diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,13128 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2183,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.3116,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0,
+      "loss": 2.4443,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5151515151515152e-07,
+      "loss": 2.299,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 2.1658,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0303030303030305e-07,
+      "loss": 2.264,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5454545454545457e-07,
+      "loss": 2.4009,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.060606060606061e-07,
+      "loss": 2.3274,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.575757575757576e-07,
+      "loss": 2.2898,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.090909090909091e-07,
+      "loss": 2.2966,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0606060606060608e-06,
+      "loss": 2.0644,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2121212121212122e-06,
+      "loss": 2.2205,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 2.0171,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5151515151515152e-06,
+      "loss": 2.1568,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 2.0365,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8181818181818183e-06,
+      "loss": 2.1208,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.96969696969697e-06,
+      "loss": 1.9468,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1212121212121216e-06,
+      "loss": 1.9272,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2727272727272728e-06,
+      "loss": 2.0293,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4242424242424244e-06,
+      "loss": 1.9671,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.575757575757576e-06,
+      "loss": 2.0938,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 2.0327,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8787878787878793e-06,
+      "loss": 2.0776,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 1.9891,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.181818181818182e-06,
+      "loss": 1.9221,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.9059,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4848484848484854e-06,
+      "loss": 1.9644,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6363636363636366e-06,
+      "loss": 2.0763,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7878787878787882e-06,
+      "loss": 1.8119,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.93939393939394e-06,
+      "loss": 1.9721,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0909090909090915e-06,
+      "loss": 2.0631,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.242424242424243e-06,
+      "loss": 1.8763,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.393939393939394e-06,
+      "loss": 1.8895,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5454545454545455e-06,
+      "loss": 1.9066,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.696969696969698e-06,
+      "loss": 1.8424,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.848484848484849e-06,
+      "loss": 1.8139,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5e-06,
+      "loss": 1.8814,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.151515151515152e-06,
+      "loss": 1.7488,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.303030303030303e-06,
+      "loss": 1.8175,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4545454545454545e-06,
+      "loss": 1.7793,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.606060606060606e-06,
+      "loss": 1.8755,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7575757575757586e-06,
+      "loss": 1.8576,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.90909090909091e-06,
+      "loss": 1.8591,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 1.8603,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.212121212121213e-06,
+      "loss": 1.7547,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.363636363636364e-06,
+      "loss": 1.8274,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.515151515151516e-06,
+      "loss": 1.936,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.7835,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.818181818181818e-06,
+      "loss": 1.8579,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.969696969696971e-06,
+      "loss": 1.8045,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.121212121212122e-06,
+      "loss": 1.8113,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.803,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.272727272727273e-06,
+      "loss": 1.985,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.424242424242425e-06,
+      "loss": 1.7711,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.5757575757575764e-06,
+      "loss": 1.978,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.727272727272727e-06,
+      "loss": 1.8875,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.87878787878788e-06,
+      "loss": 1.7774,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.03030303030303e-06,
+      "loss": 1.8698,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.181818181818183e-06,
+      "loss": 1.7954,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.8322,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.484848484848486e-06,
+      "loss": 1.8505,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.636363636363637e-06,
+      "loss": 1.8603,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.787878787878788e-06,
+      "loss": 1.754,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.93939393939394e-06,
+      "loss": 1.7675,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 1.7466,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.242424242424244e-06,
+      "loss": 1.8745,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.393939393939396e-06,
+      "loss": 1.769,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.545454545454547e-06,
+      "loss": 1.7856,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.696969696969698e-06,
+      "loss": 1.698,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.84848484848485e-06,
+      "loss": 1.8281,
+      "step": 69
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.8458,
+      "step": 70
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994494484898e-06,
+      "loss": 1.6841,
+      "step": 71
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999977977951714e-06,
+      "loss": 1.7922,
+      "step": 72
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999950450436825e-06,
+      "loss": 1.7274,
+      "step": 73
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999911912000844e-06,
+      "loss": 1.8196,
+      "step": 74
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99986236272865e-06,
+      "loss": 1.805,
+      "step": 75
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999801802729355e-06,
+      "loss": 1.7949,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999730232136324e-06,
+      "loss": 1.9433,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999647651107174e-06,
+      "loss": 1.843,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999554059823764e-06,
+      "loss": 1.6954,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9994494584922e-06,
+      "loss": 1.717,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999333847342839e-06,
+      "loss": 1.7683,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999207226630277e-06,
+      "loss": 1.7478,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999069596633358e-06,
+      "loss": 1.7982,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998920957655176e-06,
+      "loss": 1.7016,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998761310023063e-06,
+      "loss": 1.8007,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998590654088593e-06,
+      "loss": 1.7538,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998408990227588e-06,
+      "loss": 1.8722,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99821631884011e-06,
+      "loss": 1.7873,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998012640350462e-06,
+      "loss": 1.8208,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997797955207182e-06,
+      "loss": 1.8416,
+      "step": 90
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997572263883054e-06,
+      "loss": 1.7317,
+      "step": 91
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997335566875096e-06,
+      "loss": 1.8471,
+      "step": 92
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997087864704564e-06,
+      "loss": 1.755,
+      "step": 93
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996829157916948e-06,
+      "loss": 1.8759,
+      "step": 94
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996559447081975e-06,
+      "loss": 1.8325,
+      "step": 95
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996278732793601e-06,
+      "loss": 1.7353,
+      "step": 96
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995987015670021e-06,
+      "loss": 1.8072,
+      "step": 97
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995684296353655e-06,
+      "loss": 1.7426,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995370575511151e-06,
+      "loss": 1.7185,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.995045853833388e-06,
+      "loss": 1.7952,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994710132035472e-06,
+      "loss": 1.8242,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994363410856729e-06,
+      "loss": 1.7205,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99400569106071e-06,
+      "loss": 1.7003,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993636973435192e-06,
+      "loss": 1.8897,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993257258792164e-06,
+      "loss": 1.6777,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992866547967834e-06,
+      "loss": 1.7894,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992464841822632e-06,
+      "loss": 1.7648,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992052141241195e-06,
+      "loss": 1.7036,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991628447132375e-06,
+      "loss": 1.6607,
+      "step": 109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991193760429235e-06,
+      "loss": 1.8439,
+      "step": 110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990748082089043e-06,
+      "loss": 1.7479,
+      "step": 111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990291413093276e-06,
+      "loss": 1.6732,
+      "step": 112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989823754447611e-06,
+      "loss": 1.6816,
+      "step": 113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989345107181934e-06,
+      "loss": 1.7329,
+      "step": 114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988855472350317e-06,
+      "loss": 1.6812,
+      "step": 115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988354851031044e-06,
+      "loss": 1.8068,
+      "step": 116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987843244326581e-06,
+      "loss": 1.728,
+      "step": 117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987320653363595e-06,
+      "loss": 1.7296,
+      "step": 118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986787079292936e-06,
+      "loss": 1.7027,
+      "step": 119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986242523289646e-06,
+      "loss": 1.7465,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98568698655295e-06,
+      "loss": 1.7284,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.985120470306253e-06,
+      "loss": 1.7682,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.984542975797143e-06,
+      "loss": 1.6816,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983954504297378e-06,
+      "loss": 1.5699,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983355057102894e-06,
+      "loss": 1.719,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982744635533801e-06,
+      "loss": 1.7225,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98212324093437e-06,
+      "loss": 1.7204,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98149087467304e-06,
+      "loss": 1.5817,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980847538142413e-06,
+      "loss": 1.8018,
+      "step": 129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980193232759247e-06,
+      "loss": 1.7954,
+      "step": 130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979527959964459e-06,
+      "loss": 1.7492,
+      "step": 131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978851721223118e-06,
+      "loss": 1.7421,
+      "step": 132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978164518024436e-06,
+      "loss": 1.8022,
+      "step": 133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97746635188178e-06,
+      "loss": 1.7864,
+      "step": 134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976757224332656e-06,
+      "loss": 1.6565,
+      "step": 135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976037136938707e-06,
+      "loss": 1.858,
+      "step": 136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975306091285715e-06,
+      "loss": 1.625,
+      "step": 137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974564088983592e-06,
+      "loss": 1.6859,
+      "step": 138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97381113166638e-06,
+      "loss": 1.6953,
+      "step": 139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973047220992249e-06,
+      "loss": 1.7634,
+      "step": 140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972272358643485e-06,
+      "loss": 1.6537,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.971486546326494e-06,
+      "loss": 1.6967,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.970689785771798e-06,
+      "loss": 1.7003,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969882078734028e-06,
+      "loss": 1.6881,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.969063426991922e-06,
+      "loss": 1.6465,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.968233832348317e-06,
+      "loss": 1.7757,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967393296630155e-06,
+      "loss": 1.7269,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966541821688465e-06,
+      "loss": 1.7021,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965679409398375e-06,
+      "loss": 1.7148,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964806061659089e-06,
+      "loss": 1.7926,
+      "step": 150
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963921780393903e-06,
+      "loss": 1.8619,
+      "step": 151
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963026567550183e-06,
+      "loss": 1.7416,
+      "step": 152
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962120425099377e-06,
+      "loss": 1.6847,
+      "step": 153
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961203355036992e-06,
+      "loss": 1.7508,
+      "step": 154
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96027535938261e-06,
+      "loss": 1.6549,
+      "step": 155
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959336440179863e-06,
+      "loss": 1.7012,
+      "step": 156
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95838659949645e-06,
+      "loss": 1.7428,
+      "step": 157
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957425839424115e-06,
+      "loss": 1.7579,
+      "step": 158
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956454162078648e-06,
+      "loss": 1.738,
+      "step": 159
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955471569599881e-06,
+      "loss": 1.6351,
+      "step": 160
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95447806415169e-06,
+      "loss": 1.6951,
+      "step": 161
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953473647921974e-06,
+      "loss": 1.6922,
+      "step": 162
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952458323122666e-06,
+      "loss": 1.6966,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.951432091989721e-06,
+      "loss": 1.802,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.950394956783114e-06,
+      "loss": 1.6566,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.949346919786824e-06,
+      "loss": 1.7781,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.948287983308851e-06,
+      "loss": 1.6543,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.947218149681185e-06,
+      "loss": 1.6117,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946137421259828e-06,
+      "loss": 1.5883,
+      "step": 169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945045800424758e-06,
+      "loss": 1.7729,
+      "step": 170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943943289579955e-06,
+      "loss": 1.8322,
+      "step": 171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942829891153374e-06,
+      "loss": 1.7653,
+      "step": 172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941705607596945e-06,
+      "loss": 1.6262,
+      "step": 173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940570441386574e-06,
+      "loss": 1.6685,
+      "step": 174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939424395022132e-06,
+      "loss": 1.6884,
+      "step": 175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938267471027447e-06,
+      "loss": 1.6107,
+      "step": 176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937099671950306e-06,
+      "loss": 1.7655,
+      "step": 177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935921000362444e-06,
+      "loss": 1.8156,
+      "step": 178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934731458859533e-06,
+      "loss": 1.7259,
+      "step": 179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933531050061194e-06,
+      "loss": 1.4785,
+      "step": 180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932319776610974e-06,
+      "loss": 1.8502,
+      "step": 181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931097641176344e-06,
+      "loss": 1.7199,
+      "step": 182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9298646464487e-06,
+      "loss": 1.7281,
+      "step": 183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928620795143351e-06,
+      "loss": 1.6366,
+      "step": 184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927366089999513e-06,
+      "loss": 1.7473,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.926100533780304e-06,
+      "loss": 1.613,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.924824129272741e-06,
+      "loss": 1.6629,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.923536879287729e-06,
+      "loss": 1.6517,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.922238786660061e-06,
+      "loss": 1.7602,
+      "step": 189
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920929854248399e-06,
+      "loss": 1.7398,
+      "step": 190
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919610084935285e-06,
+      "loss": 1.7627,
+      "step": 191
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918279481627122e-06,
+      "loss": 1.8459,
+      "step": 192
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916938047254175e-06,
+      "loss": 1.7799,
+      "step": 193
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915585784770553e-06,
+      "loss": 1.6332,
+      "step": 194
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914222697154224e-06,
+      "loss": 1.5952,
+      "step": 195
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912848787406983e-06,
+      "loss": 1.8259,
+      "step": 196
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911464058554466e-06,
+      "loss": 1.5771,
+      "step": 197
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910068513646125e-06,
+      "loss": 1.7151,
+      "step": 198
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908662155755244e-06,
+      "loss": 1.6757,
+      "step": 199
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90724498797891e-06,
+      "loss": 1.7478,
+      "step": 200
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905817013438018e-06,
+      "loss": 1.6642,
+      "step": 201
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904378235277263e-06,
+      "loss": 1.7585,
+      "step": 202
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90292865666513e-06,
+      "loss": 1.6802,
+      "step": 203
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901468280793891e-06,
+      "loss": 1.6798,
+      "step": 204
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899997110879595e-06,
+      "loss": 1.8138,
+      "step": 205
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89851515016206e-06,
+      "loss": 1.6632,
+      "step": 206
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897022401904867e-06,
+      "loss": 1.4913,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89551886939536e-06,
+      "loss": 1.7985,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.894004555944625e-06,
+      "loss": 1.748,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.89247946488749e-06,
+      "loss": 1.6592,
+      "step": 210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.890943599582524e-06,
+      "loss": 1.6431,
+      "step": 211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889396963412014e-06,
+      "loss": 1.6748,
+      "step": 212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887839559781976e-06,
+      "loss": 1.6459,
+      "step": 213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88627139212213e-06,
+      "loss": 1.6893,
+      "step": 214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88469246388591e-06,
+      "loss": 1.8014,
+      "step": 215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883102778550434e-06,
+      "loss": 1.812,
+      "step": 216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881502339616521e-06,
+      "loss": 1.8156,
+      "step": 217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879891150608667e-06,
+      "loss": 1.6975,
+      "step": 218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87826921507504e-06,
+      "loss": 1.8109,
+      "step": 219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87663653658748e-06,
+      "loss": 1.6321,
+      "step": 220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874993118741477e-06,
+      "loss": 1.672,
+      "step": 221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873338965156179e-06,
+      "loss": 1.6618,
+      "step": 222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871674079474372e-06,
+      "loss": 1.6143,
+      "step": 223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869998465362478e-06,
+      "loss": 1.6376,
+      "step": 224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868312126510542e-06,
+      "loss": 1.539,
+      "step": 225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866615066632232e-06,
+      "loss": 1.6216,
+      "step": 226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864907289464823e-06,
+      "loss": 1.6573,
+      "step": 227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863188798769192e-06,
+      "loss": 1.8353,
+      "step": 228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861459598329808e-06,
+      "loss": 1.6579,
+      "step": 229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.859719691954732e-06,
+      "loss": 1.6236,
+      "step": 230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85796908347559e-06,
+      "loss": 1.5851,
+      "step": 231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.856207776747585e-06,
+      "loss": 1.7528,
+      "step": 232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.854435775649479e-06,
+      "loss": 1.7842,
+      "step": 233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852653084083581e-06,
+      "loss": 1.7454,
+      "step": 234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850859705975746e-06,
+      "loss": 1.6648,
+      "step": 235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849055645275363e-06,
+      "loss": 1.783,
+      "step": 236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847240905955345e-06,
+      "loss": 1.6808,
+      "step": 237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84541549201212e-06,
+      "loss": 1.6182,
+      "step": 238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843579407465629e-06,
+      "loss": 1.7736,
+      "step": 239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841732656359305e-06,
+      "loss": 1.597,
+      "step": 240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839875242760077e-06,
+      "loss": 1.6286,
+      "step": 241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838007170758351e-06,
+      "loss": 1.7545,
+      "step": 242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836128444468007e-06,
+      "loss": 1.721,
+      "step": 243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834239068026388e-06,
+      "loss": 1.6425,
+      "step": 244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83233904559429e-06,
+      "loss": 1.5364,
+      "step": 245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830428381355952e-06,
+      "loss": 1.6493,
+      "step": 246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828507079519053e-06,
+      "loss": 1.702,
+      "step": 247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826575144314694e-06,
+      "loss": 1.6733,
+      "step": 248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824632579997394e-06,
+      "loss": 1.8164,
+      "step": 249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82267939084508e-06,
+      "loss": 1.8015,
+      "step": 250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82071558115908e-06,
+      "loss": 1.6783,
+      "step": 251
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.818741155264104e-06,
+      "loss": 1.6873,
+      "step": 252
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.816756117508244e-06,
+      "loss": 1.8048,
+      "step": 253
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.814760472262966e-06,
+      "loss": 1.6717,
+      "step": 254
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81275422392309e-06,
+      "loss": 1.675,
+      "step": 255
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810737376906785e-06,
+      "loss": 1.7082,
+      "step": 256
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80870993565557e-06,
+      "loss": 1.5686,
+      "step": 257
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806671904634284e-06,
+      "loss": 1.5714,
+      "step": 258
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804623288331092e-06,
+      "loss": 1.6215,
+      "step": 259
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80256409125747e-06,
+      "loss": 1.701,
+      "step": 260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800494317948193e-06,
+      "loss": 1.6514,
+      "step": 261
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798413972961329e-06,
+      "loss": 1.6892,
+      "step": 262
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796323060878227e-06,
+      "loss": 1.7518,
+      "step": 263
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794221586303505e-06,
+      "loss": 1.588,
+      "step": 264
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792109553865044e-06,
+      "loss": 1.862,
+      "step": 265
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789986968213975e-06,
+      "loss": 1.6423,
+      "step": 266
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787853834024667e-06,
+      "loss": 1.8554,
+      "step": 267
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785710155994721e-06,
+      "loss": 1.6416,
+      "step": 268
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783555938844962e-06,
+      "loss": 1.6351,
+      "step": 269
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781391187319415e-06,
+      "loss": 1.7694,
+      "step": 270
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77921590618531e-06,
+      "loss": 1.6163,
+      "step": 271
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777030100233066e-06,
+      "loss": 1.6783,
+      "step": 272
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.774833774276278e-06,
+      "loss": 1.6065,
+      "step": 273
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.772626933151705e-06,
+      "loss": 1.7209,
+      "step": 274
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.770409581719269e-06,
+      "loss": 1.7292,
+      "step": 275
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.768181724862035e-06,
+      "loss": 1.6839,
+      "step": 276
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765943367486201e-06,
+      "loss": 1.6271,
+      "step": 277
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763694514521092e-06,
+      "loss": 1.6894,
+      "step": 278
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761435170919145e-06,
+      "loss": 1.6193,
+      "step": 279
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7591653416559e-06,
+      "loss": 1.7061,
+      "step": 280
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756885031729992e-06,
+      "loss": 1.5983,
+      "step": 281
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754594246163128e-06,
+      "loss": 1.7138,
+      "step": 282
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752292990000094e-06,
+      "loss": 1.7307,
+      "step": 283
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749981268308727e-06,
+      "loss": 1.7488,
+      "step": 284
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747659086179916e-06,
+      "loss": 1.7277,
+      "step": 285
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745326448727589e-06,
+      "loss": 1.7015,
+      "step": 286
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742983361088685e-06,
+      "loss": 1.6921,
+      "step": 287
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740629828423171e-06,
+      "loss": 1.687,
+      "step": 288
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738265855914014e-06,
+      "loss": 1.7457,
+      "step": 289
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735891448767163e-06,
+      "loss": 1.5315,
+      "step": 290
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733506612211552e-06,
+      "loss": 1.6384,
+      "step": 291
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731111351499087e-06,
+      "loss": 1.6853,
+      "step": 292
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72870567190462e-06,
+      "loss": 1.6834,
+      "step": 293
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726289578725961e-06,
+      "loss": 1.8206,
+      "step": 294
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.723863077283836e-06,
+      "loss": 1.7153,
+      "step": 295
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.721426172921906e-06,
+      "loss": 1.5856,
+      "step": 296
