
+      "loss": 0.3693,
+      "step": 3266
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.027047924512698e-10,
+      "loss": 0.3927,
+      "step": 3267
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.904683259938537e-10,
+      "loss": 0.4833,
+      "step": 3268
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.879911855426578e-10,
+      "loss": 0.3829,
+      "step": 3269
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9527347111323334e-10,
+      "loss": 0.373,
+      "step": 3270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1231527319763864e-10,
+      "loss": 0.4144,
+      "step": 3271
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3911667276110826e-10,
+      "loss": 0.3935,
+      "step": 3272
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.7567774124649384e-10,
+      "loss": 0.3997,
+      "step": 3273
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2199854056760273e-10,
+      "loss": 0.4199,
+      "step": 3274
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.807912311696974e-11,
+      "loss": 0.3677,
+      "step": 3275
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.3919531758085385e-11,
+      "loss": 0.4332,
+      "step": 3276
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.951979982983687e-11,
+      "loss": 0.427,
+      "step": 3277
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.8799511487285235e-12,
+      "loss": 0.3856,
+      "step": 3278
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.3848,
+      "step": 3279
+    },
+    {
+      "epoch": 3.0,
+      "step": 3279,
+      "total_flos": 5.1664382177837056e+17,
+      "train_loss": 0.6346134240669461,
+      "train_runtime": 82540.7152,
+      "train_samples_per_second": 2.544,
+      "train_steps_per_second": 0.04
+    }
+  ],
+  "max_steps": 3279,
+  "num_train_epochs": 3,
+  "total_flos": 5.1664382177837056e+17,
+  "trial_name": null,
+  "trial_params": null
+}