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.718978871006735e-06,
+      "loss": 1.7085,
+      "step": 297
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71652117692779e-06,
+      "loss": 1.6063,
+      "step": 298
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.714053096097413e-06,
+      "loss": 1.6824,
+      "step": 299
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711574633950831e-06,
+      "loss": 1.5363,
+      "step": 300
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709085795946128e-06,
+      "loss": 1.6667,
+      "step": 301
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706586587564236e-06,
+      "loss": 1.5651,
+      "step": 302
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.704077014308929e-06,
+      "loss": 1.702,
+      "step": 303
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701557081706803e-06,
+      "loss": 1.6029,
+      "step": 304
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699026795307272e-06,
+      "loss": 1.6767,
+      "step": 305
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696486160682543e-06,
+      "loss": 1.7301,
+      "step": 306
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693935183427619e-06,
+      "loss": 1.6391,
+      "step": 307
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691373869160278e-06,
+      "loss": 1.5971,
+      "step": 308
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688802223521063e-06,
+      "loss": 1.7054,
+      "step": 309
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686220252173266e-06,
+      "loss": 1.6328,
+      "step": 310
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683627960802918e-06,
+      "loss": 1.6551,
+      "step": 311
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681025355118783e-06,
+      "loss": 1.6772,
+      "step": 312
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678412440852331e-06,
+      "loss": 1.7143,
+      "step": 313
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67578922375774e-06,
+      "loss": 1.7057,
+      "step": 314
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673155709611875e-06,
+      "loss": 1.6636,
+      "step": 315
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670511904214273e-06,
+      "loss": 1.7056,
+      "step": 316
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.667857813387142e-06,
+      "loss": 1.5951,
+      "step": 317
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.665193442975337e-06,
+      "loss": 1.6008,
+      "step": 318
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.662518798846348e-06,
+      "loss": 1.7538,
+      "step": 319
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.659833886890295e-06,
+      "loss": 1.6958,
+      "step": 320
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657138713019904e-06,
+      "loss": 1.5976,
+      "step": 321
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654433283170506e-06,
+      "loss": 1.7866,
+      "step": 322
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651717603300016e-06,
+      "loss": 1.8341,
+      "step": 323
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648991679388915e-06,
+      "loss": 1.6232,
+      "step": 324
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646255517440255e-06,
+      "loss": 1.7241,
+      "step": 325
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643509123479628e-06,
+      "loss": 1.6465,
+      "step": 326
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640752503555156e-06,
+      "loss": 1.5925,
+      "step": 327
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637985663737486e-06,
+      "loss": 1.6765,
+      "step": 328
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635208610119769e-06,
+      "loss": 1.6461,
+      "step": 329
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63242134881765e-06,
+      "loss": 1.9406,
+      "step": 330
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62962388596925e-06,
+      "loss": 1.5495,
+      "step": 331
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626816227735161e-06,
+      "loss": 1.6354,
+      "step": 332
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623998380298426e-06,
+      "loss": 1.7783,
+      "step": 333
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621170349864524e-06,
+      "loss": 1.6127,
+      "step": 334
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61833214266136e-06,
+      "loss": 1.5844,
+      "step": 335
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615483764939252e-06,
+      "loss": 1.6665,
+      "step": 336
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612625222970915e-06,
+      "loss": 1.5923,
+      "step": 337
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609756523051448e-06,
+      "loss": 1.6404,
+      "step": 338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.606877671498317e-06,
+      "loss": 1.634,
+      "step": 339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.603988674651346e-06,
+      "loss": 1.6494,
+      "step": 340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.601089538872705e-06,
+      "loss": 1.7287,
+      "step": 341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.598180270546886e-06,
+      "loss": 1.5243,
+      "step": 342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595260876080697e-06,
+      "loss": 1.7048,
+      "step": 343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592331361903247e-06,
+      "loss": 1.6151,
+      "step": 344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589391734465928e-06,
+      "loss": 1.5798,
+      "step": 345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586442000242407e-06,
+      "loss": 1.7473,
+      "step": 346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583482165728608e-06,
+      "loss": 1.6146,
+      "step": 347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580512237442694e-06,
+      "loss": 1.8188,
+      "step": 348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577532221925058e-06,
+      "loss": 1.6592,
+      "step": 349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57454212573831e-06,
+      "loss": 1.6146,
+      "step": 350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57154195546726e-06,
+      "loss": 1.677,
+      "step": 351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568531717718897e-06,
+      "loss": 1.6102,
+      "step": 352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565511419122387e-06,
+      "loss": 1.6078,
+      "step": 353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56248106632905e-06,
+      "loss": 1.7001,
+      "step": 354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559440666012345e-06,
+      "loss": 1.6702,
+      "step": 355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556390224867862e-06,
+      "loss": 1.6726,
+      "step": 356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553329749613301e-06,
+      "loss": 1.5896,
+      "step": 357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55025924698846e-06,
+      "loss": 1.6584,
+      "step": 358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547178723755215e-06,
+      "loss": 1.692,
+      "step": 359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544088186697515e-06,
+      "loss": 1.6816,
+      "step": 360
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.54098764262136e-06,
+      "loss": 1.7146,
+      "step": 361
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.537877098354787e-06,
+      "loss": 1.583,
+      "step": 362
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.534756560747853e-06,
+      "loss": 1.6038,
+      "step": 363
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.531626036672625e-06,
+      "loss": 1.659,
+      "step": 364
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528485533023166e-06,
+      "loss": 1.5356,
+      "step": 365
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525335056715508e-06,
+      "loss": 1.6093,
+      "step": 366
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522174614687652e-06,
+      "loss": 1.656,
+      "step": 367
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519004213899539e-06,
+      "loss": 1.6288,
+      "step": 368
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515823861333048e-06,
+      "loss": 1.5543,
+      "step": 369
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512633563991968e-06,
+      "loss": 1.7403,
+      "step": 370
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509433328901993e-06,
+      "loss": 1.4493,
+      "step": 371
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506223163110699e-06,
+      "loss": 1.6584,
+      "step": 372
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503003073687535e-06,
+      "loss": 1.5836,
+      "step": 373
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499773067723797e-06,
+      "loss": 1.5647,
+      "step": 374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496533152332628e-06,
+      "loss": 1.7452,
+      "step": 375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493283334648985e-06,
+      "loss": 1.5035,
+      "step": 376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49002362182964e-06,
+      "loss": 1.5682,
+      "step": 377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.486754021053149e-06,
+      "loss": 1.6937,
+      "step": 378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48347453951985e-06,
+      "loss": 1.7118,
+      "step": 379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480185184451833e-06,
+      "loss": 1.7493,
+      "step": 380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47688596309294e-06,
+      "loss": 1.569,
+      "step": 381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473576882708735e-06,
+      "loss": 1.5273,
+      "step": 382
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.47025795058649e-06,
+      "loss": 1.6697,
+      "step": 383
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.466929174035184e-06,
+      "loss": 1.6498,
+      "step": 384
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.463590560385464e-06,
+      "loss": 1.7833,
+      "step": 385
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.46024211698965e-06,
+      "loss": 1.4737,
+      "step": 386
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4568838512217e-06,
+      "loss": 1.8554,
+      "step": 387
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453515770477208e-06,
+      "loss": 1.6775,
+      "step": 388
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450137882173385e-06,
+      "loss": 1.7146,
+      "step": 389
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446750193749033e-06,
+      "loss": 1.622,
+      "step": 390
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44335271266454e-06,
+      "loss": 1.815,
+      "step": 391
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439945446401861e-06,
+      "loss": 1.7394,
+      "step": 392
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436528402464498e-06,
+      "loss": 1.6319,
+      "step": 393
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433101588377487e-06,
+      "loss": 1.5736,
+      "step": 394
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429665011687377e-06,
+      "loss": 1.5385,
+      "step": 395
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42621867996222e-06,
+      "loss": 1.6779,
+      "step": 396
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422762600791545e-06,
+      "loss": 1.6568,
+      "step": 397
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419296781786354e-06,
+      "loss": 1.7352,
+      "step": 398
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415821230579092e-06,
+      "loss": 1.6749,
+      "step": 399
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41233595482364e-06,
+      "loss": 1.7727,
+      "step": 400
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408840962195293e-06,
+      "loss": 1.6907,
+      "step": 401
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405336260390745e-06,
+      "loss": 1.7903,
+      "step": 402
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40182185712807e-06,
+      "loss": 1.4394,
+      "step": 403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.398297760146713e-06,
+      "loss": 1.7368,
+      "step": 404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.394763977207456e-06,
+      "loss": 1.6041,
+      "step": 405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.39122051609242e-06,
+      "loss": 1.6817,
+      "step": 406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.387667384605037e-06,
+      "loss": 1.712,
+      "step": 407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.384104590570034e-06,
+      "loss": 1.7385,
+      "step": 408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380532141833416e-06,
+      "loss": 1.7865,
+      "step": 409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376950046262451e-06,
+      "loss": 1.536,
+      "step": 410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373358311745656e-06,
+      "loss": 1.6681,
+      "step": 411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369756946192764e-06,
+      "loss": 1.5602,
+      "step": 412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366145957534727e-06,
+      "loss": 1.7009,
+      "step": 413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362525353723689e-06,
+      "loss": 1.6603,
+      "step": 414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358895142732962e-06,
+      "loss": 1.6076,
+      "step": 415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355255332557018e-06,
+      "loss": 1.7784,
+      "step": 416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351605931211472e-06,
+      "loss": 1.7384,
+      "step": 417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347946946733055e-06,
+      "loss": 1.7087,
+      "step": 418
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344278387179607e-06,
+      "loss": 1.561,
+      "step": 419
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34060026063005e-06,
+      "loss": 1.6578,
+      "step": 420
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336912575184379e-06,
+      "loss": 1.6408,
+      "step": 421
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333215338963632e-06,
+      "loss": 1.7276,
+      "step": 422
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32950856010989e-06,
+      "loss": 1.6675,
+      "step": 423
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325792246786242e-06,
+      "loss": 1.5001,
+      "step": 424
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322066407176778e-06,
+      "loss": 1.7856,
+      "step": 425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.318331049486562e-06,
+      "loss": 1.6848,
+      "step": 426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.314586181941619e-06,
+      "loss": 1.7112,
+      "step": 427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.310831812788924e-06,
+      "loss": 1.6182,
+      "step": 428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.307067950296366e-06,
+      "loss": 1.6503,
+      "step": 429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.303294602752751e-06,
+      "loss": 1.6458,
+      "step": 430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299511778467766e-06,
+      "loss": 1.6922,
+      "step": 431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295719485771967e-06,
+      "loss": 1.795,
+      "step": 432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291917733016768e-06,
+      "loss": 1.7366,
+      "step": 433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288106528574407e-06,
+      "loss": 1.6582,
+      "step": 434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284285880837947e-06,
+      "loss": 1.7453,
+      "step": 435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280455798221238e-06,
+      "loss": 1.6151,
+      "step": 436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27661628915891e-06,
+      "loss": 1.6299,
+      "step": 437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272767362106356e-06,
+      "loss": 1.6394,
+      "step": 438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268909025539707e-06,
+      "loss": 1.7597,
+      "step": 439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26504128795581e-06,
+      "loss": 1.5198,
+      "step": 440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261164157872225e-06,
+      "loss": 1.6417,
+      "step": 441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257277643827191e-06,
+      "loss": 1.6835,
+      "step": 442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25338175437961e-06,
+      "loss": 1.7627,
+      "step": 443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249476498109034e-06,
+      "loss": 1.626,
+      "step": 444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245561883615643e-06,
+      "loss": 1.618,
+      "step": 445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241637919520225e-06,
+      "loss": 1.642,
+      "step": 446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237704614464157e-06,
+      "loss": 1.578,
+      "step": 447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.233761977109386e-06,
+      "loss": 1.6227,
+      "step": 448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.229810016138414e-06,
+      "loss": 1.6223,
+      "step": 449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.22584874025427e-06,
+      "loss": 1.7729,
+      "step": 450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.221878158180503e-06,
+      "loss": 1.6242,
+      "step": 451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217898278661151e-06,
+      "loss": 1.6343,
+      "step": 452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21390911046073e-06,
+      "loss": 1.681,
+      "step": 453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209910662364207e-06,
+      "loss": 1.6164,
+      "step": 454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205902943176994e-06,
+      "loss": 1.6701,
+      "step": 455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201885961724913e-06,
+      "loss": 1.6833,
+      "step": 456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19785972685418e-06,
+      "loss": 1.5536,
+      "step": 457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193824247431398e-06,
+      "loss": 1.66,
+      "step": 458
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189779532343523e-06,
+      "loss": 1.6281,
+      "step": 459
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185725590497851e-06,
+      "loss": 1.7059,
+      "step": 460
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181662430821999e-06,
+      "loss": 1.6889,
+      "step": 461
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177590062263878e-06,
+      "loss": 1.6382,
+      "step": 462
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173508493791687e-06,
+      "loss": 1.665,
+      "step": 463
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169417734393878e-06,
+      "loss": 1.7574,
+      "step": 464
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165317793079147e-06,
+      "loss": 1.7101,
+      "step": 465
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16120867887641e-06,
+      "loss": 1.6805,
+      "step": 466
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157090400834781e-06,
+      "loss": 1.6043,
+      "step": 467
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152962968023558e-06,
+      "loss": 1.5882,
+      "step": 468
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1488263895322e-06,
+      "loss": 1.7174,
+      "step": 469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.144680674470302e-06,
+      "loss": 1.6687,
+      "step": 470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.140525831967585e-06,
+      "loss": 1.6289,
+      "step": 471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.136361871173869e-06,
+      "loss": 1.6981,
+      "step": 472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.13218880125905e-06,
+      "loss": 1.4992,
+      "step": 473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128006631413093e-06,
+      "loss": 1.6618,
+      "step": 474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123815370845995e-06,
+      "loss": 1.6536,
+      "step": 475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119615028787771e-06,
+      "loss": 1.7763,
+      "step": 476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115405614488449e-06,
+      "loss": 1.687,
+      "step": 477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111187137218019e-06,
+      "loss": 1.6459,
+      "step": 478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10695960626644e-06,
+      "loss": 1.7341,
+      "step": 479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102723030943604e-06,
+      "loss": 1.5562,
+      "step": 480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098477420579326e-06,
+      "loss": 1.6262,
+      "step": 481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094222784523314e-06,
+      "loss": 1.5209,
+      "step": 482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089959132145152e-06,
+      "loss": 1.565,
+      "step": 483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085686472834281e-06,
+      "loss": 1.716,
+      "step": 484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081404815999978e-06,
+      "loss": 1.588,
+      "step": 485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077114171071333e-06,
+      "loss": 1.7594,
+      "step": 486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07281454749723e-06,
+      "loss": 1.5441,
+      "step": 487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068505954746329e-06,
+      "loss": 1.7985,
+      "step": 488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064188402307036e-06,
+      "loss": 1.5834,
+      "step": 489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05986189968749e-06,
+      "loss": 1.6734,
+      "step": 490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055526456415543e-06,
+      "loss": 1.67,
+      "step": 491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.051182082038734e-06,
+      "loss": 1.4873,
+      "step": 492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.046828786124271e-06,
+      "loss": 1.7052,
+      "step": 493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.042466578259009e-06,
+      "loss": 1.717,
+      "step": 494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.038095468049426e-06,
+      "loss": 1.5706,
+      "step": 495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033715465121608e-06,
+      "loss": 1.6833,
+      "step": 496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029326579121228e-06,
+      "loss": 1.7279,
+      "step": 497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024928819713513e-06,
+      "loss": 1.7942,
+      "step": 498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020522196583236e-06,
+      "loss": 1.5445,
+      "step": 499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016106719434689e-06,
+      "loss": 1.6861,
+      "step": 500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011682397991664e-06,
+      "loss": 1.8839,
+      "step": 501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007249241997428e-06,
+      "loss": 1.6127,
+      "step": 502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002807261214702e-06,
+      "loss": 1.7089,
+      "step": 503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998356465425644e-06,
+      "loss": 1.6719,
+      "step": 504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993896864431825e-06,
+      "loss": 1.5737,
+      "step": 505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989428468054204e-06,
+      "loss": 1.732,
+      "step": 506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98495128613311e-06,
+      "loss": 1.7786,
+      "step": 507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98046532852822e-06,
+      "loss": 1.56,
+      "step": 508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975970605118536e-06,
+      "loss": 1.6731,
+      "step": 509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971467125802369e-06,
+      "loss": 1.6773,
+      "step": 510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966954900497304e-06,
+      "loss": 1.5665,
+      "step": 511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962433939140194e-06,
+      "loss": 1.7239,
+      "step": 512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957904251687125e-06,
+      "loss": 1.642,
+      "step": 513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.953365848113404e-06,
+      "loss": 1.6096,
+      "step": 514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.948818738413529e-06,
+      "loss": 1.598,
+      "step": 515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.944262932601174e-06,
+      "loss": 1.5892,
+      "step": 516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.939698440709162e-06,
+      "loss": 1.7064,
+      "step": 517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935125272789443e-06,
+      "loss": 1.7369,
+      "step": 518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930543438913076e-06,
+      "loss": 1.5435,
+      "step": 519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925952949170202e-06,
+      "loss": 1.5836,
+      "step": 520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92135381367003e-06,
+      "loss": 1.5724,
+      "step": 521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916746042540796e-06,
+      "loss": 1.7229,
+      "step": 522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912129645929768e-06,
+      "loss": 1.5288,
+      "step": 523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9075046340032e-06,
+      "loss": 1.5647,
+      "step": 524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902871016946324e-06,
+      "loss": 1.5462,
+      "step": 525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898228804963313e-06,
+      "loss": 1.581,
+      "step": 526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893578008277281e-06,
+      "loss": 1.6043,
+      "step": 527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88891863713024e-06,
+      "loss": 1.6771,
+      "step": 528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884250701783079e-06,
+      "loss": 1.6009,
+      "step": 529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87957421251556e-06,
+      "loss": 1.5923,
+      "step": 530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874889179626273e-06,
+      "loss": 1.6965,
+      "step": 531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870195613432626e-06,
+      "loss": 1.652,
+      "step": 532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86549352427082e-06,
+      "loss": 1.6527,
+      "step": 533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860782922495821e-06,
+      "loss": 1.5745,
+      "step": 534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.856063818481349e-06,
+      "loss": 1.6206,
+      "step": 535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.851336222619842e-06,
+      "loss": 1.6837,
+      "step": 536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.846600145322438e-06,
+      "loss": 1.6556,
+      "step": 537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.841855597018956e-06,
+      "loss": 1.6988,
+      "step": 538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83710258815787e-06,
+      "loss": 1.5552,
+      "step": 539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832341129206286e-06,
+      "loss": 1.6448,
+      "step": 540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827571230649913e-06,
+      "loss": 1.591,
+      "step": 541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822792902993056e-06,
+      "loss": 1.619,
+      "step": 542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818006156758574e-06,
+      "loss": 1.4554,
+      "step": 543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813211002487869e-06,
+      "loss": 1.5839,
+      "step": 544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808407450740857e-06,
+      "loss": 1.7492,
+      "step": 545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.80359551209595e-06,
+      "loss": 1.6386,
+      "step": 546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798775197150032e-06,
+      "loss": 1.7555,
+      "step": 547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793946516518423e-06,
+      "loss": 1.6852,
+      "step": 548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789109480834876e-06,
+      "loss": 1.6909,
+      "step": 549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784264100751542e-06,
+      "loss": 1.5958,
+      "step": 550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779410386938943e-06,
+      "loss": 1.6175,
+      "step": 551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.77454835008596e-06,
+      "loss": 1.7568,
+      "step": 552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.769678000899796e-06,
+      "loss": 1.6232,
+      "step": 553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764799350105966e-06,
+      "loss": 1.7335,
+      "step": 554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759912408448266e-06,
+      "loss": 1.6768,
+      "step": 555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755017186688745e-06,
+      "loss": 1.6406,
+      "step": 556
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.750113695607692e-06,
+      "loss": 1.4326,
+      "step": 557
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.745201946003602e-06,
+      "loss": 1.5892,
+      "step": 558
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.740281948693164e-06,
+      "loss": 1.6298,
+      "step": 559
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.735353714511222e-06,
+      "loss": 1.6099,
+      "step": 560
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730417254310763e-06,
+      "loss": 1.7088,
+      "step": 561
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725472578962894e-06,
+      "loss": 1.7419,
+      "step": 562
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720519699356804e-06,
+      "loss": 1.6018,
+      "step": 563
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715558626399757e-06,
+      "loss": 1.5695,
+      "step": 564
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710589371017057e-06,
+      "loss": 1.7766,
+      "step": 565
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705611944152027e-06,
+      "loss": 1.7232,
+      "step": 566
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70062635676599e-06,
+      "loss": 1.6617,
+      "step": 567
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695632619838234e-06,
+      "loss": 1.6079,
+      "step": 568
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690630744365998e-06,
+      "loss": 1.7013,
+      "step": 569
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68562074136444e-06,
+      "loss": 1.7383,
+      "step": 570
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680602621866621e-06,
+      "loss": 1.6288,
+      "step": 571
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675576396923473e-06,
+      "loss": 1.6702,
+      "step": 572
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670542077603779e-06,
+      "loss": 1.6808,
+      "step": 573
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66549967499415e-06,
+      "loss": 1.5805,
+      "step": 574
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66044920019899e-06,
+      "loss": 1.4821,
+      "step": 575
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.655390664340491e-06,
+      "loss": 1.6351,
+      "step": 576
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650324078558587e-06,
+      "loss": 1.5956,
+      "step": 577
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645249454010946e-06,
+      "loss": 1.5821,
+      "step": 578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.640166801872938e-06,
+      "loss": 1.7436,
+      "step": 579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.635076133337606e-06,
+      "loss": 1.5841,
+      "step": 580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.629977459615655e-06,
+      "loss": 1.6936,
+      "step": 581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.624870791935413e-06,
+      "loss": 1.7167,
+      "step": 582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619756141542815e-06,
+      "loss": 1.5682,
+      "step": 583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614633519701377e-06,
+      "loss": 1.6899,
+      "step": 584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609502937692163e-06,
+      "loss": 1.7219,
+      "step": 585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604364406813775e-06,
+      "loss": 1.574,
+      "step": 586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599217938382319e-06,
+      "loss": 1.6072,
+      "step": 587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594063543731372e-06,
+      "loss": 1.547,
+      "step": 588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588901234211979e-06,
+      "loss": 1.5857,
+      "step": 589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583731021192607e-06,
+      "loss": 1.6155,
+      "step": 590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57855291605913e-06,
+      "loss": 1.6036,
+      "step": 591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573366930214807e-06,
+      "loss": 1.6594,
+      "step": 592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56817307508024e-06,
+      "loss": 1.5154,
+      "step": 593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562971362093371e-06,
+      "loss": 1.71,
+      "step": 594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557761802709446e-06,
+      "loss": 1.631,
+      "step": 595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552544408400986e-06,
+      "loss": 1.7544,
+      "step": 596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547319190657768e-06,
+      "loss": 1.6344,
+      "step": 597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5420861609868e-06,
+      "loss": 1.6524,
+      "step": 598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.53684533091229e-06,
+      "loss": 1.5741,
+      "step": 599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531596711975626e-06,
+      "loss": 1.6193,
+      "step": 600
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.526340315735349e-06,
+      "loss": 1.4681,
+      "step": 601
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.521076153767125e-06,
+      "loss": 1.4532,
+      "step": 602
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.515804237663725e-06,
+      "loss": 1.6565,
+      "step": 603
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.510524579034994e-06,
+      "loss": 1.6146,
+      "step": 604
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505237189507825e-06,
+      "loss": 1.5829,
+      "step": 605
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499942080726144e-06,
+      "loss": 1.7601,
+      "step": 606
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49463926435087e-06,
+      "loss": 1.5549,
+      "step": 607
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489328752059898e-06,
+      "loss": 1.5096,
+      "step": 608
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484010555548067e-06,
+      "loss": 1.5928,
+      "step": 609
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478684686527143e-06,
+      "loss": 1.7011,
+      "step": 610
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473351156725788e-06,
+      "loss": 1.6366,
+      "step": 611
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468009977889535e-06,
+      "loss": 1.719,
+      "step": 612
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462661161780756e-06,
+      "loss": 1.5957,
+      "step": 613
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457304720178649e-06,
+      "loss": 1.5483,
+      "step": 614
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451940664879204e-06,
+      "loss": 1.5742,
+      "step": 615
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44656900769517e-06,
+      "loss": 1.656,
+      "step": 616
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441189760456047e-06,
+      "loss": 1.5481,
+      "step": 617
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435802935008048e-06,
+      "loss": 1.5002,
+      "step": 618
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430408543214067e-06,
+      "loss": 1.5726,
+      "step": 619
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42500659695367e-06,
+      "loss": 1.582,
+      "step": 620
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419597108123054e-06,
+      "loss": 1.6418,
+      "step": 621
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414180088635029e-06,
+      "loss": 1.5659,
+      "step": 622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.408755550418987e-06,
+      "loss": 1.4733,
+      "step": 623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.40332350542088e-06,
+      "loss": 1.5616,
+      "step": 624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.397883965603189e-06,
+      "loss": 1.7877,
+      "step": 625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.392436942944903e-06,
+      "loss": 1.5858,
+      "step": 626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386982449441488e-06,
+      "loss": 1.5958,
+      "step": 627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38152049710486e-06,
+      "loss": 1.5214,
+      "step": 628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376051097963367e-06,
+      "loss": 1.6181,
+      "step": 629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37057426406175e-06,
+      "loss": 1.5886,
+      "step": 630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365090007461129e-06,
+      "loss": 1.6104,
+      "step": 631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359598340238964e-06,
+      "loss": 1.6162,
+      "step": 632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354099274489037e-06,
+      "loss": 1.5563,
+      "step": 633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348592822321427e-06,
+      "loss": 1.8521,
+      "step": 634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343078995862475e-06,
+      "loss": 1.553,
+      "step": 635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33755780725476e-06,
+      "loss": 1.6736,
+      "step": 636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.332029268657084e-06,
+      "loss": 1.558,
+      "step": 637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326493392244422e-06,
+      "loss": 1.6179,
+      "step": 638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320950190207915e-06,
+      "loss": 1.5896,
+      "step": 639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315399674754835e-06,
+      "loss": 1.6013,
+      "step": 640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309841858108567e-06,
+      "loss": 1.6421,
+      "step": 641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304276752508562e-06,
+      "loss": 1.7222,
+      "step": 642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298704370210334e-06,
+      "loss": 1.5829,
+      "step": 643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.29312472348541e-06,
+      "loss": 1.6531,
+      "step": 644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.28753782462133e-06,
+      "loss": 1.6774,
+      "step": 645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.28194368592159e-06,
+      "loss": 1.5491,
+      "step": 646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.27634231970564e-06,
+      "loss": 1.6847,
+      "step": 647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.27073373830884e-06,
+      "loss": 1.5859,
+      "step": 648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.265117954082443e-06,
+      "loss": 1.6338,
+      "step": 649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.259494979393563e-06,
+      "loss": 1.7074,
+      "step": 650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25386482662515e-06,
+      "loss": 1.5052,
+      "step": 651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248227508175957e-06,
+      "loss": 1.6599,
+      "step": 652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242583036460523e-06,
+      "loss": 1.4906,
+      "step": 653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23693142390914e-06,
+      "loss": 1.5965,
+      "step": 654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231272682967817e-06,
+      "loss": 1.5407,
+      "step": 655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225606826098273e-06,
+      "loss": 1.6375,
+      "step": 656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219933865777892e-06,
+      "loss": 1.57,
+      "step": 657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2142538144997e-06,
+      "loss": 1.6261,
+      "step": 658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20856668477234e-06,
+      "loss": 1.6587,
+      "step": 659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202872489120042e-06,
+      "loss": 1.556,
+      "step": 660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1971712400826e-06,
+      "loss": 1.6262,
+      "step": 661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19146295021534e-06,
+      "loss": 1.5161,
+      "step": 662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185747632089091e-06,
+      "loss": 1.6929,
+      "step": 663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180025298290162e-06,
+      "loss": 1.6646,
+      "step": 664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174295961420309e-06,
+      "loss": 1.6347,
+      "step": 665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.168559634096716e-06,
+      "loss": 1.6521,
+      "step": 666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.162816328951953e-06,
+      "loss": 1.4769,
+      "step": 667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.157066058633964e-06,
+      "loss": 1.4749,
+      "step": 668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.15130883580603e-06,
+      "loss": 1.5138,
+      "step": 669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14554467314674e-06,
+      "loss": 1.7624,
+      "step": 670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139773583349969e-06,
+      "loss": 1.6183,
+      "step": 671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133995579124844e-06,
+      "loss": 1.594,
+      "step": 672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128210673195722e-06,
+      "loss": 1.7036,
+      "step": 673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122418878302159e-06,
+      "loss": 1.6264,
+      "step": 674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11662020719888e-06,
+      "loss": 1.671,
+      "step": 675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110814672655752e-06,
+      "loss": 1.503,
+      "step": 676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10500228745776e-06,
+      "loss": 1.6521,
+      "step": 677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099183064404972e-06,
+      "loss": 1.5102,
+      "step": 678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093357016312518e-06,
+      "loss": 1.7219,
+      "step": 679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087524156010556e-06,
+      "loss": 1.7528,
+      "step": 680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081684496344245e-06,
+      "loss": 1.6459,
+      "step": 681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07583805017372e-06,
+      "loss": 1.6763,
+      "step": 682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069984830374059e-06,
+      "loss": 1.6334,
+      "step": 683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06412484983526e-06,
+      "loss": 1.6006,
+      "step": 684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058258121462206e-06,
+      "loss": 1.6158,
+      "step": 685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052384658174641e-06,
+      "loss": 1.6477,
+      "step": 686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046504472907142e-06,
+      "loss": 1.55,
+      "step": 687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.04061757860909e-06,
+      "loss": 1.631,
+      "step": 688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.034723988244636e-06,
+      "loss": 1.5648,
+      "step": 689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.028823714792684e-06,
+      "loss": 1.6313,
+      "step": 690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.022916771246849e-06,
+      "loss": 1.6559,
+      "step": 691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.017003170615439e-06,
+      "loss": 1.7337,
+      "step": 692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011082925921421e-06,
+      "loss": 1.6529,
+      "step": 693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005156050202395e-06,
+      "loss": 1.6298,
+      "step": 694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99922255651056e-06,
+      "loss": 1.5004,
+      "step": 695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99328245791269e-06,
+      "loss": 1.6418,
+      "step": 696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987335767490112e-06,
+      "loss": 1.5807,
+      "step": 697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981382498338658e-06,
+      "loss": 1.5926,
+      "step": 698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975422663568657e-06,
+      "loss": 1.5394,
+      "step": 699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969456276304892e-06,
+      "loss": 1.8249,
+      "step": 700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963483349686576e-06,
+      "loss": 1.644,
+      "step": 701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957503896867324e-06,
+      "loss": 1.5995,
+      "step": 702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951517931015125e-06,
+      "loss": 1.6808,
+      "step": 703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945525465312307e-06,
+      "loss": 1.6569,
+      "step": 704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939526512955516e-06,
+      "loss": 1.6681,
+      "step": 705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933521087155682e-06,
+      "loss": 1.5141,
+      "step": 706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927509201137986e-06,
+      "loss": 1.7399,
+      "step": 707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921490868141843e-06,
+      "loss": 1.646,
+      "step": 708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.915466101420861e-06,
+      "loss": 1.5792,
+      "step": 709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.909434914242818e-06,
+      "loss": 1.4834,
+      "step": 710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.903397319889632e-06,
+      "loss": 1.7505,
+      "step": 711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.897353331657327e-06,
+      "loss": 1.721,
+      "step": 712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.891302962856013e-06,
+      "loss": 1.6941,
+      "step": 713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885246226809847e-06,
+      "loss": 1.5541,
+      "step": 714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879183136857012e-06,
+      "loss": 1.5411,
+      "step": 715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873113706349677e-06,
+      "loss": 1.7476,
+      "step": 716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867037948653984e-06,
+      "loss": 1.5465,
+      "step": 717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.860955877149998e-06,
+      "loss": 1.5736,
+      "step": 718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854867505231698e-06,
+      "loss": 1.7104,
+      "step": 719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848772846306932e-06,
+      "loss": 1.615,
+      "step": 720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842671913797395e-06,
+      "loss": 1.4162,
+      "step": 721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836564721138595e-06,
+      "loss": 1.5918,
+      "step": 722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83045128177983e-06,
+      "loss": 1.611,
+      "step": 723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.824331609184155e-06,
+      "loss": 1.6522,
+      "step": 724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.818205716828346e-06,
+      "loss": 1.5473,
+      "step": 725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812073618202884e-06,
+      "loss": 1.658,
+      "step": 726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805935326811913e-06,
+      "loss": 1.5168,
+      "step": 727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799790856173212e-06,
+      "loss": 1.6736,
+      "step": 728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793640219818178e-06,
+      "loss": 1.5163,
+      "step": 729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787483431291772e-06,
+      "loss": 1.7131,
+      "step": 730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781320504152515e-06,
+      "loss": 1.4596,
+      "step": 731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.775151451972443e-06,
+      "loss": 1.5873,
+      "step": 732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.768976288337077e-06,
+      "loss": 1.6976,
+      "step": 733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.762795026845402e-06,
+      "loss": 1.6817,
+      "step": 734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.756607681109829e-06,
+      "loss": 1.6778,
+      "step": 735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750414264756168e-06,
+      "loss": 1.5947,
+      "step": 736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744214791423597e-06,
+      "loss": 1.5568,
+      "step": 737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738009274764636e-06,
+      "loss": 1.6819,
+      "step": 738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731797728445107e-06,
+      "loss": 1.5921,
+      "step": 739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72558016614412e-06,
+      "loss": 1.6644,
+      "step": 740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719356601554025e-06,
+      "loss": 1.5194,
+      "step": 741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713127048380394e-06,
+      "loss": 1.5576,
+      "step": 742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706891520341987e-06,
+      "loss": 1.698,
+      "step": 743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700650031170722e-06,
+      "loss": 1.5852,
+      "step": 744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694402594611643e-06,
+      "loss": 1.6992,
+      "step": 745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.688149224422895e-06,
+      "loss": 1.3974,
+      "step": 746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681889934375683e-06,
+      "loss": 1.5729,
+      "step": 747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675624738254258e-06,
+      "loss": 1.5006,
+      "step": 748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669353649855869e-06,
+      "loss": 1.6353,
+      "step": 749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663076682990748e-06,
+      "loss": 1.4386,
+      "step": 750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656793851482068e-06,
+      "loss": 1.5174,
+      "step": 751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.650505169165918e-06,
+      "loss": 1.7085,
+      "step": 752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644210649891273e-06,
+      "loss": 1.6091,
+      "step": 753
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.637910307519961e-06,
+      "loss": 1.5525,
+      "step": 754
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.631604155926635e-06,
+      "loss": 1.6297,
+      "step": 755
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6252922089987384e-06,
+      "loss": 1.6804,
+      "step": 756
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.618974480636478e-06,
+      "loss": 1.6918,
+      "step": 757
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612650984752797e-06,
+      "loss": 1.4885,
+      "step": 758
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606321735273334e-06,
+      "loss": 1.5668,
+      "step": 759
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599986746136401e-06,
+      "loss": 1.4729,
+      "step": 760
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593646031292948e-06,
+      "loss": 1.6528,
+      "step": 761
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587299604706537e-06,
+      "loss": 1.6222,
+      "step": 762
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580947480353307e-06,
+      "loss": 1.5229,
+      "step": 763
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574589672221944e-06,
+      "loss": 1.6441,
+      "step": 764
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56822619431365e-06,
+      "loss": 1.5664,
+      "step": 765
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.56185706064212e-06,
+      "loss": 1.7114,
+      "step": 766
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5554822852334905e-06,
+      "loss": 1.6031,
+      "step": 767
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549101882126338e-06,
+      "loss": 1.619,
+      "step": 768
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54271586537162e-06,
+      "loss": 1.6155,
+      "step": 769
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536324249032663e-06,
+      "loss": 1.7181,
+      "step": 770
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529927047185122e-06,
+      "loss": 1.6278,
+      "step": 771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523524273916954e-06,
+      "loss": 1.6941,
+      "step": 772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5171159433283865e-06,
+      "loss": 1.6017,
+      "step": 773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510702069531881e-06,
+      "loss": 1.5602,
+      "step": 774
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.504282666652113e-06,
+      "loss": 1.578,
+      "step": 775
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.497857748825926e-06,
+      "loss": 1.6078,
+      "step": 776
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.491427330202317e-06,
+      "loss": 1.5506,
+      "step": 777
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.484991424942388e-06,
+      "loss": 1.494,
+      "step": 778
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.478550047219333e-06,
+      "loss": 1.5295,
+      "step": 779
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472103211218391e-06,
+      "loss": 1.5964,
+      "step": 780
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465650931136823e-06,
+      "loss": 1.5598,
+      "step": 781
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459193221183878e-06,
+      "loss": 1.6539,
+      "step": 782
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452730095580767e-06,
+      "loss": 1.6104,
+      "step": 783
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446261568560622e-06,
+      "loss": 1.5815,
+      "step": 784
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439787654368472e-06,
+      "loss": 1.5755,
+      "step": 785
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433308367261211e-06,
+      "loss": 1.652,
+      "step": 786
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426823721507565e-06,
+      "loss": 1.4838,
+      "step": 787
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4203337313880575e-06,
+      "loss": 1.7026,
+      "step": 788
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413838411194986e-06,
+      "loss": 1.5793,
+      "step": 789
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407337775232384e-06,
+      "loss": 1.6387,
+      "step": 790
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40083183781599e-06,
+      "loss": 1.587,
+      "step": 791
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394320613273218e-06,
+      "loss": 1.5098,
+      "step": 792
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387804115943129e-06,
+      "loss": 1.6223,
+      "step": 793
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.381282360176389e-06,
+      "loss": 1.5844,
+      "step": 794
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374755360335253e-06,
+      "loss": 1.5363,
+      "step": 795
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368223130793515e-06,
+      "loss": 1.5939,
+      "step": 796
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.361685685936491e-06,
+      "loss": 1.6806,
+      "step": 797
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.355143040160983e-06,
+      "loss": 1.5438,
+      "step": 798
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.348595207875245e-06,
+      "loss": 1.5235,
+      "step": 799
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.342042203498952e-06,
+      "loss": 1.6416,
+      "step": 800
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33548404146317e-06,
+      "loss": 1.6552,
+      "step": 801
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3289207362103234e-06,
+      "loss": 1.7453,
+      "step": 802
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322352302194162e-06,
+      "loss": 1.6113,
+      "step": 803
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315778753879732e-06,
+      "loss": 1.4639,
+      "step": 804
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309200105743339e-06,
+      "loss": 1.628,
+      "step": 805
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302616372272524e-06,
+      "loss": 1.3998,
+      "step": 806
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296027567966024e-06,
+      "loss": 1.5745,
+      "step": 807
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289433707333744e-06,
+      "loss": 1.5189,
+      "step": 808
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282834804896721e-06,
+      "loss": 1.7066,
+      "step": 809
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276230875187102e-06,
+      "loss": 1.5413,
+      "step": 810
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269621932748099e-06,
+      "loss": 1.6314,
+      "step": 811
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263007992133963e-06,
+      "loss": 1.6685,
+      "step": 812
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256389067909958e-06,
+      "loss": 1.5881,
+      "step": 813
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249765174652316e-06,
+      "loss": 1.6432,
+      "step": 814
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2431363269482144e-06,
+      "loss": 1.5849,
+      "step": 815
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236502539395743e-06,
+      "loss": 1.6415,
+      "step": 816
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.22986382660387e-06,
+      "loss": 1.6085,
+      "step": 817
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2232202031924055e-06,
+      "loss": 1.6047,
+      "step": 818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.21657168379198e-06,
+      "loss": 1.6048,
+      "step": 819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.209918283044001e-06,
+      "loss": 1.6601,
+      "step": 820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.2032600156006315e-06,
+      "loss": 1.5682,
+      "step": 821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.196596896124745e-06,
+      "loss": 1.5878,
+      "step": 822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189928939289902e-06,
+      "loss": 1.6858,
+      "step": 823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183256159780321e-06,
+      "loss": 1.6023,
+      "step": 824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1765785722908345e-06,
+      "loss": 1.7581,
+      "step": 825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169896191526868e-06,
+      "loss": 1.583,
+      "step": 826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1632090322044e-06,
+      "loss": 1.5182,
+      "step": 827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156517109049934e-06,
+      "loss": 1.6377,
+      "step": 828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149820436800462e-06,
+      "loss": 1.6173,
+      "step": 829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143119030203438e-06,
+      "loss": 1.6668,
+      "step": 830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136412904016738e-06,
+      "loss": 1.7843,
+      "step": 831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129702073008635e-06,
+      "loss": 1.5893,
+      "step": 832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1229865519577624e-06,
+      "loss": 1.666,
+      "step": 833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.116266355653078e-06,
+      "loss": 1.6611,
+      "step": 834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1095414988938415e-06,
+      "loss": 1.5289,
+      "step": 835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102811996489572e-06,
+      "loss": 1.4887,
+      "step": 836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096077863260022e-06,
+      "loss": 1.5631,
+      "step": 837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089339114035138e-06,
+      "loss": 1.5487,
+      "step": 838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.082595763655036e-06,
+      "loss": 1.4981,
+      "step": 839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07584782696996e-06,
+      "loss": 1.62,
+      "step": 840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.069095318840261e-06,
+      "loss": 1.4979,
+      "step": 841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.062338254136351e-06,
+      "loss": 1.6832,
+      "step": 842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.05557664773868e-06,
+      "loss": 1.5738,
+      "step": 843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.048810514537697e-06,
+      "loss": 1.5481,
+      "step": 844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042039869433821e-06,
+      "loss": 1.6221,
+      "step": 845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0352647273374095e-06,
+      "loss": 1.7252,
+      "step": 846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028485103168719e-06,
+      "loss": 1.4357,
+      "step": 847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021701011857882e-06,
+      "loss": 1.7237,
+      "step": 848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014912468344863e-06,
+      "loss": 1.6294,
+      "step": 849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008119487579435e-06,
+      "loss": 1.7142,
+      "step": 850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001322084521138e-06,
+      "loss": 1.6856,
+      "step": 851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994520274139258e-06,
+      "loss": 1.6075,
+      "step": 852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987714071412781e-06,
+      "loss": 1.6701,
+      "step": 853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.980903491330368e-06,
+      "loss": 1.7054,
+      "step": 854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974088548890319e-06,
+      "loss": 1.5991,
+      "step": 855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967269259100542e-06,
+      "loss": 1.5652,
+      "step": 856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96044563697852e-06,
+      "loss": 1.6667,
+      "step": 857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953617697551272e-06,
+      "loss": 1.7271,
+      "step": 858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946785455855328e-06,
+      "loss": 1.5776,
+      "step": 859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939948926936692e-06,
+      "loss": 1.5512,
+      "step": 860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933108125850813e-06,
+      "loss": 1.7373,
+      "step": 861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92626306766254e-06,
+      "loss": 1.6763,
+      "step": 862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.919413767446102e-06,
+      "loss": 1.617,
+      "step": 863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.912560240285069e-06,
+      "loss": 1.5208,
+      "step": 864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.905702501272322e-06,
+      "loss": 1.5545,
+      "step": 865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.898840565510014e-06,
+      "loss": 1.6305,
+      "step": 866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.891974448109542e-06,
+      "loss": 1.6399,
+      "step": 867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885104164191511e-06,
+      "loss": 1.5394,
+      "step": 868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878229728885702e-06,
+      "loss": 1.7775,
+      "step": 869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871351157331038e-06,
+      "loss": 1.5825,
+      "step": 870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864468464675549e-06,
+      "loss": 1.4853,
+      "step": 871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8575816660763435e-06,
+      "loss": 1.6843,
+      "step": 872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.850690776699574e-06,
+      "loss": 1.6592,
+      "step": 873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843795811720393e-06,
+      "loss": 1.4754,
+      "step": 874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836896786322939e-06,
+      "loss": 1.6712,
+      "step": 875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829993715700283e-06,
+      "loss": 1.6473,
+      "step": 876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823086615054413e-06,
+      "loss": 1.6145,
+      "step": 877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816175499596185e-06,
+      "loss": 1.5537,
+      "step": 878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.809260384545302e-06,
+      "loss": 1.6639,
+      "step": 879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802341285130269e-06,
+      "loss": 1.5547,
+      "step": 880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.79541821658837e-06,
+      "loss": 1.558,
+      "step": 881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788491194165629e-06,
+      "loss": 1.598,
+      "step": 882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781560233116775e-06,
+      "loss": 1.6716,
+      "step": 883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774625348705214e-06,
+      "loss": 1.5223,
+      "step": 884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7676865562029884e-06,
+      "loss": 1.6661,
+      "step": 885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7607438708907505e-06,
+      "loss": 1.5622,
+      "step": 886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.753797308057722e-06,
+      "loss": 1.6268,
+      "step": 887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.74684688300167e-06,
+      "loss": 1.4495,
+      "step": 888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739892611028858e-06,
+      "loss": 1.6227,
+      "step": 889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732934507454026e-06,
+      "loss": 1.6473,
+      "step": 890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725972587600356e-06,
+      "loss": 1.5197,
+      "step": 891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7190068667994234e-06,
+      "loss": 1.5806,
+      "step": 892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7120373603911855e-06,
+      "loss": 1.54,
+      "step": 893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705064083723931e-06,
+      "loss": 1.661,
+      "step": 894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698087052154251e-06,
+      "loss": 1.6289,
+      "step": 895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.691106281047004e-06,
+      "loss": 1.6554,
+      "step": 896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684121785775292e-06,
+      "loss": 1.5565,
+      "step": 897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677133581720407e-06,
+      "loss": 1.6105,
+      "step": 898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6701416842718195e-06,
+      "loss": 1.6183,
+      "step": 899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663146108827123e-06,
+      "loss": 1.6889,
+      "step": 900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6561468707920195e-06,
+      "loss": 1.541,
+      "step": 901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649143985580273e-06,
+      "loss": 1.8512,
+      "step": 902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642137468613677e-06,
+      "loss": 1.6098,
+      "step": 903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635127335322029e-06,
+      "loss": 1.4744,
+      "step": 904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628113601143085e-06,
+      "loss": 1.6305,
+      "step": 905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.621096281522533e-06,
+      "loss": 1.595,
+      "step": 906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.614075391913956e-06,
+      "loss": 1.5033,
+      "step": 907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.607050947778802e-06,
+      "loss": 1.5533,
+      "step": 908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.60002296458634e-06,
+      "loss": 1.5973,
+      "step": 909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.592991457813642e-06,
+      "loss": 1.6429,
+      "step": 910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585956442945531e-06,
+      "loss": 1.6545,
+      "step": 911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578917935474561e-06,
+      "loss": 1.6653,
+      "step": 912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571875950900975e-06,
+      "loss": 1.5608,
+      "step": 913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564830504732674e-06,
+      "loss": 1.5864,
+      "step": 914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557781612485184e-06,
+      "loss": 1.6089,
+      "step": 915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550729289681613e-06,
+      "loss": 1.6087,
+      "step": 916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.543673551852635e-06,
+      "loss": 1.5885,
+      "step": 917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536614414536434e-06,
+      "loss": 1.5763,
+      "step": 918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529551893278688e-06,
+      "loss": 1.5715,
+      "step": 919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522486003632522e-06,
+      "loss": 1.7381,
+      "step": 920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515416761158482e-06,
+      "loss": 1.6205,
+      "step": 921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.508344181424495e-06,
+      "loss": 1.5679,
+      "step": 922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5012682800058395e-06,
+      "loss": 1.6963,
+      "step": 923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494189072485109e-06,
+      "loss": 1.566,
+      "step": 924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4871065744521754e-06,
+      "loss": 1.6951,
+      "step": 925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480020801504162e-06,
+      "loss": 1.5399,
+      "step": 926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472931769245397e-06,
+      "loss": 1.6544,
+      "step": 927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.465839493287393e-06,
+      "loss": 1.4787,
+      "step": 928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4587439892488014e-06,
+      "loss": 1.639,
+      "step": 929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.451645272755385e-06,
+      "loss": 1.5058,
+      "step": 930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.444543359439979e-06,
+      "loss": 1.7305,
+      "step": 931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.437438264942458e-06,
+      "loss": 1.5463,
+      "step": 932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430330004909709e-06,
+      "loss": 1.6689,
+      "step": 933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4232185949955814e-06,
+      "loss": 1.6011,
+      "step": 934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.416104050860867e-06,
+      "loss": 1.5276,
+      "step": 935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4089863881732565e-06,
+      "loss": 1.5291,
+      "step": 936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.4018656226073104e-06,
+      "loss": 1.644,
+      "step": 937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394741769844422e-06,
+      "loss": 1.5717,
+      "step": 938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3876148455727815e-06,
+      "loss": 1.4927,
+      "step": 939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380484865487346e-06,
+      "loss": 1.7474,
+      "step": 940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373351845289799e-06,
+      "loss": 1.6303,
+      "step": 941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366215800688523e-06,
+      "loss": 1.6439,
+      "step": 942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359076747398557e-06,
+      "loss": 1.6528,
+      "step": 943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351934701141568e-06,
+      "loss": 1.7122,
+      "step": 944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.344789677645813e-06,
+      "loss": 1.4457,
+      "step": 945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337641692646106e-06,
+      "loss": 1.5765,
+      "step": 946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330490761883781e-06,
+      "loss": 1.4618,
+      "step": 947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323336901106665e-06,
+      "loss": 1.7737,
+      "step": 948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.316180126069029e-06,
+      "loss": 1.7201,
+      "step": 949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.30902045253157e-06,
+      "loss": 1.6752,
+      "step": 950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.301857896261363e-06,
+      "loss": 1.5481,
+      "step": 951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.29469247303183e-06,
+      "loss": 1.5519,
+      "step": 952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.287524198622714e-06,
+      "loss": 1.6138,
+      "step": 953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280353088820028e-06,
+      "loss": 1.5473,
+      "step": 954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273179159416036e-06,
+      "loss": 1.5295,
+      "step": 955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266002426209208e-06,
+      "loss": 1.5363,
+      "step": 956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258822905004189e-06,
+      "loss": 1.6089,
+      "step": 957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251640611611762e-06,
+      "loss": 1.5842,
+      "step": 958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.244455561848822e-06,
+      "loss": 1.5436,
+      "step": 959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237267771538324e-06,
+      "loss": 1.578,
+      "step": 960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230077256509266e-06,
+      "loss": 1.6249,
+      "step": 961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222884032596643e-06,
+      "loss": 1.5409,
+      "step": 962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215688115641415e-06,
+      "loss": 1.6556,
+      "step": 963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208489521490475e-06,
+      "loss": 1.4964,
+      "step": 964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20128826599661e-06,
+      "loss": 1.5812,
+      "step": 965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194084365018469e-06,
+      "loss": 1.4189,
+      "step": 966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186877834420526e-06,
+      "loss": 1.5241,
+      "step": 967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1796686900730465e-06,
+      "loss": 1.6197,
+      "step": 968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1724569478520495e-06,
+      "loss": 1.6169,
+      "step": 969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16524262363928e-06,
+      "loss": 1.6881,
+      "step": 970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158025733322166e-06,
+      "loss": 1.4889,
+      "step": 971
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.150806292793786e-06,
+      "loss": 1.651,
+      "step": 972
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.143584317952836e-06,
+      "loss": 1.5519,
+      "step": 973
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.136359824703592e-06,
+      "loss": 1.5176,
+      "step": 974
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.12913282895588e-06,
+      "loss": 1.5528,
+      "step": 975
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121903346625026e-06,
+      "loss": 1.5466,
+      "step": 976
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.114671393631849e-06,
+      "loss": 1.5963,
+      "step": 977
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107436985902593e-06,
+      "loss": 1.5948,
+      "step": 978
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.100200139368917e-06,
+      "loss": 1.6177,
+      "step": 979
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092960869967848e-06,
+      "loss": 1.5725,
+      "step": 980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.085719193641748e-06,
+      "loss": 1.6242,
+      "step": 981
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078475126338282e-06,
+      "loss": 1.631,
+      "step": 982
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.071228684010376e-06,
+      "loss": 1.6016,
+      "step": 983
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.063979882616193e-06,
+      "loss": 1.5446,
+      "step": 984
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.056728738119083e-06,
+      "loss": 1.4638,
+      "step": 985
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.049475266487563e-06,
+      "loss": 1.4953,
+      "step": 986
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.042219483695271e-06,
+      "loss": 1.5124,
+      "step": 987
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.034961405720936e-06,
+      "loss": 1.5862,
+      "step": 988
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.02770104854834e-06,
+      "loss": 1.5436,
+      "step": 989
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020438428166289e-06,
+      "loss": 1.6162,
+      "step": 990
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013173560568566e-06,
+      "loss": 1.5972,
+      "step": 991
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005906461753905e-06,
+      "loss": 1.647,
+      "step": 992
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998637147725961e-06,
+      "loss": 1.615,
+      "step": 993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.991365634493255e-06,
+      "loss": 1.6654,
+      "step": 994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.984091938069161e-06,
+      "loss": 1.5433,
+      "step": 995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.976816074471854e-06,
+      "loss": 1.6125,
+      "step": 996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.969538059724289e-06,
+      "loss": 1.557,
+      "step": 997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.96225790985415e-06,
+      "loss": 1.4421,
+      "step": 998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.954975640893831e-06,
+      "loss": 1.6645,
+      "step": 999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947691268880385e-06,
+      "loss": 1.5404,
+      "step": 1000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.940404809855502e-06,
+      "loss": 1.6766,
+      "step": 1001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933116279865467e-06,
+      "loss": 1.5355,
+      "step": 1002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925825694961122e-06,
+      "loss": 1.5389,
+      "step": 1003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918533071197839e-06,
+      "loss": 1.5039,
+      "step": 1004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911238424635477e-06,
+      "loss": 1.6934,
+      "step": 1005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903941771338353e-06,
+      "loss": 1.545,
+      "step": 1006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896643127375198e-06,
+      "loss": 1.5827,
+      "step": 1007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.889342508819133e-06,
+      "loss": 1.5401,
+      "step": 1008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.88203993174762e-06,
+      "loss": 1.54,
+      "step": 1009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874735412242443e-06,
+      "loss": 1.6248,
+      "step": 1010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867428966389656e-06,
+      "loss": 1.5178,
+      "step": 1011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860120610279558e-06,
+      "loss": 1.6287,
+      "step": 1012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.852810360006659e-06,
+      "loss": 1.5879,
+      "step": 1013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.845498231669631e-06,
+      "loss": 1.6127,
+      "step": 1014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8381842413712895e-06,
+      "loss": 1.6398,
+      "step": 1015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.830868405218548e-06,
+      "loss": 1.4827,
+      "step": 1016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.823550739322386e-06,
+      "loss": 1.6721,
+      "step": 1017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.816231259797811e-06,
+      "loss": 1.5146,
+      "step": 1018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.808909982763825e-06,
+      "loss": 1.7043,
+      "step": 1019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801586924343386e-06,
+      "loss": 1.522,
+      "step": 1020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7942621006633825e-06,
+      "loss": 1.7304,
+      "step": 1021
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.786935527854583e-06,
+      "loss": 1.6446,
+      "step": 1022
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.779607222051608e-06,
+      "loss": 1.5514,
+      "step": 1023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7722771993929e-06,
+      "loss": 1.5813,
+      "step": 1024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.764945476020678e-06,
+      "loss": 1.5994,
+      "step": 1025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.757612068080908e-06,
+      "loss": 1.565,
+      "step": 1026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7502769917232635e-06,
+      "loss": 1.6245,
+      "step": 1027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742940263101097e-06,
+      "loss": 1.4125,
+      "step": 1028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.735601898371395e-06,
+      "loss": 1.6074,
+      "step": 1029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.728261913694748e-06,
+      "loss": 1.5579,
+      "step": 1030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720920325235314e-06,
+      "loss": 1.7263,
+      "step": 1031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.713577149160787e-06,
+      "loss": 1.5221,
+      "step": 1032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.706232401642351e-06,
+      "loss": 1.638,
+      "step": 1033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.698886098854653e-06,
+      "loss": 1.496,
+      "step": 1034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.691538256975766e-06,
+      "loss": 1.5109,
+      "step": 1035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684188892187152e-06,
+      "loss": 1.7035,
+      "step": 1036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6768380206736265e-06,
+      "loss": 1.482,
+      "step": 1037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.669485658623321e-06,
+      "loss": 1.5144,
+      "step": 1038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.662131822227657e-06,
+      "loss": 1.5783,
+      "step": 1039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.654776527681293e-06,
+      "loss": 1.6447,
+      "step": 1040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6474197911821034e-06,
+      "loss": 1.6029,
+      "step": 1041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6400616289311405e-06,
+      "loss": 1.628,
+      "step": 1042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.63270205713259e-06,
+      "loss": 1.5622,
+      "step": 1043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625341091993748e-06,
+      "loss": 1.6601,
+      "step": 1044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617978749724974e-06,
+      "loss": 1.6552,
+      "step": 1045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.610615046539666e-06,
+      "loss": 1.5563,
+      "step": 1046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.603249998654212e-06,
+      "loss": 1.601,
+      "step": 1047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.595883622287968e-06,
+      "loss": 1.6218,
+      "step": 1048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58851593366321e-06,
+      "loss": 1.5442,
+      "step": 1049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.58114694900511e-06,
+      "loss": 1.7197,
+      "step": 1050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573776684541685e-06,
+      "loss": 1.6611,
+      "step": 1051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5664051565037816e-06,
+      "loss": 1.5994,
+      "step": 1052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5590323811250205e-06,
+      "loss": 1.5378,
+      "step": 1053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551658374641772e-06,
+      "loss": 1.676,
+      "step": 1054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544283153293119e-06,
+      "loss": 1.5769,
+      "step": 1055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.536906733320816e-06,
+      "loss": 1.6762,
+      "step": 1056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529529130969264e-06,
+      "loss": 1.617,
+      "step": 1057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522150362485458e-06,
+      "loss": 1.7366,
+      "step": 1058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.514770444118971e-06,
+      "loss": 1.6595,
+      "step": 1059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.507389392121901e-06,
+      "loss": 1.5711,
+      "step": 1060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.500007222748849e-06,
+      "loss": 1.4909,
+      "step": 1061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.492623952256868e-06,
+      "loss": 1.5481,
+      "step": 1062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.485239596905445e-06,
+      "loss": 1.6811,
+      "step": 1063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.477854172956451e-06,
+      "loss": 1.6689,
+      "step": 1064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.470467696674108e-06,
+      "loss": 1.4845,
+      "step": 1065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463080184324964e-06,
+      "loss": 1.4903,
+      "step": 1066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455691652177838e-06,
+      "loss": 1.6221,
+      "step": 1067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448302116503804e-06,
+      "loss": 1.6451,
+      "step": 1068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440911593576141e-06,
+      "loss": 1.5301,
+      "step": 1069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.433520099670303e-06,
+      "loss": 1.4626,
+      "step": 1070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.426127651063882e-06,
+      "loss": 1.5772,
+      "step": 1071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418734264036575e-06,
+      "loss": 1.4351,
+      "step": 1072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411339954870141e-06,
+      "loss": 1.5539,
+      "step": 1073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.403944739848373e-06,
+      "loss": 1.7214,
+      "step": 1074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3965486352570595e-06,
+      "loss": 1.5472,
+      "step": 1075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.389151657383945e-06,
+      "loss": 1.5822,
+      "step": 1076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381753822518701e-06,
+      "loss": 1.4593,
+      "step": 1077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374355146952882e-06,
+      "loss": 1.5723,
+      "step": 1078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3669556469798975e-06,
+      "loss": 1.6024,
+      "step": 1079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35955533889497e-06,
+      "loss": 1.5234,
+      "step": 1080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.352154238995103e-06,
+      "loss": 1.5594,
+      "step": 1081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.344752363579043e-06,
+      "loss": 1.6114,
+      "step": 1082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.337349728947247e-06,
+      "loss": 1.5965,
+      "step": 1083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.329946351401839e-06,
+      "loss": 1.5268,
+      "step": 1084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.322542247246583e-06,
+      "loss": 1.6868,
+      "step": 1085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315137432786841e-06,
+      "loss": 1.5559,
+      "step": 1086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.30773192432954e-06,
+      "loss": 1.5749,
+      "step": 1087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3003257381831386e-06,
+      "loss": 1.6373,
+      "step": 1088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.29291889065758e-06,
+      "loss": 1.5021,
+      "step": 1089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2855113980642725e-06,
+      "loss": 1.5751,
+      "step": 1090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2781032767160375e-06,
+      "loss": 1.62,
+      "step": 1091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270694542927089e-06,
+      "loss": 1.6346,
+      "step": 1092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.263285213012982e-06,
+      "loss": 1.5032,
+      "step": 1093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255875303290586e-06,
+      "loss": 1.557,
+      "step": 1094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248464830078055e-06,
+      "loss": 1.6096,
+      "step": 1095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.241053809694771e-06,
+      "loss": 1.6856,
+      "step": 1096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.233642258461333e-06,
+      "loss": 1.6196,
+      "step": 1097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2262301926995004e-06,
+      "loss": 1.5344,
+      "step": 1098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2188176287321735e-06,
+      "loss": 1.6176,
+      "step": 1099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.211404582883341e-06,
+      "loss": 1.596,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.203991071478059e-06,
+      "loss": 1.605,
+      "step": 1101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.196577110842405e-06,
+      "loss": 1.6126,
+      "step": 1102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.189162717303453e-06,
+      "loss": 1.5931,
+      "step": 1103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.181747907189223e-06,
+      "loss": 1.524,
+      "step": 1104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.174332696828651e-06,
+      "loss": 1.6279,
+      "step": 1105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.166917102551563e-06,
+      "loss": 1.5499,
+      "step": 1106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159501140688621e-06,
+      "loss": 1.6764,
+      "step": 1107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152084827571305e-06,
+      "loss": 1.4991,
+      "step": 1108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144668179531863e-06,
+      "loss": 1.6032,
+      "step": 1109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137251212903282e-06,
+      "loss": 1.5704,
+      "step": 1110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12983394401925e-06,
+      "loss": 1.6656,
+      "step": 1111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122416389214124e-06,
+      "loss": 1.5421,
+      "step": 1112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114998564822884e-06,
+      "loss": 1.6037,
+      "step": 1113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107580487181112e-06,
+      "loss": 1.5512,
+      "step": 1114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.10016217262494e-06,
+      "loss": 1.4875,
+      "step": 1115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092743637491027e-06,
+      "loss": 1.4801,
+      "step": 1116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085324898116515e-06,
+      "loss": 1.5084,
+      "step": 1117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077905970838996e-06,
+      "loss": 1.5493,
+      "step": 1118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.070486871996479e-06,
+      "loss": 1.6255,
+      "step": 1119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.063067617927344e-06,
+      "loss": 1.6285,
+      "step": 1120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.055648224970322e-06,
+      "loss": 1.5346,
+      "step": 1121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048228709464443e-06,
+      "loss": 1.6907,
+      "step": 1122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.040809087749006e-06,
+      "loss": 1.649,
+      "step": 1123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033389376163553e-06,
+      "loss": 1.6119,
+      "step": 1124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.025969591047813e-06,
+      "loss": 1.6193,
+      "step": 1125
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.0185497487416815e-06,
+      "loss": 1.5636,
+      "step": 1126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.011129865585182e-06,
+      "loss": 1.5372,
+      "step": 1127
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.003709957918426e-06,
+      "loss": 1.633,
+      "step": 1128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.996290042081576e-06,
+      "loss": 1.4675,
+      "step": 1129
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98887013441482e-06,
+      "loss": 1.6265,
+      "step": 1130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.981450251258319e-06,
+      "loss": 1.6209,
+      "step": 1131
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97403040895219e-06,
+      "loss": 1.4959,
+      "step": 1132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9666106238364485e-06,
+      "loss": 1.5399,
+      "step": 1133
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9591909122509944e-06,
+      "loss": 1.4537,
+      "step": 1134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.95177129053556e-06,
+      "loss": 1.6434,
+      "step": 1135
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94435177502968e-06,
+      "loss": 1.639,
+      "step": 1136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936932382072657e-06,
+      "loss": 1.6698,
+      "step": 1137
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929513128003524e-06,
+      "loss": 1.4816,
+      "step": 1138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.922094029161006e-06,
+      "loss": 1.6226,
+      "step": 1139
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9146751018834875e-06,
+      "loss": 1.5844,
+      "step": 1140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907256362508976e-06,
+      "loss": 1.6686,
+      "step": 1141
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.899837827375061e-06,
+      "loss": 1.5781,
+      "step": 1142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89241951281889e-06,
+      "loss": 1.7336,
+      "step": 1143
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885001435177117e-06,
+      "loss": 1.6326,
+      "step": 1144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877583610785877e-06,
+      "loss": 1.705,
+      "step": 1145
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87016605598075e-06,
+      "loss": 1.6206,
+      "step": 1146
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8627487870967186e-06,
+      "loss": 1.6259,
+      "step": 1147
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.855331820468138e-06,
+      "loss": 1.6391,
+      "step": 1148
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.847915172428695e-06,
+      "loss": 1.6019,
+      "step": 1149
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.84049885931138e-06,
+      "loss": 1.5234,
+      "step": 1150
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.833082897448439e-06,
+      "loss": 1.7208,
+      "step": 1151
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.82566730317135e-06,
+      "loss": 1.5721,
+      "step": 1152
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.818252092810779e-06,
+      "loss": 1.6494,
+      "step": 1153
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8108372826965475e-06,
+      "loss": 1.6075,
+      "step": 1154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8034228891575955e-06,
+      "loss": 1.7053,
+      "step": 1155
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.796008928521943e-06,
+      "loss": 1.6065,
+      "step": 1156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.788595417116662e-06,
+      "loss": 1.5508,
+      "step": 1157
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781182371267828e-06,
+      "loss": 1.4651,
+      "step": 1158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7737698073005e-06,
+      "loss": 1.6699,
+      "step": 1159
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.766357741538668e-06,
+      "loss": 1.6174,
+      "step": 1160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.75894619030523e-06,
+      "loss": 1.535,
+      "step": 1161
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751535169921947e-06,
+      "loss": 1.6957,
+      "step": 1162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7441246967094145e-06,
+      "loss": 1.5607,
+      "step": 1163
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7441246967094145e-06,
+      "loss": 1.4492,
+      "step": 1164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736714786987021e-06,
+      "loss": 1.5655,
+      "step": 1165
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.729305457072913e-06,
+      "loss": 1.5214,
+      "step": 1166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.721896723283963e-06,
+      "loss": 1.636,
+      "step": 1167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.714488601935729e-06,
+      "loss": 1.6357,
+      "step": 1168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.707081109342422e-06,
+      "loss": 1.5228,
+      "step": 1169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.699674261816863e-06,
+      "loss": 1.6031,
+      "step": 1170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.692268075670461e-06,
+      "loss": 1.4295,
+      "step": 1171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.68486256721316e-06,
+      "loss": 1.603,
+      "step": 1172
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6774577527534195e-06,
+      "loss": 1.5868,
+      "step": 1173
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670053648598164e-06,
+      "loss": 1.5614,
+      "step": 1174
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662650271052755e-06,
+      "loss": 1.5193,
+      "step": 1175
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.655247636420958e-06,
+      "loss": 1.4782,
+      "step": 1176
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6478457610048985e-06,
+      "loss": 1.465,
+      "step": 1177
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640444661105032e-06,
+      "loss": 1.5954,
+      "step": 1178
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.633044353020104e-06,
+      "loss": 1.534,
+      "step": 1179
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.625644853047119e-06,
+      "loss": 1.6097,
+      "step": 1180
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6182461774813004e-06,
+      "loss": 1.5027,
+      "step": 1181
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610848342616056e-06,
+      "loss": 1.5023,
+      "step": 1182
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603451364742942e-06,
+      "loss": 1.486,
+      "step": 1183
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596055260151628e-06,
+      "loss": 1.5667,
+      "step": 1184
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.58866004512986e-06,
+      "loss": 1.5886,
+      "step": 1185
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5812657359634275e-06,
+      "loss": 1.5466,
+      "step": 1186
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57387234893612e-06,
+      "loss": 1.6424,
+      "step": 1187
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566479900329698e-06,
+      "loss": 1.6089,
+      "step": 1188
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.559088406423861e-06,
+      "loss": 1.5484,
+      "step": 1189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.551697883496197e-06,
+      "loss": 1.5782,
+      "step": 1190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.544308347822163e-06,
+      "loss": 1.6084,
+      "step": 1191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.536919815675038e-06,
+      "loss": 1.6902,
+      "step": 1192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.529532303325893e-06,
+      "loss": 1.6896,
+      "step": 1193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.522145827043551e-06,
+      "loss": 1.6159,
+      "step": 1194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514760403094556e-06,
+      "loss": 1.5718,
+      "step": 1195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507376047743134e-06,
+      "loss": 1.5888,
+      "step": 1196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4999927772511525e-06,
+      "loss": 1.5158,
+      "step": 1197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4926106078781004e-06,
+      "loss": 1.7069,
+      "step": 1198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485229555881031e-06,
+      "loss": 1.597,
+      "step": 1199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477849637514544e-06,
+      "loss": 1.6912,
+      "step": 1200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.470470869030739e-06,
+      "loss": 1.6711,
+      "step": 1201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463093266679185e-06,
+      "loss": 1.5967,
+      "step": 1202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455716846706883e-06,
+      "loss": 1.5855,
+      "step": 1203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4483416253582305e-06,
+      "loss": 1.6684,
+      "step": 1204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.440967618874982e-06,
+      "loss": 1.5145,
+      "step": 1205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.43359484349622e-06,
+      "loss": 1.6024,
+      "step": 1206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.426223315458316e-06,
+      "loss": 1.6511,
+      "step": 1207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418853050994893e-06,
+      "loss": 1.4937,
+      "step": 1208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411484066336792e-06,
+      "loss": 1.5134,
+      "step": 1209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.404116377712033e-06,
+      "loss": 1.4929,
+      "step": 1210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3967500013457895e-06,
+      "loss": 1.6958,
+      "step": 1211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.389384953460336e-06,
+      "loss": 1.4627,
+      "step": 1212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.382021250275028e-06,
+      "loss": 1.5785,
+      "step": 1213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.374658908006255e-06,
+      "loss": 1.5096,
+      "step": 1214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.367297942867413e-06,
+      "loss": 1.6834,
+      "step": 1215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.359938371068861e-06,
+      "loss": 1.5377,
+      "step": 1216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.352580208817897e-06,
+      "loss": 1.534,
+      "step": 1217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34522347231871e-06,
+      "loss": 1.6534,
+      "step": 1218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.337868177772344e-06,
+      "loss": 1.5288,
+      "step": 1219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3305143413766795e-06,
+      "loss": 1.6005,
+      "step": 1220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323161979326376e-06,
+      "loss": 1.4993,
+      "step": 1221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315811107812851e-06,
+      "loss": 1.6035,
+      "step": 1222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308461743024237e-06,
+      "loss": 1.5974,
+      "step": 1223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30111390114535e-06,
+      "loss": 1.4704,
+      "step": 1224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.293767598357651e-06,
+      "loss": 1.5824,
+      "step": 1225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.286422850839215e-06,
+      "loss": 1.4817,
+      "step": 1226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.279079674764687e-06,
+      "loss": 1.6432,
+      "step": 1227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271738086305254e-06,
+      "loss": 1.6123,
+      "step": 1228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264398101628608e-06,
+      "loss": 1.549,
+      "step": 1229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257059736898904e-06,
+      "loss": 1.5449,
+      "step": 1230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.249723008276737e-06,
+      "loss": 1.5511,
+      "step": 1231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.242387931919095e-06,
+      "loss": 1.5506,
+      "step": 1232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.235054523979324e-06,
+      "loss": 1.6526,
+      "step": 1233
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.2277228006071014e-06,
+      "loss": 1.5457,
+      "step": 1234
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.220392777948393e-06,
+      "loss": 1.7607,
+      "step": 1235
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.21306447214542e-06,
+      "loss": 1.6653,
+      "step": 1236
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.205737899336618e-06,
+      "loss": 1.5653,
+      "step": 1237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1984130756566144e-06,
+      "loss": 1.5694,
+      "step": 1238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191090017236177e-06,
+      "loss": 1.5949,
+      "step": 1239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183768740202191e-06,
+      "loss": 1.6563,
+      "step": 1240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.176449260677614e-06,
+      "loss": 1.5411,
+      "step": 1241
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169131594781453e-06,
+      "loss": 1.5573,
+      "step": 1242
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.161815758628711e-06,
+      "loss": 1.5975,
+      "step": 1243
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.154501768330371e-06,
+      "loss": 1.5793,
+      "step": 1244
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.147189639993344e-06,
+      "loss": 1.4482,
+      "step": 1245
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139879389720443e-06,
+      "loss": 1.4582,
+      "step": 1246
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.132571033610346e-06,
+      "loss": 1.6079,
+      "step": 1247
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125264587757559e-06,
+      "loss": 1.6041,
+      "step": 1248
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.117960068252381e-06,
+      "loss": 1.6004,
+      "step": 1249
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.11065749118087e-06,
+      "loss": 1.4798,
+      "step": 1250
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1033568726248035e-06,
+      "loss": 1.5615,
+      "step": 1251
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096058228661648e-06,
+      "loss": 1.6215,
+      "step": 1252
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088761575364524e-06,
+      "loss": 1.5702,
+      "step": 1253
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081466928802163e-06,
+      "loss": 1.4768,
+      "step": 1254
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.07417430503888e-06,
+      "loss": 1.5395,
+      "step": 1255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0668837201345345e-06,
+      "loss": 1.5744,
+      "step": 1256
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.059595190144499e-06,
+      "loss": 1.4333,
+      "step": 1257
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.052308731119616e-06,
+      "loss": 1.4744,
+      "step": 1258
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.045024359106171e-06,
+      "loss": 1.5689,
+      "step": 1259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.037742090145851e-06,
+      "loss": 1.5015,
+      "step": 1260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030461940275713e-06,
+      "loss": 1.5443,
+      "step": 1261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.023183925528148e-06,
+      "loss": 1.5473,
+      "step": 1262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.015908061930842e-06,
+      "loss": 1.5484,
+      "step": 1263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008634365506747e-06,
+      "loss": 1.6442,
+      "step": 1264
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.001362852274041e-06,
+      "loss": 1.5823,
+      "step": 1265
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994093538246096e-06,
+      "loss": 1.6052,
+      "step": 1266
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9868264394314375e-06,
+      "loss": 1.5545,
+      "step": 1267
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979561571833713e-06,
+      "loss": 1.6153,
+      "step": 1268
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.972298951451661e-06,
+      "loss": 1.5628,
+      "step": 1269
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.965038594279065e-06,
+      "loss": 1.5607,
+      "step": 1270
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957780516304731e-06,
+      "loss": 1.4985,
+      "step": 1271
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950524733512439e-06,
+      "loss": 1.5747,
+      "step": 1272
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.943271261880919e-06,
+      "loss": 1.6765,
+      "step": 1273
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9360201173838085e-06,
+      "loss": 1.6151,
+      "step": 1274
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.928771315989625e-06,
+      "loss": 1.6119,
+      "step": 1275
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.92152487366172e-06,
+      "loss": 1.4454,
+      "step": 1276
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9142808063582526e-06,
+      "loss": 1.5783,
+      "step": 1277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.907039130032154e-06,
+      "loss": 1.5816,
+      "step": 1278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.899799860631084e-06,
+      "loss": 1.4853,
+      "step": 1279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8925630140974085e-06,
+      "loss": 1.5664,
+      "step": 1280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.885328606368154e-06,
+      "loss": 1.5897,
+      "step": 1281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8780966533749745e-06,
+      "loss": 1.4509,
+      "step": 1282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870867171044123e-06,
+      "loss": 1.5946,
+      "step": 1283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8636401752964085e-06,
+      "loss": 1.6261,
+      "step": 1284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856415682047166e-06,
+      "loss": 1.559,
+      "step": 1285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.849193707206215e-06,
+      "loss": 1.6076,
+      "step": 1286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.841974266677835e-06,
+      "loss": 1.6022,
+      "step": 1287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834757376360721e-06,
+      "loss": 1.5214,
+      "step": 1288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.827543052147952e-06,
+      "loss": 1.5274,
+      "step": 1289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820331309926956e-06,
+      "loss": 1.627,
+      "step": 1290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8131221655794758e-06,
+      "loss": 1.528,
+      "step": 1291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8059156349815316e-06,
+      "loss": 1.689,
+      "step": 1292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7987117340033913e-06,
+      "loss": 1.4686,
+      "step": 1293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7915104785095273e-06,
+      "loss": 1.4688,
+      "step": 1294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784311884358587e-06,
+      "loss": 1.5487,
+      "step": 1295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7771159674033586e-06,
+      "loss": 1.4988,
+      "step": 1296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7699227434907354e-06,
+      "loss": 1.695,
+      "step": 1297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7627322284616774e-06,
+      "loss": 1.5893,
+      "step": 1298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7555444381511796e-06,
+      "loss": 1.6281,
+      "step": 1299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7483593883882384e-06,
+      "loss": 1.5368,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7411770949958126e-06,
+      "loss": 1.6136,
+      "step": 1301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.733997573790794e-06,
+      "loss": 1.637,
+      "step": 1302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7268208405839664e-06,
+      "loss": 1.6947,
+      "step": 1303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7196469111799733e-06,
+      "loss": 1.5064,
+      "step": 1304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7124758013772878e-06,
+      "loss": 1.3707,
+      "step": 1305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705307526968171e-06,
+      "loss": 1.6263,
+      "step": 1306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.69814210373864e-06,
+      "loss": 1.6694,
+      "step": 1307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6909795474684307e-06,
+      "loss": 1.6591,
+      "step": 1308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6838198739309716e-06,
+      "loss": 1.5707,
+      "step": 1309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6766630988933365e-06,
+      "loss": 1.5367,
+      "step": 1310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.66950923811622e-06,
+      "loss": 1.5777,
+      "step": 1311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662358307353897e-06,
+      "loss": 1.48,
+      "step": 1312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655210322354189e-06,
+      "loss": 1.573,
+      "step": 1313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648065298858433e-06,
+      "loss": 1.5256,
+      "step": 1314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640923252601445e-06,
+      "loss": 1.5391,
+      "step": 1315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6337841993114792e-06,
+      "loss": 1.5922,
+      "step": 1316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.626648154710202e-06,
+      "loss": 1.5538,
+      "step": 1317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6195151345126556e-06,
+      "loss": 1.5212,
+      "step": 1318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.61238515442722e-06,
+      "loss": 1.4543,
+      "step": 1319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6052582301555804e-06,
+      "loss": 1.6303,
+      "step": 1320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.598134377392692e-06,
+      "loss": 1.5366,
+      "step": 1321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.591013611826745e-06,
+      "loss": 1.4879,
+      "step": 1322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5838959491391346e-06,
+      "loss": 1.8109,
+      "step": 1323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.57678140500442e-06,
+      "loss": 1.6597,
+      "step": 1324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.569669995090293e-06,
+      "loss": 1.5543,
+      "step": 1325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5625617350575425e-06,
+      "loss": 1.5634,
+      "step": 1326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5554566405600234e-06,
+      "loss": 1.6406,
+      "step": 1327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.548354727244617e-06,
+      "loss": 1.5853,
+      "step": 1328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5412560107512007e-06,
+      "loss": 1.4869,
+      "step": 1329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534160506712609e-06,
+      "loss": 1.6278,
+      "step": 1330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.527068230754604e-06,
+      "loss": 1.5946,
+      "step": 1331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5199791984958394e-06,
+      "loss": 1.4605,
+      "step": 1332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5128934255478254e-06,
+      "loss": 1.6528,
+      "step": 1333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5058109275148934e-06,
+      "loss": 1.5753,
+      "step": 1334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.498731719994162e-06,
+      "loss": 1.6278,
+      "step": 1335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4916558185755068e-06,
+      "loss": 1.5712,
+      "step": 1336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.48458323884152e-06,
+      "loss": 1.6626,
+      "step": 1337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47751399636748e-06,
+      "loss": 1.4239,
+      "step": 1338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4704481067213134e-06,
+      "loss": 1.5683,
+      "step": 1339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4633855854635674e-06,
+      "loss": 1.5651,
+      "step": 1340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.456326448147367e-06,
+      "loss": 1.6219,
+      "step": 1341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4492707103183883e-06,
+      "loss": 1.5884,
+      "step": 1342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4422183875148196e-06,
+      "loss": 1.5907,
+      "step": 1343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4351694952673276e-06,
+      "loss": 1.4419,
+      "step": 1344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.428124049099026e-06,
+      "loss": 1.5263,
+      "step": 1345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4210820645254407e-06,
+      "loss": 1.4845,
+      "step": 1346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4140435570544708e-06,
+      "loss": 1.5263,
+      "step": 1347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4070085421863595e-06,
+      "loss": 1.6339,
+      "step": 1348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3999770354136607e-06,
+      "loss": 1.6103,
+      "step": 1349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3929490522212e-06,
+      "loss": 1.533,
+      "step": 1350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3859246080860454e-06,
+      "loss": 1.4609,
+      "step": 1351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.378903718477469e-06,
+      "loss": 1.5675,
+      "step": 1352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.371886398856917e-06,
+      "loss": 1.4245,
+      "step": 1353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3648726646779716e-06,
+      "loss": 1.5487,
+      "step": 1354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357862531386324e-06,
+      "loss": 1.6305,
+      "step": 1355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.35085601441973e-06,
+      "loss": 1.6458,
+      "step": 1356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3438531292079817e-06,
+      "loss": 1.4481,
+      "step": 1357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3368538911728776e-06,
+      "loss": 1.4821,
+      "step": 1358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3298583157281825e-06,
+      "loss": 1.5319,
+      "step": 1359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322866418279594e-06,
+      "loss": 1.4405,
+      "step": 1360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3158782142247105e-06,
+      "loss": 1.5655,
+      "step": 1361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.308893718952997e-06,
+      "loss": 1.5705,
+      "step": 1362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.301912947845751e-06,
+      "loss": 1.5653,
+      "step": 1363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2949359162760707e-06,
+      "loss": 1.4983,
+      "step": 1364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.287962639608816e-06,
+      "loss": 1.6631,
+      "step": 1365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.280993133200578e-06,
+      "loss": 1.6336,
+      "step": 1366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2740274123996465e-06,
+      "loss": 1.4896,
+      "step": 1367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2670654925459748e-06,
+      "loss": 1.4616,
+      "step": 1368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2601073889711445e-06,
+      "loss": 1.6309,
+      "step": 1369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2531531169983327e-06,
+      "loss": 1.4302,
+      "step": 1370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2462026919422785e-06,
+      "loss": 1.5758,
+      "step": 1371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.239256129109251e-06,
+      "loss": 1.5222,
+      "step": 1372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2323134437970137e-06,
+      "loss": 1.7368,
+      "step": 1373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.225374651294789e-06,
+      "loss": 1.5559,
+      "step": 1374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2184397668832267e-06,
+      "loss": 1.5562,
+      "step": 1375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2184397668832267e-06,
+      "loss": 1.6041,
+      "step": 1376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2115088058343725e-06,
+      "loss": 1.5606,
+      "step": 1377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.204581783411631e-06,
+      "loss": 1.5947,
+      "step": 1378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197658714869733e-06,
+      "loss": 1.6726,
+      "step": 1379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1907396154547004e-06,
+      "loss": 1.5833,
+      "step": 1380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183824500403816e-06,
+      "loss": 1.5534,
+      "step": 1381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.176913384945588e-06,
+      "loss": 1.6301,
+      "step": 1382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170006284299718e-06,
+      "loss": 1.5269,
+      "step": 1383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.163103213677064e-06,
+      "loss": 1.7108,
+      "step": 1384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1562041882796084e-06,
+      "loss": 1.4823,
+      "step": 1385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.149309223300428e-06,
+      "loss": 1.5819,
+      "step": 1386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.142418333923657e-06,
+      "loss": 1.7736,
+      "step": 1387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.135531535324453e-06,
+      "loss": 1.6477,
+      "step": 1388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.128648842668964e-06,
+      "loss": 1.4452,
+      "step": 1389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1217702711142984e-06,
+      "loss": 1.436,
+      "step": 1390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.11489583580849e-06,
+      "loss": 1.4696,
+      "step": 1391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.108025551890459e-06,
+      "loss": 1.5309,
+      "step": 1392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1011594344899875e-06,
+      "loss": 1.5897,
+      "step": 1393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09429749872768e-06,
+      "loss": 1.6589,
+      "step": 1394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.087439759714932e-06,
+      "loss": 1.5582,
+      "step": 1395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0805862325539005e-06,
+      "loss": 1.6431,
+      "step": 1396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0737369323374635e-06,
+      "loss": 1.5703,
+      "step": 1397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066891874149189e-06,
+      "loss": 1.6122,
+      "step": 1398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0600510730633073e-06,
+      "loss": 1.3909,
+      "step": 1399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0532145441446725e-06,
+      "loss": 1.3788,
+      "step": 1400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.04638230244873e-06,
+      "loss": 1.5168,
+      "step": 1401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0395543630214823e-06,
+      "loss": 1.5864,
+      "step": 1402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0327307408994587e-06,
+      "loss": 1.5479,
+      "step": 1403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.025911451109682e-06,
+      "loss": 1.5687,
+      "step": 1404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0190965086696342e-06,
+      "loss": 1.5793,
+      "step": 1405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0122859285872214e-06,
+      "loss": 1.5281,
+      "step": 1406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0054797258607438e-06,
+      "loss": 1.5894,
+      "step": 1407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9986779154788627e-06,
+      "loss": 1.6785,
+      "step": 1408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9918805124205672e-06,
+      "loss": 1.5697,
+      "step": 1409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9850875316551387e-06,
+      "loss": 1.5803,
+      "step": 1410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.97829898814212e-06,
+      "loss": 1.4929,
+      "step": 1411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.971514896831281e-06,
+      "loss": 1.4259,
+      "step": 1412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9647352726625917e-06,
+      "loss": 1.5168,
+      "step": 1413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.95796013056618e-06,
+      "loss": 1.6275,
+      "step": 1414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9511894854623056e-06,
+      "loss": 1.4759,
+      "step": 1415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.944423352261322e-06,
+      "loss": 1.6466,
+      "step": 1416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.937661745863649e-06,
+      "loss": 1.481,
+      "step": 1417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9309046811597395e-06,
+      "loss": 1.4927,
+      "step": 1418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9241521730300416e-06,
+      "loss": 1.7615,
+      "step": 1419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9174042363449685e-06,
+      "loss": 1.5624,
+      "step": 1420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.910660885964863e-06,
+      "loss": 1.5077,
+      "step": 1421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9039221367399808e-06,
+      "loss": 1.5624,
+      "step": 1422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8971880035104294e-06,
+      "loss": 1.5007,
+      "step": 1423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8904585011061614e-06,
+      "loss": 1.6802,
+      "step": 1424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8837336443469223e-06,
+      "loss": 1.6115,
+      "step": 1425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8770134480422405e-06,
+      "loss": 1.6191,
+      "step": 1426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8702979269913663e-06,
+      "loss": 1.7364,
+      "step": 1427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8635870959832645e-06,
+      "loss": 1.5036,
+      "step": 1428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.856880969796564e-06,
+      "loss": 1.5157,
+      "step": 1429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.850179563199539e-06,
+      "loss": 1.6746,
+      "step": 1430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.843482890950067e-06,
+      "loss": 1.6267,
+      "step": 1431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.836790967795602e-06,
+      "loss": 1.5889,
+      "step": 1432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8301038084731336e-06,
+      "loss": 1.5901,
+      "step": 1433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8234214277091663e-06,
+      "loss": 1.5957,
+      "step": 1434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.816743840219681e-06,
+      "loss": 1.5461,
+      "step": 1435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.810071060710101e-06,
+      "loss": 1.6353,
+      "step": 1436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8034031038752587e-06,
+      "loss": 1.5593,
+      "step": 1437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.796739984399372e-06,
+      "loss": 1.585,
+      "step": 1438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7900817169559977e-06,
+      "loss": 1.5276,
+      "step": 1439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7834283162080217e-06,
+      "loss": 1.5738,
+      "step": 1440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7767797968075953e-06,
+      "loss": 1.5367,
+      "step": 1441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7701361733961334e-06,
+      "loss": 1.6353,
+      "step": 1442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7634974606042568e-06,
+      "loss": 1.6079,
+      "step": 1443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7568636730517872e-06,
+      "loss": 1.4739,
+      "step": 1444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7502348253476862e-06,
+      "loss": 1.6258,
+      "step": 1445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.743610932090045e-06,
+      "loss": 1.3888,
+      "step": 1446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7369920078660367e-06,
+      "loss": 1.4607,
+      "step": 1447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7303780672519033e-06,
+      "loss": 1.5994,
+      "step": 1448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.723769124812899e-06,
+      "loss": 1.588,
+      "step": 1449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7171651951032805e-06,
+      "loss": 1.5365,
+      "step": 1450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7105662926662592e-06,
+      "loss": 1.5185,
+      "step": 1451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.7039724320339777e-06,
+      "loss": 1.5526,
+      "step": 1452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6973836277274766e-06,
+      "loss": 1.5919,
+      "step": 1453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6907998942566637e-06,
+      "loss": 1.5918,
+      "step": 1454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.684221246120271e-06,
+      "loss": 1.457,
+      "step": 1455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6776476978058397e-06,
+      "loss": 1.4268,
+      "step": 1456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.671079263789678e-06,
+      "loss": 1.545,
+      "step": 1457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6645159585368323e-06,
+      "loss": 1.5295,
+      "step": 1458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.65795779650105e-06,
+      "loss": 1.4476,
+      "step": 1459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6514047921247584e-06,
+      "loss": 1.6824,
+      "step": 1460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.644856959839017e-06,
+      "loss": 1.6368,
+      "step": 1461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.638314314063511e-06,
+      "loss": 1.5219,
+      "step": 1462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.631776869206487e-06,
+      "loss": 1.5155,
+      "step": 1463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6252446396647503e-06,
+      "loss": 1.5092,
+      "step": 1464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.61871763982361e-06,
+      "loss": 1.5754,
+      "step": 1465
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6121958840568732e-06,
+      "loss": 1.6143,
+      "step": 1466
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6056793867267834e-06,
+      "loss": 1.6852,
+      "step": 1467
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5991681621840136e-06,
+      "loss": 1.5402,
+      "step": 1468
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5926622247676193e-06,
+      "loss": 1.5414,
+      "step": 1469
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5861615888050158e-06,
+      "loss": 1.5501,
+      "step": 1470
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5796662686119433e-06,
+      "loss": 1.5605,
+      "step": 1471
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.573176278492438e-06,
+      "loss": 1.467,
+      "step": 1472
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5666916327387907e-06,
+      "loss": 1.4805,
+      "step": 1473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.56021234563153e-06,
+      "loss": 1.5741,
+      "step": 1474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5537384314393793e-06,
+      "loss": 1.6251,
+      "step": 1475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5472699044192354e-06,
+      "loss": 1.5898,
+      "step": 1476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5408067788161236e-06,
+      "loss": 1.622,
+      "step": 1477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5343490688631805e-06,
+      "loss": 1.4468,
+      "step": 1478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5278967887816096e-06,
+      "loss": 1.4921,
+      "step": 1479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5214499527806692e-06,
+      "loss": 1.5693,
+      "step": 1480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5150085750576126e-06,
+      "loss": 1.5572,
+      "step": 1481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5085726697976864e-06,
+      "loss": 1.6297,
+      "step": 1482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.502142251174074e-06,
+      "loss": 1.5306,
+      "step": 1483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.495717333347889e-06,
+      "loss": 1.6136,
+      "step": 1484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.48929793046812e-06,
+      "loss": 1.4997,
+      "step": 1485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.482884056671617e-06,
+      "loss": 1.4229,
+      "step": 1486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.476475726083048e-06,
+      "loss": 1.6277,
+      "step": 1487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4700729528148804e-06,
+      "loss": 1.5681,
+      "step": 1488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4636757509673393e-06,
+      "loss": 1.4954,
+      "step": 1489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4572841346283833e-06,
+      "loss": 1.5606,
+      "step": 1490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4508981178736645e-06,
+      "loss": 1.5382,
+      "step": 1491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.44451771476651e-06,
+      "loss": 1.5148,
+      "step": 1492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.438142939357882e-06,
+      "loss": 1.6235,
+      "step": 1493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.431773805686351e-06,
+      "loss": 1.5707,
+      "step": 1494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425410327778058e-06,
+      "loss": 1.5493,
+      "step": 1495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4190525196466945e-06,
+      "loss": 1.4647,
+      "step": 1496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4127003952934635e-06,
+      "loss": 1.6035,
+      "step": 1497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4063539687070542e-06,
+      "loss": 1.4484,
+      "step": 1498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.4000132538636013e-06,
+      "loss": 1.5443,
+      "step": 1499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3936782647266687e-06,
+      "loss": 1.569,
+      "step": 1500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3873490152472036e-06,
+      "loss": 1.5508,
+      "step": 1501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3810255193635233e-06,
+      "loss": 1.6494,
+      "step": 1502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.374707791001264e-06,
+      "loss": 1.5837,
+      "step": 1503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3683958440733685e-06,
+      "loss": 1.6233,
+      "step": 1504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.362089692480039e-06,
+      "loss": 1.6208,
+      "step": 1505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3557893501087286e-06,
+      "loss": 1.6347,
+      "step": 1506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3494948308340825e-06,
+      "loss": 1.4928,
+      "step": 1507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3432061485179345e-06,
+      "loss": 1.6192,
+      "step": 1508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3369233170092536e-06,
+      "loss": 1.5875,
+      "step": 1509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3306463501441324e-06,
+      "loss": 1.4887,
+      "step": 1510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324375261745744e-06,
+      "loss": 1.5408,
+      "step": 1511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3181100656243195e-06,
+      "loss": 1.6001,
+      "step": 1512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3118507755771082e-06,
+      "loss": 1.5459,
+      "step": 1513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3055974053883585e-06,
+      "loss": 1.624,
+      "step": 1514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2993499688292793e-06,
+      "loss": 1.557,
+      "step": 1515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293108479658015e-06,
+      "loss": 1.6587,
+      "step": 1516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2868729516196075e-06,
+      "loss": 1.5191,
+      "step": 1517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.280643398445978e-06,
+      "loss": 1.6246,
+      "step": 1518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.274419833855881e-06,
+      "loss": 1.5929,
+      "step": 1519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.268202271554894e-06,
+      "loss": 1.6434,
+      "step": 1520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.261990725235366e-06,
+      "loss": 1.5345,
+      "step": 1521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2557852085764053e-06,
+      "loss": 1.5293,
+      "step": 1522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.249585735243833e-06,
+      "loss": 1.5882,
+      "step": 1523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243392318890173e-06,
+      "loss": 1.4733,
+      "step": 1524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2372049731545993e-06,
+      "loss": 1.634,
+      "step": 1525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2310237116629253e-06,
+      "loss": 1.5402,
+      "step": 1526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2248485480275593e-06,
+      "loss": 1.4247,
+      "step": 1527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.218679495847486e-06,
+      "loss": 1.5597,
+      "step": 1528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2125165687082288e-06,
+      "loss": 1.495,
+      "step": 1529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2063597801818253e-06,
+      "loss": 1.5841,
+      "step": 1530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.200209143826789e-06,
+      "loss": 1.544,
+      "step": 1531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1940646731880887e-06,
+      "loss": 1.5987,
+      "step": 1532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.187926381797116e-06,
+      "loss": 1.5224,
+      "step": 1533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1817942831716556e-06,
+      "loss": 1.6472,
+      "step": 1534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.175668390815848e-06,
+      "loss": 1.5659,
+      "step": 1535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169548718220173e-06,
+      "loss": 1.442,
+      "step": 1536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1634352788614064e-06,
+      "loss": 1.3766,
+      "step": 1537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157328086202608e-06,
+      "loss": 1.4673,
+      "step": 1538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.151227153693069e-06,
+      "loss": 1.5096,
+      "step": 1539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1451324947683037e-06,
+      "loss": 1.5248,
+      "step": 1540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1390441228500013e-06,
+      "loss": 1.5412,
+      "step": 1541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1329620513460183e-06,
+      "loss": 1.6072,
+      "step": 1542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.126886293650323e-06,
+      "loss": 1.4913,
+      "step": 1543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1208168631429914e-06,
+      "loss": 1.4256,
+      "step": 1544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.114753773190153e-06,
+      "loss": 1.5598,
+      "step": 1545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1086970371439884e-06,
+      "loss": 1.5094,
+      "step": 1546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.102646668342674e-06,
+      "loss": 1.5552,
+      "step": 1547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0966026801103705e-06,
+      "loss": 1.5971,
+      "step": 1548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0905650857571837e-06,
+      "loss": 1.5031,
+      "step": 1549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08453389857914e-06,
+      "loss": 1.6484,
+      "step": 1550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0785091318581577e-06,
+      "loss": 1.4563,
+      "step": 1551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0724907988620156e-06,
+      "loss": 1.4946,
+      "step": 1552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.06647891284432e-06,
+      "loss": 1.6781,
+      "step": 1553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.060473487044484e-06,
+      "loss": 1.4882,
+      "step": 1554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0544745346876926e-06,
+      "loss": 1.6493,
+      "step": 1555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0484820689848767e-06,
+      "loss": 1.7769,
+      "step": 1556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0424961031326766e-06,
+      "loss": 1.6807,
+      "step": 1557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.036516650313427e-06,
+      "loss": 1.6248,
+      "step": 1558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0305437236951085e-06,
+      "loss": 1.5345,
+      "step": 1559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0245773364313443e-06,
+      "loss": 1.6243,
+      "step": 1560
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0186175016613425e-06,
+      "loss": 1.4839,
+      "step": 1561
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0126642325098906e-06,
+      "loss": 1.4873,
+      "step": 1562
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0067175420873093e-06,
+      "loss": 1.5422,
+      "step": 1563
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.0007774434894427e-06,
+      "loss": 1.4841,
+      "step": 1564
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.994843949797607e-06,
+      "loss": 1.4194,
+      "step": 1565
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9889170740785805e-06,
+      "loss": 1.5597,
+      "step": 1566
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9829968293845624e-06,
+      "loss": 1.5183,
+      "step": 1567
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9770832287531523e-06,
+      "loss": 1.5586,
+      "step": 1568
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.971176285207317e-06,
+      "loss": 1.4739,
+      "step": 1569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9652760117553653e-06,
+      "loss": 1.5435,
+      "step": 1570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9593824213909124e-06,
+      "loss": 1.5946,
+      "step": 1571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.953495527092859e-06,
+      "loss": 1.6315,
+      "step": 1572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.94761534182536e-06,
+      "loss": 1.5773,
+      "step": 1573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9417418785377965e-06,
+      "loss": 1.49,
+      "step": 1574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.935875150164742e-06,
+      "loss": 1.5962,
+      "step": 1575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9300151696259434e-06,
+      "loss": 1.6485,
+      "step": 1576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9241619498262807e-06,
+      "loss": 1.6923,
+      "step": 1577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.918315503655757e-06,
+      "loss": 1.5203,
+      "step": 1578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912475843989446e-06,
+      "loss": 1.6468,
+      "step": 1579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9066429836874844e-06,
+      "loss": 1.5616,
+      "step": 1580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9008169355950284e-06,
+      "loss": 1.5352,
+      "step": 1581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8949977125422426e-06,
+      "loss": 1.5748,
+      "step": 1582
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8891853273442495e-06,
+      "loss": 1.6784,
+      "step": 1583
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8833797928011227e-06,
+      "loss": 1.6108,
+      "step": 1584
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.877581121697843e-06,
+      "loss": 1.5936,
+      "step": 1585
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8717893268042786e-06,
+      "loss": 1.4762,
+      "step": 1586
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8660044208751571e-06,
+      "loss": 1.5441,
+      "step": 1587
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.860226416650034e-06,
+      "loss": 1.5014,
+      "step": 1588
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8544553268532617e-06,
+      "loss": 1.5415,
+      "step": 1589
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8486911641939715e-06,
+      "loss": 1.5809,
+      "step": 1590
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8429339413660363e-06,
+      "loss": 1.5087,
+      "step": 1591
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.837183671048049e-06,
+      "loss": 1.5285,
+      "step": 1592
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8314403659032865e-06,
+      "loss": 1.6423,
+      "step": 1593
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8257040385796937e-06,
+      "loss": 1.482,
+      "step": 1594
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.819974701709839e-06,
+      "loss": 1.553,
+      "step": 1595
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.814252367910911e-06,
+      "loss": 1.5581,
+      "step": 1596
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.808537049784661e-06,
+      "loss": 1.5009,
+      "step": 1597
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802828759917402e-06,
+      "loss": 1.5656,
+      "step": 1598
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7971275108799586e-06,
+      "loss": 1.6372,
+      "step": 1599
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7914333152276631e-06,
+      "loss": 1.5714,
+      "step": 1600
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7857461855003017e-06,
+      "loss": 1.5399,
+      "step": 1601
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.78006613422211e-06,
+      "loss": 1.6065,
+      "step": 1602
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7743931739017257e-06,
+      "loss": 1.6599,
+      "step": 1603
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687273170321834e-06,
+      "loss": 1.6044,
+      "step": 1604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630685760908623e-06,
+      "loss": 1.5818,
+      "step": 1605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7574169635394784e-06,
+      "loss": 1.5363,
+      "step": 1606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7517724918240446e-06,
+      "loss": 1.5155,
+      "step": 1607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7461351733748522e-06,
+      "loss": 1.4983,
+      "step": 1608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7405050206064372e-06,
+      "loss": 1.6948,
+      "step": 1609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7348820459175587e-06,
+      "loss": 1.611,
+      "step": 1610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7292662616911615e-06,
+      "loss": 1.5107,
+      "step": 1611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7236576802943616e-06,
+      "loss": 1.568,
+      "step": 1612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71805631407841e-06,
+      "loss": 1.4606,
+      "step": 1613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7124621753786725e-06,
+      "loss": 1.3939,
+      "step": 1614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.70687527651459e-06,
+      "loss": 1.573,
+      "step": 1615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7012956297896698e-06,
+      "loss": 1.4859,
+      "step": 1616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6957232474914376e-06,
+      "loss": 1.5204,
+      "step": 1617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6901581418914342e-06,
+      "loss": 1.5027,
+      "step": 1618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6846003252451643e-06,
+      "loss": 1.542,
+      "step": 1619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6790498097920881e-06,
+      "loss": 1.5242,
+      "step": 1620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6735066077555795e-06,
+      "loss": 1.7478,
+      "step": 1621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6679707313429183e-06,
+      "loss": 1.5906,
+      "step": 1622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6624421927452394e-06,
+      "loss": 1.5586,
+      "step": 1623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6569210041375278e-06,
+      "loss": 1.4955,
+      "step": 1624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6514071776785745e-06,
+      "loss": 1.5255,
+      "step": 1625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6459007255109637e-06,
+      "loss": 1.5456,
+      "step": 1626
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.640401659761038e-06,
+      "loss": 1.5027,
+      "step": 1627
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.634909992538874e-06,
+      "loss": 1.4755,
+      "step": 1628
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6294257359382509e-06,
+      "loss": 1.7004,
+      "step": 1629
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6239489020366346e-06,
+      "loss": 1.6087,
+      "step": 1630
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6184795028951406e-06,
+      "loss": 1.6013,
+      "step": 1631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6130175505585149e-06,
+      "loss": 1.5463,
+      "step": 1632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6075630570550982e-06,
+      "loss": 1.6127,
+      "step": 1633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6021160343968134e-06,
+      "loss": 1.6547,
+      "step": 1634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5966764945791208e-06,
+      "loss": 1.4376,
+      "step": 1635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5912444495810147e-06,
+      "loss": 1.4767,
+      "step": 1636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.585819911364973e-06,
+      "loss": 1.3541,
+      "step": 1637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5804028918769488e-06,
+      "loss": 1.6713,
+      "step": 1638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.574993403046331e-06,
+      "loss": 1.554,
+      "step": 1639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5695914567859345e-06,
+      "loss": 1.6109,
+      "step": 1640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.564197064991954e-06,
+      "loss": 1.5835,
+      "step": 1641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5588102395439542e-06,
+      "loss": 1.6049,
+      "step": 1642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5534309923048313e-06,
+      "loss": 1.7061,
+      "step": 1643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5480593351207984e-06,
+      "loss": 1.4972,
+      "step": 1644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5426952798213507e-06,
+      "loss": 1.5194,
+      "step": 1645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.537338838219245e-06,
+      "loss": 1.64,
+      "step": 1646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319900221104666e-06,
+      "loss": 1.5682,
+      "step": 1647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5266488432742116e-06,
+      "loss": 1.4004,
+      "step": 1648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.521315313472857e-06,
+      "loss": 1.4892,
+      "step": 1649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5159894444519353e-06,
+      "loss": 1.5348,
+      "step": 1650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5106712479401043e-06,
+      "loss": 1.5384,
+      "step": 1651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.50536073564913e-06,
+      "loss": 1.5218,
+      "step": 1652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.5000579192738557e-06,
+      "loss": 1.5089,
+      "step": 1653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4947628104921757e-06,
+      "loss": 1.4287,
+      "step": 1654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4894754209650086e-06,
+      "loss": 1.5887,
+      "step": 1655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.484195762336278e-06,
+      "loss": 1.4841,
+      "step": 1656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4789238462328752e-06,
+      "loss": 1.5711,
+      "step": 1657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4736596842646534e-06,
+      "loss": 1.5912,
+      "step": 1658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4684032880243753e-06,
+      "loss": 1.4947,
+      "step": 1659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4631546690877124e-06,
+      "loss": 1.5314,
+      "step": 1660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4579138390132003e-06,
+      "loss": 1.4794,
+      "step": 1661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.452680809342233e-06,
+      "loss": 1.4619,
+      "step": 1662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4474555915990152e-06,
+      "loss": 1.5229,
+      "step": 1663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4422381972905563e-06,
+      "loss": 1.5706,
+      "step": 1664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4370286379066306e-06,
+      "loss": 1.5945,
+      "step": 1665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.431826924919762e-06,
+      "loss": 1.7066,
+      "step": 1666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4266330697851955e-06,
+      "loss": 1.5455,
+      "step": 1667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4214470839408705e-06,
+      "loss": 1.5552,
+      "step": 1668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4162689788073941e-06,
+      "loss": 1.5487,
+      "step": 1669
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.411098765788022e-06,
+      "loss": 1.5461,
+      "step": 1670
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4059364562686283e-06,
+      "loss": 1.5591,
+      "step": 1671
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.4007820616176842e-06,
+      "loss": 1.475,
+      "step": 1672
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3956355931862258e-06,
+      "loss": 1.4752,
+      "step": 1673
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3904970623078395e-06,
+      "loss": 1.5287,
+      "step": 1674
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3853664802986245e-06,
+      "loss": 1.5423,
+      "step": 1675
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3802438584571858e-06,
+      "loss": 1.5712,
+      "step": 1676
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3751292080645878e-06,
+      "loss": 1.6107,
+      "step": 1677
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.370022540384347e-06,
+      "loss": 1.4899,
+      "step": 1678
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3649238666623938e-06,
+      "loss": 1.5675,
+      "step": 1679
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.359833198127064e-06,
+      "loss": 1.5947,
+      "step": 1680
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3547505459890541e-06,
+      "loss": 1.6244,
+      "step": 1681
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3496759214414147e-06,
+      "loss": 1.5024,
+      "step": 1682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3446093356595114e-06,
+      "loss": 1.4103,
+      "step": 1683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.339550799801011e-06,
+      "loss": 1.5609,
+      "step": 1684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3345003250058525e-06,
+      "loss": 1.5669,
+      "step": 1685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.329457922396223e-06,
+      "loss": 1.5434,
+      "step": 1686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.324423603076529e-06,
+      "loss": 1.4535,
+      "step": 1687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3193973781333808e-06,
+      "loss": 1.5133,
+      "step": 1688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3143792586355614e-06,
+      "loss": 1.4612,
+      "step": 1689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3093692556340049e-06,
+      "loss": 1.5485,
+      "step": 1690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3043673801617674e-06,
+      "loss": 1.417,
+      "step": 1691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2993736432340121e-06,
+      "loss": 1.5662,
+      "step": 1692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2943880558479721e-06,
+      "loss": 1.5997,
+      "step": 1693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2894106289829444e-06,
+      "loss": 1.5078,
+      "step": 1694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2844413736002442e-06,
+      "loss": 1.7191,
+      "step": 1695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2794803006431984e-06,
+      "loss": 1.5439,
+      "step": 1696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2745274210371073e-06,
+      "loss": 1.4446,
+      "step": 1697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.269582745689238e-06,
+      "loss": 1.4527,
+      "step": 1698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2646462854887804e-06,
+      "loss": 1.304,
+      "step": 1699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2597180513068397e-06,
+      "loss": 1.5015,
+      "step": 1700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2547980539963984e-06,
+      "loss": 1.645,
+      "step": 1701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.249886304392311e-06,
+      "loss": 1.7214,
+      "step": 1702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2449828133112568e-06,
+      "loss": 1.5054,
+      "step": 1703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.240087591551737e-06,
+      "loss": 1.4718,
+      "step": 1704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235200649894035e-06,
+      "loss": 1.5925,
+      "step": 1705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2303219991002052e-06,
+      "loss": 1.4896,
+      "step": 1706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2254516499140417e-06,
+      "loss": 1.6888,
+      "step": 1707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220589613061059e-06,
+      "loss": 1.7037,
+      "step": 1708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2157358992484597e-06,
+      "loss": 1.4334,
+      "step": 1709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.210890519165125e-06,
+      "loss": 1.4209,
+      "step": 1710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.206053483481578e-06,
+      "loss": 1.4907,
+      "step": 1711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2012248028499713e-06,
+      "loss": 1.3152,
+      "step": 1712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1964044879040503e-06,
+      "loss": 1.5607,
+      "step": 1713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1915925492591456e-06,
+      "loss": 1.5511,
+      "step": 1714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1867889975121327e-06,
+      "loss": 1.5662,
+      "step": 1715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.181993843241428e-06,
+      "loss": 1.5466,
+      "step": 1716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.177207097006945e-06,
+      "loss": 1.4688,
+      "step": 1717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1724287693500879e-06,
+      "loss": 1.5671,
+      "step": 1718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1676588707937147e-06,
+      "loss": 1.5158,
+      "step": 1719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1628974118421305e-06,
+      "loss": 1.5002,
+      "step": 1720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.158144402981044e-06,
+      "loss": 1.5513,
+      "step": 1721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1533998546775643e-06,
+      "loss": 1.4348,
+      "step": 1722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1486637773801601e-06,
+      "loss": 1.5262,
+      "step": 1723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1439361815186516e-06,
+      "loss": 1.6298,
+      "step": 1724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1392170775041788e-06,
+      "loss": 1.455,
+      "step": 1725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1345064757291823e-06,
+      "loss": 1.7083,
+      "step": 1726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1298043865673757e-06,
+      "loss": 1.6208,
+      "step": 1727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1251108203737282e-06,
+      "loss": 1.4962,
+      "step": 1728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.120425787484441e-06,
+      "loss": 1.5262,
+      "step": 1729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1157492982169227e-06,
+      "loss": 1.5479,
+      "step": 1730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.111081362869763e-06,
+      "loss": 1.5116,
+      "step": 1731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1064219917227204e-06,
+      "loss": 1.623,
+      "step": 1732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1017711950366867e-06,
+      "loss": 1.5465,
+      "step": 1733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0971289830536786e-06,
+      "loss": 1.4156,
+      "step": 1734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0924953659968008e-06,
+      "loss": 1.5609,
+      "step": 1735
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.087870354070234e-06,
+      "loss": 1.5177,
+      "step": 1736
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.083253957459205e-06,
+      "loss": 1.4687,
+      "step": 1737
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0786461863299741e-06,
+      "loss": 1.5121,
+      "step": 1738
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0740470508297996e-06,
+      "loss": 1.5545,
+      "step": 1739
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0694565610869278e-06,
+      "loss": 1.5957,
+      "step": 1740
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0648747272105602e-06,
+      "loss": 1.529,
+      "step": 1741
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0603015592908405e-06,
+      "loss": 1.5513,
+      "step": 1742
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0557370673988276e-06,
+      "loss": 1.6652,
+      "step": 1743
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.051181261586473e-06,
+      "loss": 1.5689,
+      "step": 1744
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.046634151886598e-06,
+      "loss": 1.5562,
+      "step": 1745
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0420957483128757e-06,
+      "loss": 1.4859,
+      "step": 1746
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.037566060859807e-06,
+      "loss": 1.4607,
+      "step": 1747
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.033045099502697e-06,
+      "loss": 1.5622,
+      "step": 1748
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0285328741976325e-06,
+      "loss": 1.5445,
+      "step": 1749
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.024029394881464e-06,
+      "loss": 1.4963,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0195346714717813e-06,
+      "loss": 1.5887,
+      "step": 1751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0150487138668925e-06,
+      "loss": 1.519,
+      "step": 1752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0105715319457977e-06,
+      "loss": 1.4911,
+      "step": 1753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0061031355681766e-06,
+      "loss": 1.4752,
+      "step": 1754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.001643534574356e-06,
+      "loss": 1.6108,
+      "step": 1755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.971927387852992e-07,
+      "loss": 1.3417,
+      "step": 1756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.927507580025736e-07,
+      "loss": 1.5847,
+      "step": 1757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.88317602008338e-07,
+      "loss": 1.5199,
+      "step": 1758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.838932805653106e-07,
+      "loss": 1.5034,
+      "step": 1759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.794778034167658e-07,
+      "loss": 1.6118,
+      "step": 1760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.750711802864888e-07,
+      "loss": 1.6011,
+      "step": 1761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.706734208787745e-07,
+      "loss": 1.4509,
+      "step": 1762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.662845348783922e-07,
+      "loss": 1.6624,
+      "step": 1763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61904531950576e-07,
+      "loss": 1.6534,
+      "step": 1764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.575334217409933e-07,
+      "loss": 1.4703,
+      "step": 1765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.53171213875731e-07,
+      "loss": 1.6261,
+      "step": 1766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.488179179612677e-07,
+      "loss": 1.5433,
+      "step": 1767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.444735435844587e-07,
+      "loss": 1.4337,
+      "step": 1768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.401381003125115e-07,
+      "loss": 1.5745,
+      "step": 1769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.358115976929671e-07,
+      "loss": 1.5574,
+      "step": 1770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.314940452536725e-07,
+      "loss": 1.5024,
+      "step": 1771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.271854525027713e-07,
+      "loss": 1.5328,
+      "step": 1772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.228858289286674e-07,
+      "loss": 1.5676,
+      "step": 1773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.18595184000024e-07,
+      "loss": 1.5526,
+      "step": 1774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.143135271657205e-07,
+      "loss": 1.465,
+      "step": 1775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.100408678548512e-07,
+      "loss": 1.4926,
+      "step": 1776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.057772154766875e-07,
+      "loss": 1.4938,
+      "step": 1777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.015225794206756e-07,
+      "loss": 1.6286,
+      "step": 1778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972769690563976e-07,
+      "loss": 1.6519,
+      "step": 1779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.930403937335635e-07,
+      "loss": 1.5991,
+      "step": 1780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.888128627819842e-07,
+      "loss": 1.6985,
+      "step": 1781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.845943855115535e-07,
+      "loss": 1.4546,
+      "step": 1782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.803849712122292e-07,
+      "loss": 1.6048,
+      "step": 1783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.761846291540088e-07,
+      "loss": 1.4964,
+      "step": 1784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.719933685869087e-07,
+      "loss": 1.534,
+      "step": 1785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.678111987409505e-07,
+      "loss": 1.5315,
+      "step": 1786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636381288261325e-07,
+      "loss": 1.5051,
+      "step": 1787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.594741680324165e-07,
+      "loss": 1.5179,
+      "step": 1788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.553193255296994e-07,
+      "loss": 1.5226,
+      "step": 1789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.51173610467802e-07,
+      "loss": 1.4915,
+      "step": 1790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.470370319764426e-07,
+      "loss": 1.4298,
+      "step": 1791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.429095991652209e-07,
+      "loss": 1.695,
+      "step": 1792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.38791321123592e-07,
+      "loss": 1.4562,
+      "step": 1793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.346822069208549e-07,
+      "loss": 1.5752,
+      "step": 1794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.305822656061218e-07,
+      "loss": 1.5464,
+      "step": 1795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.264915062083139e-07,
+      "loss": 1.5302,
+      "step": 1796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.224099377361216e-07,
+      "loss": 1.5287,
+      "step": 1797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.183375691780032e-07,
+      "loss": 1.4676,
+      "step": 1798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.142744095021498e-07,
+      "loss": 1.5125,
+      "step": 1799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.102204676564784e-07,
+      "loss": 1.616,
+      "step": 1800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.061757525686032e-07,
+      "loss": 1.5005,
+      "step": 1801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.021402731458222e-07,
+      "loss": 1.5628,
+      "step": 1802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.981140382750902e-07,
+      "loss": 1.5722,
+      "step": 1803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.940970568230066e-07,
+      "loss": 1.5286,
+      "step": 1804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.900893376357927e-07,
+      "loss": 1.3484,
+      "step": 1805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.860908895392722e-07,
+      "loss": 1.4369,
+      "step": 1806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.821017213388504e-07,
+      "loss": 1.4392,
+      "step": 1807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.781218418194986e-07,
+      "loss": 1.4103,
+      "step": 1808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.741512597457307e-07,
+      "loss": 1.58,
+      "step": 1809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.701899838615889e-07,
+      "loss": 1.5749,
+      "step": 1810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.662380228906153e-07,
+      "loss": 1.6019,
+      "step": 1811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622953855358456e-07,
+      "loss": 1.4878,
+      "step": 1812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.583620804797753e-07,
+      "loss": 1.5713,
+      "step": 1813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.544381163843578e-07,
+      "loss": 1.5174,
+      "step": 1814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.505235018909668e-07,
+      "loss": 1.4994,
+      "step": 1815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.466182456203924e-07,
+      "loss": 1.569,
+      "step": 1816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.427223561728103e-07,
+      "loss": 1.5663,
+      "step": 1817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.388358421277758e-07,
+      "loss": 1.4959,
+      "step": 1818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.349587120441903e-07,
+      "loss": 1.5514,
+      "step": 1819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.310909744602957e-07,
+      "loss": 1.5409,
+      "step": 1820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.272326378936446e-07,
+      "loss": 1.4722,
+      "step": 1821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.233837108410907e-07,
+      "loss": 1.6203,
+      "step": 1822
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.195442017787635e-07,
+      "loss": 1.6405,
+      "step": 1823
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.157141191620548e-07,
+      "loss": 1.3772,
+      "step": 1824
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.118934714255931e-07,
+      "loss": 1.4944,
+      "step": 1825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.080822669832332e-07,
+      "loss": 1.5464,
+      "step": 1826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.042805142280329e-07,
+      "loss": 1.4356,
+      "step": 1827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.004882215322356e-07,
+      "loss": 1.4088,
+      "step": 1828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.967053972472498e-07,
+      "loss": 1.5718,
+      "step": 1829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.92932049703634e-07,
+      "loss": 1.6074,
+      "step": 1830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.891681872110778e-07,
+      "loss": 1.5111,
+      "step": 1831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.854138180583825e-07,
+      "loss": 1.5413,
+      "step": 1832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.816689505134406e-07,
+      "loss": 1.5775,
+      "step": 1833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.779335928232239e-07,
+      "loss": 1.5164,
+      "step": 1834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.74207753213757e-07,
+      "loss": 1.4366,
+      "step": 1835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.704914398901102e-07,
+      "loss": 1.572,
+      "step": 1836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.667846610363687e-07,
+      "loss": 1.5673,
+      "step": 1837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.630874248156244e-07,
+      "loss": 1.4791,
+      "step": 1838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.593997393699514e-07,
+      "loss": 1.6806,
+      "step": 1839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.557216128203942e-07,
+      "loss": 1.5726,
+      "step": 1840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.52053053266945e-07,
+      "loss": 1.4943,
+      "step": 1841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.483940687885293e-07,
+      "loss": 1.6148,
+      "step": 1842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.447446674429824e-07,
+      "loss": 1.551,
+      "step": 1843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.411048572670398e-07,
+      "loss": 1.5786,
+      "step": 1844
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.374746462763121e-07,
+      "loss": 1.565,
+      "step": 1845
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.338540424652734e-07,
+      "loss": 1.5446,
+      "step": 1846
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.302430538072373e-07,
+      "loss": 1.5677,
+      "step": 1847
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.266416882543463e-07,
+      "loss": 1.395,
+      "step": 1848
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.230499537375495e-07,
+      "loss": 1.43,
+      "step": 1849
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.194678581665864e-07,
+      "loss": 1.5542,
+      "step": 1850
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.15895409429968e-07,
+      "loss": 1.6357,
+      "step": 1851
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.123326153949633e-07,
+      "loss": 1.4846,
+      "step": 1852
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.087794839075795e-07,
+      "loss": 1.5727,
+      "step": 1853
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.052360227925446e-07,
+      "loss": 1.669,
+      "step": 1854
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.017022398532879e-07,
+      "loss": 1.4306,
+      "step": 1855
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.981781428719302e-07,
+      "loss": 1.5903,
+      "step": 1856
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.946637396092558e-07,
+      "loss": 1.5489,
+      "step": 1857
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.911590378047088e-07,
+      "loss": 1.6121,
+      "step": 1858
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.876640451763621e-07,
+      "loss": 1.563,
+      "step": 1859
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.841787694209106e-07,
+      "loss": 1.4763,
+      "step": 1860
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.807032182136485e-07,
+      "loss": 1.4875,
+      "step": 1861
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.772373992084563e-07,
+      "loss": 1.4949,
+      "step": 1862
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.737813200377817e-07,
+      "loss": 1.4309,
+      "step": 1863
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.703349883126241e-07,
+      "loss": 1.473,
+      "step": 1864
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.66898411622514e-07,
+      "loss": 1.6392,
+      "step": 1865
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.63471597535502e-07,
+      "loss": 1.5753,
+      "step": 1866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.600545535981395e-07,
+      "loss": 1.5747,
+      "step": 1867
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.566472873354617e-07,
+      "loss": 1.6834,
+      "step": 1868
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.532498062509695e-07,
+      "loss": 1.5369,
+      "step": 1869
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.498621178266167e-07,
+      "loss": 1.5005,
+      "step": 1870
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.464842295227917e-07,
+      "loss": 1.5786,
+      "step": 1871
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.431161487783009e-07,
+      "loss": 1.5555,
+      "step": 1872
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.397578830103512e-07,
+      "loss": 1.5101,
+      "step": 1873
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.364094396145364e-07,
+      "loss": 1.579,
+      "step": 1874
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.330708259648166e-07,
+      "loss": 1.6101,
+      "step": 1875
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.297420494135103e-07,
+      "loss": 1.6383,
+      "step": 1876
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.264231172912676e-07,
+      "loss": 1.5101,
+      "step": 1877
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.231140369070614e-07,
+      "loss": 1.5956,
+      "step": 1878
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.198148155481675e-07,
+      "loss": 1.616,
+      "step": 1879
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165254604801512e-07,
+      "loss": 1.4969,
+      "step": 1880
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.132459789468514e-07,
+      "loss": 1.345,
+      "step": 1881
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.099763781703621e-07,
+      "loss": 1.542,
+      "step": 1882
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.067166653510158e-07,
+      "loss": 1.5369,
+      "step": 1883
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.034668476673737e-07,
+      "loss": 1.5665,
+      "step": 1884
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.002269322762032e-07,
+      "loss": 1.4804,
+      "step": 1885
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.969969263124669e-07,
+      "loss": 1.6174,
+      "step": 1886
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.937768368893009e-07,
+      "loss": 1.5556,
+      "step": 1887
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.905666710980072e-07,
+      "loss": 1.5052,
+      "step": 1888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.873664360080321e-07,
+      "loss": 1.5201,
+      "step": 1889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.841761386669535e-07,
+      "loss": 1.4301,
+      "step": 1890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.809957861004621e-07,
+      "loss": 1.5816,
+      "step": 1891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.778253853123493e-07,
+      "loss": 1.5601,
+      "step": 1892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7466494328449187e-07,
+      "loss": 1.5707,
+      "step": 1893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.715144669768351e-07,
+      "loss": 1.6315,
+      "step": 1894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6837396332737483e-07,
+      "loss": 1.5557,
+      "step": 1895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.652434392521493e-07,
+      "loss": 1.3917,
+      "step": 1896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6212290164521554e-07,
+      "loss": 1.6164,
+      "step": 1897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5901235737864094e-07,
+      "loss": 1.6707,
+      "step": 1898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5591181330248534e-07,
+      "loss": 1.5433,
+      "step": 1899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5282127624478646e-07,
+      "loss": 1.5942,
+      "step": 1900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.497407530115422e-07,
+      "loss": 1.6851,
+      "step": 1901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.466702503866993e-07,
+      "loss": 1.6091,
+      "step": 1902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.436097751321383e-07,
+      "loss": 1.7154,
+      "step": 1903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.405593339876563e-07,
+      "loss": 1.4696,
+      "step": 1904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3751893367095233e-07,
+      "loss": 1.6008,
+      "step": 1905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344885808776145e-07,
+      "loss": 1.5471,
+      "step": 1906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.314682822811039e-07,
+      "loss": 1.5452,
+      "step": 1907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2845804453274186e-07,
+      "loss": 1.5111,
+      "step": 1908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.254578742616905e-07,
+      "loss": 1.3847,
+      "step": 1909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.224677780749431e-07,
+      "loss": 1.5945,
+      "step": 1910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1948776255730796e-07,
+      "loss": 1.4697,
+      "step": 1911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.165178342713938e-07,
+      "loss": 1.6651,
+      "step": 1912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1355799975759336e-07,
+      "loss": 1.456,
+      "step": 1913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.1060826553407376e-07,
+      "loss": 1.4588,
+      "step": 1914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.076686380967543e-07,
+      "loss": 1.481,
+      "step": 1915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0473912391930434e-07,
+      "loss": 1.457,
+      "step": 1916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.018197294531151e-07,
+      "loss": 1.4473,
+      "step": 1917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.989104611272965e-07,
+      "loss": 1.6009,
+      "step": 1918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.960113253486547e-07,
+      "loss": 1.4387,
+      "step": 1919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9312232850168495e-07,
+      "loss": 1.4745,
+      "step": 1920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.902434769485536e-07,
+      "loss": 1.5054,
+      "step": 1921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8737477702908587e-07,
+      "loss": 1.5815,
+      "step": 1922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.845162350607484e-07,
+      "loss": 1.608,
+      "step": 1923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8166785733864066e-07,
+      "loss": 1.4793,
+      "step": 1924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.788296501354771e-07,
+      "loss": 1.6317,
+      "step": 1925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7600161970157534e-07,
+      "loss": 1.4106,
+      "step": 1926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.73183772264839e-07,
+      "loss": 1.5988,
+      "step": 1927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70376114030751e-07,
+      "loss": 1.5454,
+      "step": 1928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6757865118235205e-07,
+      "loss": 1.48,
+      "step": 1929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6479138988023297e-07,
+      "loss": 1.5264,
+      "step": 1930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.620143362625156e-07,
+      "loss": 1.5022,
+      "step": 1931
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5924749644484504e-07,
+      "loss": 1.414,
+      "step": 1932
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.564908765203728e-07,
+      "loss": 1.4832,
+      "step": 1933
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5374448255974525e-07,
+      "loss": 1.4999,
+      "step": 1934
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5100832061108537e-07,
+      "loss": 1.5996,
+      "step": 1935
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.482823966999871e-07,
+      "loss": 1.5772,
+      "step": 1936
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4556671682949517e-07,
+      "loss": 1.4896,
+      "step": 1937
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4286128698009737e-07,
+      "loss": 1.6272,
+      "step": 1938
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.401661131097073e-07,
+      "loss": 1.4613,
+      "step": 1939
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.374812011536538e-07,
+      "loss": 1.5678,
+      "step": 1940
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3480655702466446e-07,
+      "loss": 1.5307,
+      "step": 1941
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3214218661285825e-07,
+      "loss": 1.4813,
+      "step": 1942
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.294880957857272e-07,
+      "loss": 1.5186,
+      "step": 1943
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2684429038812747e-07,
+      "loss": 1.5142,
+      "step": 1944
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.242107762422608e-07,
+      "loss": 1.3978,
+      "step": 1945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215875591476697e-07,
+      "loss": 1.6011,
+      "step": 1946
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1897464488121863e-07,
+      "loss": 1.5412,
+      "step": 1947
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1637203919708313e-07,
+      "loss": 1.5964,
+      "step": 1948
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1377974782673627e-07,
+      "loss": 1.5899,
+      "step": 1949
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.111977764789381e-07,
+      "loss": 1.5735,
+      "step": 1950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08626130839722e-07,
+      "loss": 1.4705,
+      "step": 1951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.060648165723823e-07,
+      "loss": 1.5402,
+      "step": 1952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.035138393174586e-07,
+      "loss": 1.4335,
+      "step": 1953
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0097320469272996e-07,
+      "loss": 1.5284,
+      "step": 1954
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.984429182931975e-07,
+      "loss": 1.5108,
+      "step": 1955
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.959229856910717e-07,
+      "loss": 1.5312,
+      "step": 1956
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.934134124357646e-07,
+      "loss": 1.4194,
+      "step": 1957
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9091420405387427e-07,
+      "loss": 1.6474,
+      "step": 1958
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.884253660491704e-07,
+      "loss": 1.5505,
+      "step": 1959
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.859469039025886e-07,
+      "loss": 1.5317,
+      "step": 1960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8347882307221243e-07,
+      "loss": 1.5147,
+      "step": 1961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8102112899326575e-07,
+      "loss": 1.4515,
+      "step": 1962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.785738270780952e-07,
+      "loss": 1.5972,
+      "step": 1963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7613692271616534e-07,
+      "loss": 1.735,
+      "step": 1964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.73710421274041e-07,
+      "loss": 1.4721,
+      "step": 1965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.712943280953795e-07,
+      "loss": 1.67,
+      "step": 1966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6888864850091457e-07,
+      "loss": 1.4873,
+      "step": 1967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6649338778844845e-07,
+      "loss": 1.5428,
+      "step": 1968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6410855123283897e-07,
+      "loss": 1.5368,
+      "step": 1969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.617341440859883e-07,
+      "loss": 1.5539,
+      "step": 1970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5937017157682876e-07,
+      "loss": 1.476,
+      "step": 1971
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57016638911316e-07,
+      "loss": 1.5298,
+      "step": 1972
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5467355127241353e-07,
+      "loss": 1.4746,
+      "step": 1973
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.523409138200839e-07,
+      "loss": 1.3951,
+      "step": 1974
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5001873169127334e-07,
+      "loss": 1.5037,
+      "step": 1975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4770700999990793e-07,
+      "loss": 1.4811,
+      "step": 1976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4540575383687313e-07,
+      "loss": 1.4351,
+      "step": 1977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.4311496827000957e-07,
+      "loss": 1.545,
+      "step": 1978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.408346583441001e-07,
+      "loss": 1.4491,
+      "step": 1979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.385648290808562e-07,
+      "loss": 1.4993,
+      "step": 1980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3630548547890931e-07,
+      "loss": 1.5946,
+      "step": 1981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.340566325138005e-07,
+      "loss": 1.677,
+      "step": 1982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3181827513796606e-07,
+      "loss": 1.4671,
+      "step": 1983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.295904182807318e-07,
+      "loss": 1.4632,
+      "step": 1984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2737306684829596e-07,
+      "loss": 1.3228,
+      "step": 1985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2516622572372416e-07,
+      "loss": 1.5104,
+      "step": 1986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2296989976693438e-07,
+      "loss": 1.4911,
+      "step": 1987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2078409381469103e-07,
+      "loss": 1.5821,
+      "step": 1988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1860881268058632e-07,
+      "loss": 1.5845,
+      "step": 1989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1644406115503947e-07,
+      "loss": 1.5687,
+      "step": 1990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.142898440052782e-07,
+      "loss": 1.5654,
+      "step": 1991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1214616597533434e-07,
+      "loss": 1.5825,
+      "step": 1992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.100130317860266e-07,
+      "loss": 1.674,
+      "step": 1993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0789044613495624e-07,
+      "loss": 1.7067,
+      "step": 1994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0577841369649576e-07,
+      "loss": 1.5096,
+      "step": 1995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.036769391217741e-07,
+      "loss": 1.5296,
+      "step": 1996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.015860270386716e-07,
+      "loss": 1.6088,
+      "step": 1997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9950568205180877e-07,
+      "loss": 1.4776,
+      "step": 1998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9743590874253204e-07,
+      "loss": 1.4914,
+      "step": 1999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.953767116689098e-07,
+      "loss": 1.6193,
+      "step": 2000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.933280953657174e-07,
+      "loss": 1.4344,
+      "step": 2001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.9129006434443209e-07,
+      "loss": 1.4099,
+      "step": 2002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8926262309321542e-07,
+      "loss": 1.6607,
+      "step": 2003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8724577607691196e-07,
+      "loss": 1.523,
+      "step": 2004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.85239527737035e-07,
+      "loss": 1.4195,
+      "step": 2005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.832438824917565e-07,
+      "loss": 1.4825,
+      "step": 2006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.812588447358976e-07,
+      "loss": 1.6181,
+      "step": 2007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7928441884092096e-07,
+      "loss": 1.5008,
+      "step": 2008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7732060915491956e-07,
+      "loss": 1.493,
+      "step": 2009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7536742000260677e-07,
+      "loss": 1.6059,
+      "step": 2010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7342485568530742e-07,
+      "loss": 1.4266,
+      "step": 2011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.714929204809479e-07,
+      "loss": 1.4489,
+      "step": 2012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6957161864404825e-07,
+      "loss": 1.626,
+      "step": 2013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6766095440571118e-07,
+      "loss": 1.5494,
+      "step": 2014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6576093197361253e-07,
+      "loss": 1.4725,
+      "step": 2015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6387155553199363e-07,
+      "loss": 1.5421,
+      "step": 2016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.619928292416495e-07,
+      "loss": 1.6076,
+      "step": 2017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6012475723992394e-07,
+      "loss": 1.5147,
+      "step": 2018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.582673436406956e-07,
+      "loss": 1.6185,
+      "step": 2019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.564205925343726e-07,
+      "loss": 1.4919,
+      "step": 2020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5458450798788049e-07,
+      "loss": 1.5429,
+      "step": 2021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.527590940446566e-07,
+      "loss": 1.5621,
+      "step": 2022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.5094435472463753e-07,
+      "loss": 1.5056,
+      "step": 2023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.491402940242548e-07,
+      "loss": 1.4771,
+      "step": 2024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4734691591642036e-07,
+      "loss": 1.5984,
+      "step": 2025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.455642243505223e-07,
+      "loss": 1.5988,
+      "step": 2026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4379222325241527e-07,
+      "loss": 1.4281,
+      "step": 2027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.420309165244116e-07,
+      "loss": 1.4307,
+      "step": 2028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4028030804527027e-07,
+      "loss": 1.496,
+      "step": 2029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385404016701919e-07,
+      "loss": 1.4144,
+      "step": 2030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3681120123080926e-07,
+      "loss": 1.5466,
+      "step": 2031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3509271053517848e-07,
+      "loss": 1.7727,
+      "step": 2032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3338493336776947e-07,
+      "loss": 1.5866,
+      "step": 2033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3168787348945944e-07,
+      "loss": 1.4925,
+      "step": 2034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.300015346375244e-07,
+      "loss": 1.6961,
+      "step": 2035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2832592052562932e-07,
+      "loss": 1.6963,
+      "step": 2036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2666103484382186e-07,
+      "loss": 1.4493,
+      "step": 2037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2500688125852366e-07,
+      "loss": 1.5707,
+      "step": 2038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2336346341252138e-07,
+      "loss": 1.6114,
+      "step": 2039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2173078492496049e-07,
+      "loss": 1.5359,
+      "step": 2040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2010884939133382e-07,
+      "loss": 1.6691,
+      "step": 2041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1849766038347976e-07,
+      "loss": 1.528,
+      "step": 2042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1689722144956672e-07,
+      "loss": 1.5938,
+      "step": 2043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1530753611409151e-07,
+      "loss": 1.4959,
+      "step": 2044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1372860787786933e-07,
+      "loss": 1.3982,
+      "step": 2045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1216044021802485e-07,
+      "loss": 1.4656,
+      "step": 2046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1060303658798621e-07,
+      "loss": 1.5708,
+      "step": 2047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0905640041747767e-07,
+      "loss": 1.4486,
+      "step": 2048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0752053511251026e-07,
+      "loss": 1.4623,
+      "step": 2049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0599544405537676e-07,
+      "loss": 1.5991,
+      "step": 2050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.044811306046406e-07,
+      "loss": 1.5053,
+      "step": 2051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0297759809513308e-07,
+      "loss": 1.5638,
+      "step": 2052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0148484983794227e-07,
+      "loss": 1.5644,
+      "step": 2053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0000288912040635e-07,
+      "loss": 1.4313,
+      "step": 2054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.853171920610972e-08,
+      "loss": 1.5041,
+      "step": 2055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.707134333487078e-08,
+      "loss": 1.5639,
+      "step": 2056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.562176472273866e-08,
+      "loss": 1.6146,
+      "step": 2057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.41829865619831e-08,
+      "loss": 1.5912,
+      "step": 2058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.275501202109127e-08,
+      "loss": 1.6132,
+      "step": 2059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.133784424475712e-08,
+      "loss": 1.5845,
+      "step": 2060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.993148635387528e-08,
+      "loss": 1.6474,
+      "step": 2061
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.853594144553612e-08,
+      "loss": 1.5639,
+      "step": 2062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.715121259301684e-08,
+      "loss": 1.5306,
+      "step": 2063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.577730284577645e-08,
+      "loss": 1.775,
+      "step": 2064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.441421522944638e-08,
+      "loss": 1.4332,
+      "step": 2065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.306195274582707e-08,
+      "loss": 1.5409,
+      "step": 2066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.172051837287808e-08,
+      "loss": 1.4402,
+      "step": 2067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.038991506471584e-08,
+      "loss": 1.5511,
+      "step": 2068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.907014575160188e-08,
+      "loss": 1.4756,
+      "step": 2069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.77612133399408e-08,
+      "loss": 1.4475,
+      "step": 2070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.646312071227069e-08,
+      "loss": 1.6539,
+      "step": 2071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.517587072725984e-08,
+      "loss": 1.4713,
+      "step": 2072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.389946621969679e-08,
+      "loss": 1.6539,
+      "step": 2073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.263391000048858e-08,
+      "loss": 1.5809,
+      "step": 2074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.137920485665028e-08,
+      "loss": 1.5395,
+      "step": 2075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.013535355130052e-08,
+      "loss": 1.5675,
+      "step": 2076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.890235882365648e-08,
+      "loss": 1.4979,
+      "step": 2077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.76802233890278e-08,
+      "loss": 1.6089,
+      "step": 2078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.64689499388066e-08,
+      "loss": 1.5138,
+      "step": 2079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.526854114046799e-08,
+      "loss": 1.6072,
+      "step": 2080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.407899963755848e-08,
+      "loss": 1.5944,
+      "step": 2081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.290032804969426e-08,
+      "loss": 1.634,
+      "step": 2082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.173252897255289e-08,
+      "loss": 1.5318,
+      "step": 2083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.05756049778683e-08,
+      "loss": 1.4533,
+      "step": 2084
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.942955861342581e-08,
+      "loss": 1.4146,
+      "step": 2085
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.829439240305601e-08,
+      "loss": 1.4829,
+      "step": 2086
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.7170108846627546e-08,
+      "loss": 1.5073,
+      "step": 2087
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.6056710420045456e-08,
+      "loss": 1.6517,
+      "step": 2088
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.495419957524229e-08,
+      "loss": 1.5191,
+      "step": 2089
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.38625787401742e-08,
+      "loss": 1.551,
+      "step": 2090
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.2781850318814335e-08,
+      "loss": 1.456,
+      "step": 2091
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1712016691150554e-08,
+      "loss": 1.6702,
+      "step": 2092
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.065308021317661e-08,
+      "loss": 1.5,
+      "step": 2093
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9605043216888196e-08,
+      "loss": 1.517,
+      "step": 2094
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.856790801027911e-08,
+      "loss": 1.5868,
+      "step": 2095
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.754167687733458e-08,
+      "loss": 1.5397,
+      "step": 2096
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6526352078027916e-08,
+      "loss": 1.4962,
+      "step": 2097
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5521935848312217e-08,
+      "loss": 1.5166,
+      "step": 2098
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.452843040011978e-08,
+      "loss": 1.5469,
+      "step": 2099
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.354583792135436e-08,
+      "loss": 1.6114,
+      "step": 2100
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2574160575886146e-08,
+      "loss": 1.5454,
+      "step": 2101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1613400503550114e-08,
+      "loss": 1.6582,
+      "step": 2102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0663559820137144e-08,
+      "loss": 1.5218,
+      "step": 2103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.972464061739234e-08,
+      "loss": 1.6678,
+      "step": 2104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.879664496300894e-08,
+      "loss": 1.4027,
+      "step": 2105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.787957490062444e-08,
+      "loss": 1.4935,
+      "step": 2106
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.697343244981721e-08,
+      "loss": 1.4621,
+      "step": 2107
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.607821960609881e-08,
+      "loss": 1.5112,
+      "step": 2108
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5193938340911674e-08,
+      "loss": 1.46,
+      "step": 2109
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.432059060162696e-08,
+      "loss": 1.6183,
+      "step": 2110
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.345817831153508e-08,
+      "loss": 1.458,
+      "step": 2111
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.260670336984628e-08,
+      "loss": 1.4859,
+      "step": 2112
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1766167651682835e-08,
+      "loss": 1.6023,
+      "step": 2113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0936573008078504e-08,
+      "loss": 1.5727,
+      "step": 2114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.01179212659719e-08,
+      "loss": 1.5961,
+      "step": 2115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9310214228202016e-08,
+      "loss": 1.5385,
+      "step": 2116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8513453673506552e-08,
+      "loss": 1.5844,
+      "step": 2117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.77276413565164e-08,
+      "loss": 1.4014,
+      "step": 2118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6952779007751727e-08,
+      "loss": 1.5735,
+      "step": 2119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6188868333619776e-08,
+      "loss": 1.4506,
+      "step": 2120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5435911016408744e-08,
+      "loss": 1.5674,
+      "step": 2121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4693908714286675e-08,
+      "loss": 1.4574,
+      "step": 2122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.396286306129425e-08,
+      "loss": 1.5293,
+      "step": 2123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.324277566734534e-08,
+      "loss": 1.6807,
+      "step": 2124
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.253364811822034e-08,
+      "loss": 1.3958,
+      "step": 2125
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1835481975565066e-08,
+      "loss": 1.574,
+      "step": 2126
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.114827877688408e-08,
+      "loss": 1.5591,
+      "step": 2127
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0472040035541263e-08,
+      "loss": 1.4656,
+      "step": 2128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9806767240753143e-08,
+      "loss": 1.5079,
+      "step": 2129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9152461857588344e-08,
+      "loss": 1.6189,
+      "step": 2130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.850912532696092e-08,
+      "loss": 1.4821,
+      "step": 2131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7876759065631466e-08,
+      "loss": 1.5064,
+      "step": 2132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7255364466200464e-08,
+      "loss": 1.715,
+      "step": 2133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.664494289710661e-08,
+      "loss": 1.5471,
+      "step": 2134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6045495702624035e-08,
+      "loss": 1.5834,
+      "step": 2135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5457024202858996e-08,
+      "loss": 1.4656,
+      "step": 2136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4879529693747063e-08,
+      "loss": 1.574,
+      "step": 2137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4313013447050384e-08,
+      "loss": 1.5552,
+      "step": 2138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3757476710354323e-08,
+      "loss": 1.4609,
+      "step": 2139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3212920707064703e-08,
+      "loss": 1.4078,
+      "step": 2140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2679346636406686e-08,
+      "loss": 1.6417,
+      "step": 2141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.215675567342034e-08,
+      "loss": 1.4175,
+      "step": 2142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1645148968957853e-08,
+      "loss": 1.6055,
+      "step": 2143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1144527649683545e-08,
+      "loss": 1.4198,
+      "step": 2144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0654892818068307e-08,
+      "loss": 1.5223,
+      "step": 2145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0176245552388497e-08,
+      "loss": 1.4751,
+      "step": 2146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.708586906724826e-09,
+      "loss": 1.6665,
+      "step": 2147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.251917910957365e-09,
+      "loss": 1.5708,
+      "step": 2148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.80623957076554e-09,
+      "loss": 1.5524,
+      "step": 2149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.371552867625365e-09,
+      "loss": 1.4675,
+      "step": 2150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.94785875880555e-09,
+      "loss": 1.4836,
+      "step": 2151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.535158177368607e-09,
+      "loss": 1.6038,
+      "step": 2152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.133452032166421e-09,
+      "loss": 1.6189,
+      "step": 2153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.742741207837467e-09,
+      "loss": 1.636,
+      "step": 2154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3630265648084725e-09,
+      "loss": 1.5606,
+      "step": 2155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.994308939289428e-09,
+      "loss": 1.5777,
+      "step": 2156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.63658914327192e-09,
+      "loss": 1.6809,
+      "step": 2157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.289867964529127e-09,
+      "loss": 1.4412,
+      "step": 2158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.954146166612494e-09,
+      "loss": 1.5979,
+      "step": 2159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.629424488850065e-09,
+      "loss": 1.4953,
+      "step": 2160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.315703646346481e-09,
+      "loss": 1.5182,
+      "step": 2161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0129843299790975e-09,
+      "loss": 1.4742,
+      "step": 2162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.721267206399093e-09,
+      "loss": 1.4634,
+      "step": 2163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.440552918026474e-09,
+      "loss": 1.7064,
+      "step": 2164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.170842083053405e-09,
+      "loss": 1.4788,
+      "step": 2165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.912135295437546e-09,
+      "loss": 1.3883,
+      "step": 2166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6644331249048306e-09,
+      "loss": 1.5321,
+      "step": 2167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4277361169466884e-09,
+      "loss": 1.5907,
+      "step": 2168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.202044792818381e-09,
+      "loss": 1.5856,
+      "step": 2169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.987359649539e-09,
+      "loss": 1.5276,
+      "step": 2170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7836811598898053e-09,
+      "loss": 1.5695,
+      "step": 2171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.591009772412e-09,
+      "loss": 1.643,
+      "step": 2172
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4093459114078445e-09,
+      "loss": 1.6049,
+      "step": 2173
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2386899769389893e-09,
+      "loss": 1.6216,
+      "step": 2174
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.07904234482481e-09,
+      "loss": 1.4244,
+      "step": 2175
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.304033666424072e-10,
+      "loss": 1.4753,
+      "step": 2176
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.927733697249418e-10,
+      "loss": 1.4408,
+      "step": 2177
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.66152657162189e-10,
+      "loss": 1.5678,
+      "step": 2178
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.505415077999843e-10,
+      "loss": 1.5372,
+      "step": 2179
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.459401762363369e-10,
+      "loss": 1.4767,
+      "step": 2180
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5234889282642626e-10,
+      "loss": 1.4698,
+      "step": 2181
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6976786367649555e-10,
+      "loss": 1.3657,
+      "step": 2182
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9819727064718243e-10,
+      "loss": 1.5903,
+      "step": 2183
+    },
+    {
+      "epoch": 1.0,
+      "step": 2183,
+      "total_flos": 82634518457344.0,
+      "train_loss": 0.4821208130040957,
+      "train_runtime": 3140.1887,
+      "train_samples_per_second": 22.245,
+      "train_steps_per_second": 0.695
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2183,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "total_flos": 82634518457344.0,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}